ICU 71.1  71.1
uchar.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 **********************************************************************
5 * Copyright (C) 1997-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
8 *
9 * File UCHAR.H
10 *
11 * Modification History:
12 *
13 * Date Name Description
14 * 04/02/97 aliu Creation.
15 * 03/29/99 helena Updated for C APIs.
16 * 4/15/99 Madhu Updated for C Implementation and Javadoc
17 * 5/20/99 Madhu Added the function u_getVersion()
18 * 8/19/1999 srl Upgraded scripts to Unicode 3.0
19 * 8/27/1999 schererm UCharDirection constants: U_...
20 * 11/11/1999 weiv added u_isalnum(), cleaned comments
21 * 01/11/2000 helena Renamed u_getVersion to u_getUnicodeVersion().
22 ******************************************************************************
23 */
24 
25 #ifndef UCHAR_H
26 #define UCHAR_H
27 
28 #include "unicode/utypes.h"
29 #include "unicode/stringoptions.h"
30 #include "unicode/ucpmap.h"
31 
32 #if !defined(USET_DEFINED) && !defined(U_IN_DOXYGEN)
33 
34 #define USET_DEFINED
35 
44 typedef struct USet USet;
45 
46 #endif
47 
48 
50 
51 /*==========================================================================*/
52 /* Unicode version number */
53 /*==========================================================================*/
63 #define U_UNICODE_VERSION "14.0"
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_DRAFT_API
487 
536 #endif // U_HIDE_DRAFT_API
537 #ifndef U_HIDE_DEPRECATED_API
538 
543 #endif // U_HIDE_DEPRECATED_API
544 
552  UCHAR_BLOCK=0x1001,
580  UCHAR_SCRIPT=0x100A,
652 #ifndef U_HIDE_DEPRECATED_API
653 
658 #endif // U_HIDE_DEPRECATED_API
659 
671 #ifndef U_HIDE_DEPRECATED_API
672 
677 #endif // U_HIDE_DEPRECATED_API
678 
684 #ifndef U_HIDE_DEPRECATED_API
685 
690 #endif // U_HIDE_DEPRECATED_API
691 
694  UCHAR_AGE=0x4000,
703 #ifndef U_HIDE_DEPRECATED_API
704 
707 #endif /* U_HIDE_DEPRECATED_API */
708 
713  UCHAR_NAME=0x4005,
729 #ifndef U_HIDE_DEPRECATED_API
730 
735 #endif /* U_HIDE_DEPRECATED_API */
736 
742 #ifndef U_HIDE_DEPRECATED_API
743 
748 #endif // U_HIDE_DEPRECATED_API
749 
758 #ifndef U_HIDE_DEPRECATED_API
759 
764 #endif // U_HIDE_DEPRECATED_API
765 
768 } UProperty;
769 
775 typedef enum UCharCategory
776 {
777  /*
778  * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
779  * It matches pairs of lines like
780  * / ** <Unicode 2-letter General_Category value> comment... * /
781  * U_<[A-Z_]+> = <integer>,
782  */
783 
854 } UCharCategory;
855 
870 #define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES)
871 
873 #define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER)
874 
875 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
876 
877 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
878 
879 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
880 
881 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
882 
884 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
885 
886 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
887 
888 #define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK)
889 
891 #define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER)
892 
893 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
894 
895 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
896 
898 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
899 
900 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
901 
902 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
903 
905 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
906 
907 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
908 
909 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
910 
911 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
912 
914 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
915 
916 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
917 
918 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
919 
920 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
921 
922 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
923 
925 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
926 
927 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
928 
929 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
930 
931 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
932 
934 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
935 
936 #define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION)
937 
938 
940 #define U_GC_L_MASK \
941  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
942 
944 #define U_GC_LC_MASK \
945  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
946 
948 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
949 
951 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
952 
954 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
955 
957 #define U_GC_C_MASK \
958  (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
959 
961 #define U_GC_P_MASK \
962  (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
963  U_GC_PI_MASK|U_GC_PF_MASK)
964 
966 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
967 
972 typedef enum UCharDirection {
973  /*
974  * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
975  * It matches pairs of lines like
976  * / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
977  * U_<[A-Z_]+> = <integer>,
978  */
979 
1026 #ifndef U_HIDE_DEPRECATED_API
1027 
1034 #endif // U_HIDE_DEPRECATED_API
1035 } UCharDirection;
1036 
1044  /*
1045  * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
1046  * It matches lines like
1047  * U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
1048  */
1049 
1056 #ifndef U_HIDE_DEPRECATED_API
1057 
1063  U_BPT_COUNT /* 3 */
1064 #endif // U_HIDE_DEPRECATED_API
1066 
1072  /*
1073  * Note: UBlockCode constants are parsed by preparseucd.py.
1074  * It matches lines like
1075  * UBLOCK_<Unicode Block value name> = <integer>,
1076  */
1077 
1079  UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
1080 
1082  UBLOCK_BASIC_LATIN = 1, /*[0000]*/
1083 
1086 
1089 
1092 
1094  UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
1095 
1098 
1101 
1106  UBLOCK_GREEK =8, /*[0370]*/
1107 
1109  UBLOCK_CYRILLIC =9, /*[0400]*/
1110 
1112  UBLOCK_ARMENIAN =10, /*[0530]*/
1113 
1115  UBLOCK_HEBREW =11, /*[0590]*/
1116 
1118  UBLOCK_ARABIC =12, /*[0600]*/
1119 
1121  UBLOCK_SYRIAC =13, /*[0700]*/
1122 
1124  UBLOCK_THAANA =14, /*[0780]*/
1125 
1127  UBLOCK_DEVANAGARI =15, /*[0900]*/
1128 
1130  UBLOCK_BENGALI =16, /*[0980]*/
1131 
1133  UBLOCK_GURMUKHI =17, /*[0A00]*/
1134 
1136  UBLOCK_GUJARATI =18, /*[0A80]*/
1137 
1139  UBLOCK_ORIYA =19, /*[0B00]*/
1140 
1142  UBLOCK_TAMIL =20, /*[0B80]*/
1143 
1145  UBLOCK_TELUGU =21, /*[0C00]*/
1146 
1148  UBLOCK_KANNADA =22, /*[0C80]*/
1149 
1151  UBLOCK_MALAYALAM =23, /*[0D00]*/
1152 
1154  UBLOCK_SINHALA =24, /*[0D80]*/
1155 
1157  UBLOCK_THAI =25, /*[0E00]*/
1158 
1160  UBLOCK_LAO =26, /*[0E80]*/
1161 
1163  UBLOCK_TIBETAN =27, /*[0F00]*/
1164 
1166  UBLOCK_MYANMAR =28, /*[1000]*/
1167 
1169  UBLOCK_GEORGIAN =29, /*[10A0]*/
1170 
1172  UBLOCK_HANGUL_JAMO =30, /*[1100]*/
1173 
1175  UBLOCK_ETHIOPIC =31, /*[1200]*/
1176 
1178  UBLOCK_CHEROKEE =32, /*[13A0]*/
1179 
1182 
1184  UBLOCK_OGHAM =34, /*[1680]*/
1185 
1187  UBLOCK_RUNIC =35, /*[16A0]*/
1188 
1190  UBLOCK_KHMER =36, /*[1780]*/
1191 
1193  UBLOCK_MONGOLIAN =37, /*[1800]*/
1194 
1197 
1199  UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
1200 
1203 
1206 
1208  UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
1209 
1215 
1218 
1220  UBLOCK_NUMBER_FORMS =45, /*[2150]*/
1221 
1223  UBLOCK_ARROWS =46, /*[2190]*/
1224 
1227 
1230 
1232  UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
1233 
1236 
1239 
1241  UBLOCK_BOX_DRAWING =52, /*[2500]*/
1242 
1244  UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
1245 
1247  UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
1248 
1251 
1253  UBLOCK_DINGBATS =56, /*[2700]*/
1254 
1256  UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
1257 
1260 
1262  UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
1263 
1266 
1269 
1271  UBLOCK_HIRAGANA =62, /*[3040]*/
1272 
1274  UBLOCK_KATAKANA =63, /*[30A0]*/
1275 
1277  UBLOCK_BOPOMOFO =64, /*[3100]*/
1278 
1281 
1283  UBLOCK_KANBUN =66, /*[3190]*/
1284 
1287 
1290 
1293 
1296 
1299 
1301  UBLOCK_YI_SYLLABLES =72, /*[A000]*/
1302 
1304  UBLOCK_YI_RADICALS =73, /*[A490]*/
1305 
1307  UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
1308 
1310  UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
1311 
1314 
1316  UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
1317 
1327  UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
1338 
1341 
1344 
1347 
1350 
1353 
1356 
1359 
1361  UBLOCK_SPECIALS =86, /*[FFF0]*/
1362 
1365 
1366  /* New blocks in Unicode 3.1 */
1367 
1369  UBLOCK_OLD_ITALIC = 88, /*[10300]*/
1371  UBLOCK_GOTHIC = 89, /*[10330]*/
1373  UBLOCK_DESERET = 90, /*[10400]*/
1377  UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
1385  UBLOCK_TAGS = 96, /*[E0000]*/
1386 
1387  /* New blocks in Unicode 3.2 */
1388 
1397  UBLOCK_TAGALOG = 98, /*[1700]*/
1399  UBLOCK_HANUNOO = 99, /*[1720]*/
1401  UBLOCK_BUHID = 100, /*[1740]*/
1403  UBLOCK_TAGBANWA = 101, /*[1760]*/
1417  UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
1422 
1423  /* New blocks in Unicode 4 */
1424 
1426  UBLOCK_LIMBU = 111, /*[1900]*/
1428  UBLOCK_TAI_LE = 112, /*[1950]*/
1430  UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
1432  UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
1438  UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
1440  UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
1442  UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
1444  UBLOCK_UGARITIC = 120, /*[10380]*/
1446  UBLOCK_SHAVIAN = 121, /*[10450]*/
1448  UBLOCK_OSMANYA = 122, /*[10480]*/
1450  UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
1452  UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
1455 
1456  /* New blocks in Unicode 4.1 */
1457 
1461  UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
1463  UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
1465  UBLOCK_BUGINESE = 129, /*[1A00]*/
1467  UBLOCK_CJK_STROKES = 130, /*[31C0]*/
1471  UBLOCK_COPTIC = 132, /*[2C80]*/
1473  UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
1475  UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
1477  UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
1479  UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
1481  UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
1485  UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
1487  UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
1493  UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
1495  UBLOCK_TIFINAGH = 144, /*[2D30]*/
1497  UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
1498 
1499  /* New blocks in Unicode 5.0 */
1500 
1502  UBLOCK_NKO = 146, /*[07C0]*/
1504  UBLOCK_BALINESE = 147, /*[1B00]*/
1506  UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
1508  UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
1510  UBLOCK_PHAGS_PA = 150, /*[A840]*/
1512  UBLOCK_PHOENICIAN = 151, /*[10900]*/
1514  UBLOCK_CUNEIFORM = 152, /*[12000]*/
1518  UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
1519 
1520  /* New blocks in Unicode 5.1 */
1521 
1523  UBLOCK_SUNDANESE = 155, /*[1B80]*/
1525  UBLOCK_LEPCHA = 156, /*[1C00]*/
1527  UBLOCK_OL_CHIKI = 157, /*[1C50]*/
1529  UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
1531  UBLOCK_VAI = 159, /*[A500]*/
1533  UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
1535  UBLOCK_SAURASHTRA = 161, /*[A880]*/
1537  UBLOCK_KAYAH_LI = 162, /*[A900]*/
1539  UBLOCK_REJANG = 163, /*[A930]*/
1541  UBLOCK_CHAM = 164, /*[AA00]*/
1543  UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
1545  UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
1547  UBLOCK_LYCIAN = 167, /*[10280]*/
1549  UBLOCK_CARIAN = 168, /*[102A0]*/
1551  UBLOCK_LYDIAN = 169, /*[10920]*/
1553  UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
1555  UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
1556 
1557  /* New blocks in Unicode 5.2 */
1558 
1560  UBLOCK_SAMARITAN = 172, /*[0800]*/
1564  UBLOCK_TAI_THAM = 174, /*[1A20]*/
1566  UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
1568  UBLOCK_LISU = 176, /*[A4D0]*/
1570  UBLOCK_BAMUM = 177, /*[A6A0]*/
1574  UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
1578  UBLOCK_JAVANESE = 181, /*[A980]*/
1580  UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
1582  UBLOCK_TAI_VIET = 183, /*[AA80]*/
1584  UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
1588  UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
1590  UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
1592  UBLOCK_AVESTAN = 188, /*[10B00]*/
1596  UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
1598  UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
1600  UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
1602  UBLOCK_KAITHI = 193, /*[11080]*/
1604  UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
1611 
1612  /* New blocks in Unicode 6.0 */
1613 
1615  UBLOCK_MANDAIC = 198, /*[0840]*/
1617  UBLOCK_BATAK = 199, /*[1BC0]*/
1619  UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
1621  UBLOCK_BRAHMI = 201, /*[11000]*/
1623  UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
1625  UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
1627  UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
1631  UBLOCK_EMOTICONS = 206, /*[1F600]*/
1635  UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
1638 
1639  /* New blocks in Unicode 6.1 */
1640 
1642  UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
1646  UBLOCK_CHAKMA = 212, /*[11100]*/
1650  UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
1652  UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
1654  UBLOCK_MIAO = 216, /*[16F00]*/
1656  UBLOCK_SHARADA = 217, /*[11180]*/
1658  UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
1662  UBLOCK_TAKRI = 220, /*[11680]*/
1663 
1664  /* New blocks in Unicode 7.0 */
1665 
1667  UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
1669  UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
1671  UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
1675  UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
1677  UBLOCK_ELBASAN = 226, /*[10500]*/
1681  UBLOCK_GRANTHA = 228, /*[11300]*/
1683  UBLOCK_KHOJKI = 229, /*[11200]*/
1685  UBLOCK_KHUDAWADI = 230, /*[112B0]*/
1687  UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
1689  UBLOCK_LINEAR_A = 232, /*[10600]*/
1691  UBLOCK_MAHAJANI = 233, /*[11150]*/
1693  UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
1695  UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
1697  UBLOCK_MODI = 236, /*[11600]*/
1699  UBLOCK_MRO = 237, /*[16A40]*/
1701  UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
1703  UBLOCK_NABATAEAN = 239, /*[10880]*/
1705  UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
1707  UBLOCK_OLD_PERMIC = 241, /*[10350]*/
1709  UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
1711  UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
1713  UBLOCK_PALMYRENE = 244, /*[10860]*/
1715  UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
1717  UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
1721  UBLOCK_SIDDHAM = 248, /*[11580]*/
1725  UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
1727  UBLOCK_TIRHUTA = 251, /*[11480]*/
1729  UBLOCK_WARANG_CITI = 252, /*[118A0]*/
1730 
1731  /* New blocks in Unicode 8.0 */
1732 
1734  UBLOCK_AHOM = 253, /*[11700]*/
1736  UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
1738  UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
1744  UBLOCK_HATRAN = 258, /*[108E0]*/
1746  UBLOCK_MULTANI = 259, /*[11280]*/
1748  UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
1752  UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/
1753 
1754  /* New blocks in Unicode 9.0 */
1755 
1757  UBLOCK_ADLAM = 263, /*[1E900]*/
1759  UBLOCK_BHAIKSUKI = 264, /*[11C00]*/
1761  UBLOCK_CYRILLIC_EXTENDED_C = 265, /*[1C80]*/
1763  UBLOCK_GLAGOLITIC_SUPPLEMENT = 266, /*[1E000]*/
1767  UBLOCK_MARCHEN = 268, /*[11C70]*/
1769  UBLOCK_MONGOLIAN_SUPPLEMENT = 269, /*[11660]*/
1771  UBLOCK_NEWA = 270, /*[11400]*/
1773  UBLOCK_OSAGE = 271, /*[104B0]*/
1775  UBLOCK_TANGUT = 272, /*[17000]*/
1777  UBLOCK_TANGUT_COMPONENTS = 273, /*[18800]*/
1778 
1779  // New blocks in Unicode 10.0
1780 
1784  UBLOCK_KANA_EXTENDED_A = 275, /*[1B100]*/
1786  UBLOCK_MASARAM_GONDI = 276, /*[11D00]*/
1788  UBLOCK_NUSHU = 277, /*[1B170]*/
1790  UBLOCK_SOYOMBO = 278, /*[11A50]*/
1792  UBLOCK_SYRIAC_SUPPLEMENT = 279, /*[0860]*/
1794  UBLOCK_ZANABAZAR_SQUARE = 280, /*[11A00]*/
1795 
1796  // New blocks in Unicode 11.0
1797 
1799  UBLOCK_CHESS_SYMBOLS = 281, /*[1FA00]*/
1801  UBLOCK_DOGRA = 282, /*[11800]*/
1803  UBLOCK_GEORGIAN_EXTENDED = 283, /*[1C90]*/
1805  UBLOCK_GUNJALA_GONDI = 284, /*[11D60]*/
1807  UBLOCK_HANIFI_ROHINGYA = 285, /*[10D00]*/
1809  UBLOCK_INDIC_SIYAQ_NUMBERS = 286, /*[1EC70]*/
1811  UBLOCK_MAKASAR = 287, /*[11EE0]*/
1813  UBLOCK_MAYAN_NUMERALS = 288, /*[1D2E0]*/
1815  UBLOCK_MEDEFAIDRIN = 289, /*[16E40]*/
1817  UBLOCK_OLD_SOGDIAN = 290, /*[10F00]*/
1819  UBLOCK_SOGDIAN = 291, /*[10F30]*/
1820 
1821  // New blocks in Unicode 12.0
1822 
1826  UBLOCK_ELYMAIC = 293, /*[10FE0]*/
1828  UBLOCK_NANDINAGARI = 294, /*[119A0]*/
1832  UBLOCK_OTTOMAN_SIYAQ_NUMBERS = 296, /*[1ED00]*/
1834  UBLOCK_SMALL_KANA_EXTENSION = 297, /*[1B130]*/
1838  UBLOCK_TAMIL_SUPPLEMENT = 299, /*[11FC0]*/
1840  UBLOCK_WANCHO = 300, /*[1E2C0]*/
1841 
1842  // New blocks in Unicode 13.0
1843 
1845  UBLOCK_CHORASMIAN = 301, /*[10FB0]*/
1849  UBLOCK_DIVES_AKURU = 303, /*[11900]*/
1851  UBLOCK_KHITAN_SMALL_SCRIPT = 304, /*[18B00]*/
1853  UBLOCK_LISU_SUPPLEMENT = 305, /*[11FB0]*/
1857  UBLOCK_TANGUT_SUPPLEMENT = 307, /*[18D00]*/
1859  UBLOCK_YEZIDI = 308, /*[10E80]*/
1860 
1861  // New blocks in Unicode 14.0
1862 
1864  UBLOCK_ARABIC_EXTENDED_B = 309, /*[0870]*/
1866  UBLOCK_CYPRO_MINOAN = 310, /*[12F90]*/
1868  UBLOCK_ETHIOPIC_EXTENDED_B = 311, /*[1E7E0]*/
1870  UBLOCK_KANA_EXTENDED_B = 312, /*[1AFF0]*/
1872  UBLOCK_LATIN_EXTENDED_F = 313, /*[10780]*/
1874  UBLOCK_LATIN_EXTENDED_G = 314, /*[1DF00]*/
1876  UBLOCK_OLD_UYGHUR = 315, /*[10F70]*/
1878  UBLOCK_TANGSA = 316, /*[16A70]*/
1880  UBLOCK_TOTO = 317, /*[1E290]*/
1884  UBLOCK_VITHKUQI = 319, /*[10570]*/
1887 
1888 #ifndef U_HIDE_DEPRECATED_API
1889 
1896 #endif // U_HIDE_DEPRECATED_API
1897 
1900 };
1901 
1903 typedef enum UBlockCode UBlockCode;
1904 
1912 typedef enum UEastAsianWidth {
1913  /*
1914  * Note: UEastAsianWidth constants are parsed by preparseucd.py.
1915  * It matches lines like
1916  * U_EA_<Unicode East_Asian_Width value name>
1917  */
1918 
1919  U_EA_NEUTRAL, /*[N]*/
1920  U_EA_AMBIGUOUS, /*[A]*/
1921  U_EA_HALFWIDTH, /*[H]*/
1922  U_EA_FULLWIDTH, /*[F]*/
1923  U_EA_NARROW, /*[Na]*/
1924  U_EA_WIDE, /*[W]*/
1925 #ifndef U_HIDE_DEPRECATED_API
1926 
1933 #endif // U_HIDE_DEPRECATED_API
1934 } UEastAsianWidth;
1935 
1947 typedef enum UCharNameChoice {
1950 #ifndef U_HIDE_DEPRECATED_API
1951 
1957 #endif /* U_HIDE_DEPRECATED_API */
1958 
1962 #ifndef U_HIDE_DEPRECATED_API
1963 
1968 #endif // U_HIDE_DEPRECATED_API
1969 } UCharNameChoice;
1970 
1984 typedef enum UPropertyNameChoice {
1985  U_SHORT_PROPERTY_NAME,
1986  U_LONG_PROPERTY_NAME,
1987 #ifndef U_HIDE_DEPRECATED_API
1988 
1993 #endif // U_HIDE_DEPRECATED_API
1995 
2002 typedef enum UDecompositionType {
2003  /*
2004  * Note: UDecompositionType constants are parsed by preparseucd.py.
2005  * It matches lines like
2006  * U_DT_<Unicode Decomposition_Type value name>
2007  */
2008 
2009  U_DT_NONE, /*[none]*/
2010  U_DT_CANONICAL, /*[can]*/
2011  U_DT_COMPAT, /*[com]*/
2012  U_DT_CIRCLE, /*[enc]*/
2013  U_DT_FINAL, /*[fin]*/
2014  U_DT_FONT, /*[font]*/
2015  U_DT_FRACTION, /*[fra]*/
2016  U_DT_INITIAL, /*[init]*/
2017  U_DT_ISOLATED, /*[iso]*/
2018  U_DT_MEDIAL, /*[med]*/
2019  U_DT_NARROW, /*[nar]*/
2020  U_DT_NOBREAK, /*[nb]*/
2021  U_DT_SMALL, /*[sml]*/
2022  U_DT_SQUARE, /*[sqr]*/
2023  U_DT_SUB, /*[sub]*/
2024  U_DT_SUPER, /*[sup]*/
2025  U_DT_VERTICAL, /*[vert]*/
2026  U_DT_WIDE, /*[wide]*/
2027 #ifndef U_HIDE_DEPRECATED_API
2028 
2034  U_DT_COUNT /* 18 */
2035 #endif // U_HIDE_DEPRECATED_API
2037 
2044 typedef enum UJoiningType {
2045  /*
2046  * Note: UJoiningType constants are parsed by preparseucd.py.
2047  * It matches lines like
2048  * U_JT_<Unicode Joining_Type value name>
2049  */
2050 
2051  U_JT_NON_JOINING, /*[U]*/
2052  U_JT_JOIN_CAUSING, /*[C]*/
2053  U_JT_DUAL_JOINING, /*[D]*/
2054  U_JT_LEFT_JOINING, /*[L]*/
2055  U_JT_RIGHT_JOINING, /*[R]*/
2056  U_JT_TRANSPARENT, /*[T]*/
2057 #ifndef U_HIDE_DEPRECATED_API
2058 
2064  U_JT_COUNT /* 6 */
2065 #endif // U_HIDE_DEPRECATED_API
2066 } UJoiningType;
2067 
2074 typedef enum UJoiningGroup {
2075  /*
2076  * Note: UJoiningGroup constants are parsed by preparseucd.py.
2077  * It matches lines like
2078  * U_JG_<Unicode Joining_Group value name>
2079  */
2080 
2081  U_JG_NO_JOINING_GROUP,
2082  U_JG_AIN,
2083  U_JG_ALAPH,
2084  U_JG_ALEF,
2085  U_JG_BEH,
2086  U_JG_BETH,
2087  U_JG_DAL,
2088  U_JG_DALATH_RISH,
2089  U_JG_E,
2090  U_JG_FEH,
2091  U_JG_FINAL_SEMKATH,
2092  U_JG_GAF,
2093  U_JG_GAMAL,
2094  U_JG_HAH,
2096  U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
2097  U_JG_HE,
2098  U_JG_HEH,
2099  U_JG_HEH_GOAL,
2100  U_JG_HETH,
2101  U_JG_KAF,
2102  U_JG_KAPH,
2103  U_JG_KNOTTED_HEH,
2104  U_JG_LAM,
2105  U_JG_LAMADH,
2106  U_JG_MEEM,
2107  U_JG_MIM,
2108  U_JG_NOON,
2109  U_JG_NUN,
2110  U_JG_PE,
2111  U_JG_QAF,
2112  U_JG_QAPH,
2113  U_JG_REH,
2114  U_JG_REVERSED_PE,
2115  U_JG_SAD,
2116  U_JG_SADHE,
2117  U_JG_SEEN,
2118  U_JG_SEMKATH,
2119  U_JG_SHIN,
2120  U_JG_SWASH_KAF,
2121  U_JG_SYRIAC_WAW,
2122  U_JG_TAH,
2123  U_JG_TAW,
2124  U_JG_TEH_MARBUTA,
2125  U_JG_TETH,
2126  U_JG_WAW,
2127  U_JG_YEH,
2128  U_JG_YEH_BARREE,
2129  U_JG_YEH_WITH_TAIL,
2130  U_JG_YUDH,
2131  U_JG_YUDH_HE,
2132  U_JG_ZAIN,
2190 #ifndef U_HIDE_DEPRECATED_API
2191 
2198 #endif // U_HIDE_DEPRECATED_API
2199 } UJoiningGroup;
2200 
2208  /*
2209  * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2210  * It matches lines like
2211  * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2212  */
2213 
2214  U_GCB_OTHER = 0, /*[XX]*/
2215  U_GCB_CONTROL = 1, /*[CN]*/
2216  U_GCB_CR = 2, /*[CR]*/
2217  U_GCB_EXTEND = 3, /*[EX]*/
2218  U_GCB_L = 4, /*[L]*/
2219  U_GCB_LF = 5, /*[LF]*/
2220  U_GCB_LV = 6, /*[LV]*/
2221  U_GCB_LVT = 7, /*[LVT]*/
2222  U_GCB_T = 8, /*[T]*/
2223  U_GCB_V = 9, /*[V]*/
2225  U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2227  U_GCB_PREPEND = 11, /*[PP]*/
2229  U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2231  U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2233  U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2235  U_GCB_E_MODIFIER = 15, /*[EM]*/
2237  U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2239  U_GCB_ZWJ = 17, /*[ZWJ]*/
2240 
2241 #ifndef U_HIDE_DEPRECATED_API
2242 
2249 #endif // U_HIDE_DEPRECATED_API
2251 
2259 typedef enum UWordBreakValues {
2260  /*
2261  * Note: UWordBreakValues constants are parsed by preparseucd.py.
2262  * It matches lines like
2263  * U_WB_<Unicode Word_Break value name>
2264  */
2265 
2266  U_WB_OTHER = 0, /*[XX]*/
2267  U_WB_ALETTER = 1, /*[LE]*/
2268  U_WB_FORMAT = 2, /*[FO]*/
2269  U_WB_KATAKANA = 3, /*[KA]*/
2270  U_WB_MIDLETTER = 4, /*[ML]*/
2271  U_WB_MIDNUM = 5, /*[MN]*/
2272  U_WB_NUMERIC = 6, /*[NU]*/
2273  U_WB_EXTENDNUMLET = 7, /*[EX]*/
2275  U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2277  U_WB_EXTEND = 9, /*[Extend]*/
2279  U_WB_LF = 10, /*[LF]*/
2281  U_WB_MIDNUMLET =11, /*[MB]*/
2283  U_WB_NEWLINE =12, /*[NL]*/
2285  U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2287  U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2289  U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2291  U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2293  U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2295  U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2297  U_WB_E_MODIFIER = 19, /*[EM]*/
2299  U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2301  U_WB_ZWJ = 21, /*[ZWJ]*/
2303  U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/
2304 
2305 #ifndef U_HIDE_DEPRECATED_API
2306 
2313 #endif // U_HIDE_DEPRECATED_API
2315 
2322 typedef enum USentenceBreak {
2323  /*
2324  * Note: USentenceBreak constants are parsed by preparseucd.py.
2325  * It matches lines like
2326  * U_SB_<Unicode Sentence_Break value name>
2327  */
2328 
2329  U_SB_OTHER = 0, /*[XX]*/
2330  U_SB_ATERM = 1, /*[AT]*/
2331  U_SB_CLOSE = 2, /*[CL]*/
2332  U_SB_FORMAT = 3, /*[FO]*/
2333  U_SB_LOWER = 4, /*[LO]*/
2334  U_SB_NUMERIC = 5, /*[NU]*/
2335  U_SB_OLETTER = 6, /*[LE]*/
2336  U_SB_SEP = 7, /*[SE]*/
2337  U_SB_SP = 8, /*[SP]*/
2338  U_SB_STERM = 9, /*[ST]*/
2339  U_SB_UPPER = 10, /*[UP]*/
2340  U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2341  U_SB_EXTEND = 12, /*[EX]*/
2342  U_SB_LF = 13, /*[LF]*/
2343  U_SB_SCONTINUE = 14, /*[SC]*/
2344 #ifndef U_HIDE_DEPRECATED_API
2345 
2352 #endif // U_HIDE_DEPRECATED_API
2353 } USentenceBreak;
2354 
2361 typedef enum ULineBreak {
2362  /*
2363  * Note: ULineBreak constants are parsed by preparseucd.py.
2364  * It matches lines like
2365  * U_LB_<Unicode Line_Break value name>
2366  */
2367 
2368  U_LB_UNKNOWN = 0, /*[XX]*/
2369  U_LB_AMBIGUOUS = 1, /*[AI]*/
2370  U_LB_ALPHABETIC = 2, /*[AL]*/
2371  U_LB_BREAK_BOTH = 3, /*[B2]*/
2372  U_LB_BREAK_AFTER = 4, /*[BA]*/
2373  U_LB_BREAK_BEFORE = 5, /*[BB]*/
2374  U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2375  U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2376  U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2377  U_LB_COMBINING_MARK = 9, /*[CM]*/
2378  U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2379  U_LB_EXCLAMATION = 11, /*[EX]*/
2380  U_LB_GLUE = 12, /*[GL]*/
2381  U_LB_HYPHEN = 13, /*[HY]*/
2382  U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2384  U_LB_INSEPARABLE = 15, /*[IN]*/
2385  U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2386  U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2387  U_LB_LINE_FEED = 17, /*[LF]*/
2388  U_LB_NONSTARTER = 18, /*[NS]*/
2389  U_LB_NUMERIC = 19, /*[NU]*/
2390  U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2391  U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2392  U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2393  U_LB_QUOTATION = 23, /*[QU]*/
2394  U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2395  U_LB_SURROGATE = 25, /*[SG]*/
2396  U_LB_SPACE = 26, /*[SP]*/
2397  U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2398  U_LB_ZWSPACE = 28, /*[ZW]*/
2400  U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2402  U_LB_WORD_JOINER = 30, /*[WJ]*/
2404  U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2406  U_LB_H3 = 32, /*[H3]*/
2408  U_LB_JL = 33, /*[JL]*/
2410  U_LB_JT = 34, /*[JT]*/
2412  U_LB_JV = 35, /*[JV]*/
2414  U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2416  U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2418  U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2420  U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2422  U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2424  U_LB_E_MODIFIER = 41, /*[EM]*/
2426  U_LB_ZWJ = 42, /*[ZWJ]*/
2427 #ifndef U_HIDE_DEPRECATED_API
2428 
2435 #endif // U_HIDE_DEPRECATED_API
2436 } ULineBreak;
2437 
2444 typedef enum UNumericType {
2445  /*
2446  * Note: UNumericType constants are parsed by preparseucd.py.
2447  * It matches lines like
2448  * U_NT_<Unicode Numeric_Type value name>
2449  */
2450 
2451  U_NT_NONE, /*[None]*/
2452  U_NT_DECIMAL, /*[de]*/
2453  U_NT_DIGIT, /*[di]*/
2454  U_NT_NUMERIC, /*[nu]*/
2455 #ifndef U_HIDE_DEPRECATED_API
2456 
2463 #endif // U_HIDE_DEPRECATED_API
2464 } UNumericType;
2465 
2472 typedef enum UHangulSyllableType {
2473  /*
2474  * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2475  * It matches lines like
2476  * U_HST_<Unicode Hangul_Syllable_Type value name>
2477  */
2478 
2479  U_HST_NOT_APPLICABLE, /*[NA]*/
2480  U_HST_LEADING_JAMO, /*[L]*/
2481  U_HST_VOWEL_JAMO, /*[V]*/
2482  U_HST_TRAILING_JAMO, /*[T]*/
2483  U_HST_LV_SYLLABLE, /*[LV]*/
2484  U_HST_LVT_SYLLABLE, /*[LVT]*/
2485 #ifndef U_HIDE_DEPRECATED_API
2486 
2493 #endif // U_HIDE_DEPRECATED_API
2495 
2503  /*
2504  * Note: UIndicPositionalCategory constants are parsed by preparseucd.py.
2505  * It matches lines like
2506  * U_INPC_<Unicode Indic_Positional_Category value name>
2507  */
2508 
2542 
2550  /*
2551  * Note: UIndicSyllabicCategory constants are parsed by preparseucd.py.
2552  * It matches lines like
2553  * U_INSC_<Unicode Indic_Syllabic_Category value name>
2554  */
2555 
2629 
2636 typedef enum UVerticalOrientation {
2637  /*
2638  * Note: UVerticalOrientation constants are parsed by preparseucd.py.
2639  * It matches lines like
2640  * U_VO_<Unicode Vertical_Orientation value name>
2641  */
2642 
2652 
2680 U_CAPI UBool U_EXPORT2
2682 
2683 #ifndef U_HIDE_DRAFT_API
2684 
2709 U_CAPI UBool U_EXPORT2
2710 u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which);
2711 
2712 #endif // U_HIDE_DRAFT_API
2713 
2729 U_CAPI const USet * U_EXPORT2
2730 u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode);
2731 
2744 U_CAPI UBool U_EXPORT2
2746 
2759 U_CAPI UBool U_EXPORT2
2761 
2774 U_CAPI UBool U_EXPORT2
2776 
2795 U_CAPI UBool U_EXPORT2
2797 
2836 U_CAPI int32_t U_EXPORT2
2838 
2857 U_CAPI int32_t U_EXPORT2
2859 
2886 U_CAPI int32_t U_EXPORT2
2888 
2904 U_CAPI const UCPMap * U_EXPORT2
2905 u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode);
2906 
2929 U_CAPI double U_EXPORT2
2931 
2939 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
2940 
2964 U_CAPI UBool U_EXPORT2
2965 u_islower(UChar32 c);
2966 
2991 U_CAPI UBool U_EXPORT2
2992 u_isupper(UChar32 c);
2993 
3008 U_CAPI UBool U_EXPORT2
3009 u_istitle(UChar32 c);
3010 
3029 U_CAPI UBool U_EXPORT2
3030 u_isdigit(UChar32 c);
3031 
3050 U_CAPI UBool U_EXPORT2
3051 u_isalpha(UChar32 c);
3052 
3071 U_CAPI UBool U_EXPORT2
3072 u_isalnum(UChar32 c);
3073 
3094 U_CAPI UBool U_EXPORT2
3095 u_isxdigit(UChar32 c);
3096 
3110 U_CAPI UBool U_EXPORT2
3111 u_ispunct(UChar32 c);
3112 
3129 U_CAPI UBool U_EXPORT2
3130 u_isgraph(UChar32 c);
3131 
3157 U_CAPI UBool U_EXPORT2
3158 u_isblank(UChar32 c);
3159 
3182 U_CAPI UBool U_EXPORT2
3183 u_isdefined(UChar32 c);
3184 
3203 U_CAPI UBool U_EXPORT2
3204 u_isspace(UChar32 c);
3205 
3224 U_CAPI UBool U_EXPORT2
3226 
3264 U_CAPI UBool U_EXPORT2
3266 
3288 U_CAPI UBool U_EXPORT2
3289 u_iscntrl(UChar32 c);
3290 
3303 U_CAPI UBool U_EXPORT2
3305 
3321 U_CAPI UBool U_EXPORT2
3322 u_isprint(UChar32 c);
3323 
3341 U_CAPI UBool U_EXPORT2
3342 u_isbase(UChar32 c);
3343 
3360 U_CAPI UCharDirection U_EXPORT2
3362 
3378 U_CAPI UBool U_EXPORT2
3380 
3400 U_CAPI UChar32 U_EXPORT2
3402 
3419 U_CAPI UChar32 U_EXPORT2
3421 
3433 U_CAPI int8_t U_EXPORT2
3434 u_charType(UChar32 c);
3435 
3449 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3450 
3468 typedef UBool U_CALLCONV
3469 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3470 
3490 U_CAPI void U_EXPORT2
3491 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3492 
3493 #if !UCONFIG_NO_NORMALIZATION
3494 
3502 U_CAPI uint8_t U_EXPORT2
3504 
3505 #endif
3506 
3530 U_CAPI int32_t U_EXPORT2
3532 
3542 U_CAPI UBlockCode U_EXPORT2
3544 
3577 U_CAPI int32_t U_EXPORT2
3578 u_charName(UChar32 code, UCharNameChoice nameChoice,
3579  char *buffer, int32_t bufferLength,
3580  UErrorCode *pErrorCode);
3581 
3582 #ifndef U_HIDE_DEPRECATED_API
3583 
3601 U_DEPRECATED int32_t U_EXPORT2
3603  char *dest, int32_t destCapacity,
3604  UErrorCode *pErrorCode);
3605 #endif /* U_HIDE_DEPRECATED_API */
3606 
3627 U_CAPI UChar32 U_EXPORT2
3628 u_charFromName(UCharNameChoice nameChoice,
3629  const char *name,
3630  UErrorCode *pErrorCode);
3631 
3649 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3650  UChar32 code,
3651  UCharNameChoice nameChoice,
3652  const char *name,
3653  int32_t length);
3654 
3676 U_CAPI void U_EXPORT2
3677 u_enumCharNames(UChar32 start, UChar32 limit,
3678  UEnumCharNamesFn *fn,
3679  void *context,
3680  UCharNameChoice nameChoice,
3681  UErrorCode *pErrorCode);
3682 
3714 U_CAPI const char* U_EXPORT2
3715 u_getPropertyName(UProperty property,
3716  UPropertyNameChoice nameChoice);
3717 
3737 U_CAPI UProperty U_EXPORT2
3738 u_getPropertyEnum(const char* alias);
3739 
3787 U_CAPI const char* U_EXPORT2
3789  int32_t value,
3790  UPropertyNameChoice nameChoice);
3791 
3823 U_CAPI int32_t U_EXPORT2
3825  const char* alias);
3826 
3844 U_CAPI UBool U_EXPORT2
3845 u_isIDStart(UChar32 c);
3846 
3868 U_CAPI UBool U_EXPORT2
3869 u_isIDPart(UChar32 c);
3870 
3891 U_CAPI UBool U_EXPORT2
3893 
3910 U_CAPI UBool U_EXPORT2
3912 
3931 U_CAPI UBool U_EXPORT2
3933 
3956 U_CAPI UChar32 U_EXPORT2
3957 u_tolower(UChar32 c);
3958 
3981 U_CAPI UChar32 U_EXPORT2
3982 u_toupper(UChar32 c);
3983 
4006 U_CAPI UChar32 U_EXPORT2
4007 u_totitle(UChar32 c);
4008 
4031 U_CAPI UChar32 U_EXPORT2
4032 u_foldCase(UChar32 c, uint32_t options);
4033 
4072 U_CAPI int32_t U_EXPORT2
4073 u_digit(UChar32 ch, int8_t radix);
4074 
4103 U_CAPI UChar32 U_EXPORT2
4104 u_forDigit(int32_t digit, int8_t radix);
4105 
4120 U_CAPI void U_EXPORT2
4121 u_charAge(UChar32 c, UVersionInfo versionArray);
4122 
4134 U_CAPI void U_EXPORT2
4135 u_getUnicodeVersion(UVersionInfo versionArray);
4136 
4137 #if !UCONFIG_NO_NORMALIZATION
4138 
4159 U_CAPI int32_t U_EXPORT2
4160 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
4161 
4162 #endif
4163 
4164 
4166 
4167 #endif /*_UCHAR*/
4168 /*eof*/
uint8_t UVersionInfo[U_MAX_VERSION_LENGTH]
The binary form of a version on ICU APIs is an array of 4 uint8_t.
Definition: uversion.h:59
Binary property Ideographic.
Definition: uchar.h:271
Unicode 4.0.1 renames the &quot;Cyrillic Supplementary&quot; block to &quot;Cyrillic Supplement&quot;.
Definition: uchar.h:1395
Enumerated property Indic_Positional_Category.
Definition: uchar.h:637
Binary property Changes_When_Lowercased.
Definition: uchar.h:423
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition: uchar.h:275
U_CAPI UBool u_isalnum(UChar32 c)
Determines whether the specified code point is an alphanumeric character (letter or digit) according ...
Binary property Case_Ignorable.
Definition: uchar.h:421
Enumerated property NFC_Quick_Check.
Definition: uchar.h:592
U_CAPI UBool u_isbase(UChar32 c)
Non-standard: Determines whether the specified code point is a base character.
One more than the highest normal UJoiningGroup value.
Definition: uchar.h:2197
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:755
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1337
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:549
Binary property XID_Start.
Definition: uchar.h:328
Binary property Join_Control.
Definition: uchar.h:282
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition: uchar.h:286
struct UCPMap UCPMap
Abstract map from Unicode code points (U+0000..U+10FFFF) to integer values.
Definition: ucpmap.h:31
Binary property White_Space.
Definition: uchar.h:321
String property Titlecase_Mapping.
Definition: uchar.h:728
U_CAPI UBool u_isISOControl(UChar32 c)
Determines whether the specified code point is an ISO control code.
U_CAPI int32_t u_getIntPropertyValue(UChar32 c, UProperty which)
Get the property value for an enumerated or integer Unicode property for a code point.
#define U_CAPI
This is used to declare a function as a public ICU C API.
Definition: umachine.h:110
U_CAPI UBool u_isblank(UChar32 c)
Determines whether the specified code point is a &quot;blank&quot; or &quot;horizontal space&quot;, a character that visi...
One more than the highest normal UPropertyNameChoice value.
Definition: uchar.h:1992
U_CAPI UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:657
Enumerated property Numeric_Type.
Definition: uchar.h:577
Binary property xdigit (a C/POSIX character class).
Definition: uchar.h:417
Binary property Alphabetic.
Definition: uchar.h:209
First constant for double Unicode properties.
Definition: uchar.h:683
C API: Bit set option bit constants for various string and character processing functions.
U_CAPI UBool u_isdefined(UChar32 c)
Determines whether the specified code point is &quot;defined&quot;, which usually means that it is assigned a c...
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:3649
Binary property of strings RGI_Emoji.
Definition: uchar.h:535
U_CAPI UChar32 u_totitle(UChar32 c)
The given character is mapped to its titlecase equivalent according to UnicodeData.txt; if none is defined, the character itself is returned.
U_CAPI UBool u_isspace(UChar32 c)
Determines if the specified character is a space character or not.
One more than the highest normal USentenceBreak value.
Definition: uchar.h:2351
Binary property Emoji_Component.
Definition: uchar.h:468
Binary property of strings Emoji_Keycap_Sequence.
Definition: uchar.h:500
Binary property Emoji.
Definition: uchar.h:440
Cn &quot;Other, Not Assigned (no characters in [UnicodeData.txt] have this property)&quot; (same as U_UNASSIGNE...
Definition: uchar.h:787
U_CAPI UProperty u_getPropertyEnum(const char *alias)
Return the UProperty enum for a given property name, as specified in the Unicode database file Proper...
String property Simple_Case_Folding.
Definition: uchar.h:716
U_CAPI 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.
Binary property NFC_Inert.
Definition: uchar.h:364
Binary property graph (a C/POSIX character class).
Definition: uchar.h:407
String property Bidi_Mirroring_Glyph.
Definition: uchar.h:699
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:676
U_CAPI UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
Enumerated property Block.
Definition: uchar.h:552
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:767
Renamed from the misspelled &quot;inseperable&quot; in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2384
Binary property Radical (new in Unicode 3.2).
Definition: uchar.h:301
UCharDirection
This specifies the language directional property of a character set.
Definition: uchar.h:972
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition: uchar.h:279
Binary property Grapheme_Link (new in Unicode 3.2).
Definition: uchar.h:253
Enumerated property Decomposition_Type.
Definition: uchar.h:558
Binary property Regional_Indicator.
Definition: uchar.h:473
U_CAPI UChar32 u_foldCase(UChar32 c, uint32_t options)
The given character is mapped to its case folding equivalent according to UnicodeData.txt and CaseFolding.txt; if the character has no case folding equivalent, the character itself is returned.
#define U_CALLCONV
Similar to U_CDECL_BEGIN/U_CDECL_END, this qualifier is necessary in callback function typedefs to ma...
Definition: platform.h:870
String property Case_Folding.
Definition: uchar.h:702
String property Name.
Definition: uchar.h:713
This file defines an abstract map from Unicode code points to integer values.
String property Simple_Uppercase_Mapping.
Definition: uchar.h:725
U_CAPI UBool u_isxdigit(UChar32 c)
Determines whether the specified code point is a hexadecimal digit.
U_CAPI UBool u_isgraph(UChar32 c)
Determines whether the specified code point is a &quot;graphic&quot; character (printable, excluding spaces)...
Enumerated property Bidi_Class.
Definition: uchar.h:547
Enumerated property General_Category.
Definition: uchar.h:565
Binary property Prepended_Concatenation_Mark.
Definition: uchar.h:478
U_CAPI UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
Sm.
Definition: uchar.h:835
U_CAPI void u_charAge(UChar32 c, UVersionInfo versionArray)
Get the &quot;age&quot; of the code point.
String property Unicode_1_Name.
Definition: uchar.h:734
U_CAPI int32_t u_getIntPropertyMinValue(UProperty which)
Get the minimum value for an enumerated/integer/binary Unicode property.
U_CAPI void u_getUnicodeVersion(UVersionInfo versionArray)
Gets the Unicode version information.
UNumericType
Numeric Type constants.
Definition: uchar.h:2444
U_CAPI UChar32 u_charMirror(UChar32 c)
Maps the specified character to a &quot;mirror-image&quot; character.
One more than the highest normal UBidiPairedBracketType value.
Definition: uchar.h:1063
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:392
Enumerated property Vertical_Orientation.
Definition: uchar.h:651
Close paired bracket.
Definition: uchar.h:1055
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&lt;=c...
Definition: uchar.h:3469
Binary property Changes_When_Casefolded.
Definition: uchar.h:429
U_CAPI void u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context)
Enumerate efficiently all code points with their Unicode general categories.
Binary property NFD_Inert.
Definition: uchar.h:350
Binary property Diacritic.
Definition: uchar.h:234
Binary property Terminal_Punctuation.
Definition: uchar.h:310
U_CAPI UBool u_isMirrored(UChar32 c)
Determines whether the code point has the Bidi_Mirrored property.
U_CAPI const UCPMap * u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode)
Returns an immutable UCPMap for an enumerated/catalog/int-valued property.
One more than the highest normal UNumericType value.
Definition: uchar.h:2462
U_CAPI UBool u_isUWhiteSpace(UChar32 c)
Check if a code point has the White_Space Unicode property.
Enumerated property NFD_Quick_Check.
Definition: uchar.h:586
U_CAPI int32_t u_digit(UChar32 ch, int8_t radix)
Returns the decimal digit value of the code point in the specified radix.
U_CAPI uint8_t u_getCombiningClass(UChar32 c)
Returns the combining class of the code point as specified in UnicodeData.txt.
#define U_CDECL_BEGIN
This is used to begin a declaration of a library private ICU C API.
Definition: umachine.h:85
Binary property STerm (new in Unicode 4.0.1).
Definition: uchar.h:337
Enumerated property Joining_Group.
Definition: uchar.h:568
Binary property ID_Continue.
Definition: uchar.h:264
Binary property Emoji_Presentation.
Definition: uchar.h:447
U_CAPI 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...
Open paired bracket.
Definition: uchar.h:1053
Binary property blank (a C/POSIX character class).
Definition: uchar.h:402
Binary property Quotation_Mark.
Definition: uchar.h:297
#define U_DEPRECATED
This is used to declare a function as a deprecated public ICU C API.
Definition: umachine.h:116
Binary property Changes_When_NFKC_Casefolded.
Definition: uchar.h:433
U_CAPI UBlockCode ublock_getCode(UChar32 c)
Returns the Unicode allocation block that contains the character.
First constant for binary Unicode properties.
Definition: uchar.h:211
U_CAPI UBool u_isIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in an identifier accordin...
Binary property Noncharacter_Code_Point.
Definition: uchar.h:295
Binary property Hyphen.
Definition: uchar.h:259
U_CAPI UBool u_ispunct(UChar32 c)
Determines whether the specified code point is a punctuation character.
U_CAPI int32_t u_charName(UChar32 code, UCharNameChoice nameChoice, char *buffer, int32_t bufferLength, UErrorCode *pErrorCode)
Retrieve the name of a Unicode character.
Enumerated property East_Asian_Width.
Definition: uchar.h:562
ULineBreak
Line Break constants.
Definition: uchar.h:2361
U_CAPI double u_getNumericValue(UChar32 c)
Get the numeric value for a Unicode code point as defined in the Unicode Character Database...
U_CAPI UBool u_isUUppercase(UChar32 c)
Check if a code point has the Uppercase Unicode property.
One more than the highest UCharDirection value.
Definition: uchar.h:1033
U_CAPI UChar32 u_tolower(UChar32 c)
The given character is mapped to its lowercase equivalent according to UnicodeData.txt; if the character has no lowercase equivalent, the character itself is returned.
Binary property Full_Composition_Exclusion.
Definition: uchar.h:242
Bitmask property General_Category_Mask.
Definition: uchar.h:668
String property Simple_Titlecase_Mapping.
Definition: uchar.h:722
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2248
Unicode 3.2 renames this block to &quot;Combining Diacritical Marks for Symbols&quot;.
Definition: uchar.h:1214
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:2002
One more than the highest normal ULineBreak value.
Definition: uchar.h:2434
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2472
String property Simple_Lowercase_Mapping.
Definition: uchar.h:719
UIndicPositionalCategory
Indic Positional Category constants.
Definition: uchar.h:2502
Binary property print (a C/POSIX character class).
Definition: uchar.h:412
Binary property Case_Sensitive.
Definition: uchar.h:332
Standard or synthetic character name.
Definition: uchar.h:1959
Binary property Bidi_Mirrored.
Definition: uchar.h:222
UVerticalOrientation
Vertical Orientation constants.
Definition: uchar.h:2636
Binary property NFKC_Inert.
Definition: uchar.h:371
U_CAPI 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...
Binary property Changes_When_Casemapped.
Definition: uchar.h:431
First constant for string Unicode properties.
Definition: uchar.h:696
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition: uchar.h:250
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:467
U_CAPI UBool u_isUAlphabetic(UChar32 c)
Check if a code point has the Alphabetic Unicode property.
UGraphemeClusterBreak
Grapheme Cluster Break constants.
Definition: uchar.h:2207
New No_Block value in Unicode 4.
Definition: uchar.h:1079
Binary property Extender.
Definition: uchar.h:238
Double property Numeric_Value.
Definition: uchar.h:681
Binary property Math.
Definition: uchar.h:291
Unicode character name (Name property).
Definition: uchar.h:1949
U_CAPI const char * u_getPropertyName(UProperty property, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property, as given in the Unicode database file PropertyAliases...
Binary property Grapheme_Base (new in Unicode 3.2).
Definition: uchar.h:246
Binary property NFKD_Inert.
Definition: uchar.h:357
U_CAPI UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier...
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:775
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:619
Binary property Lowercase.
Definition: uchar.h:289
First constant for bit-mask Unicode properties.
Definition: uchar.h:670
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:629
Binary property of strings Basic_Emoji.
Definition: uchar.h:493
U_CAPI UChar32 u_toupper(UChar32 c)
The given character is mapped to its uppercase equivalent according to UnicodeData.txt; if the character has no uppercase equivalent, the character itself is returned.
USentenceBreak
Sentence Break constants.
Definition: uchar.h:2322
Binary property Extended_Pictographic.
Definition: uchar.h:485
Enumerated property Indic_Syllabic_Category.
Definition: uchar.h:644
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:314
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:555
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:1947
One more than the last constant for binary Unicode properties.
Definition: uchar.h:542
Enumerated property Script.
Definition: uchar.h:580
Unicode 3.2 renames this block to &quot;Greek and Coptic&quot;.
Definition: uchar.h:1106
Binary property Hex_Digit.
Definition: uchar.h:256
String property Uppercase_Mapping.
Definition: uchar.h:738
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:1984
U_CAPI UBool u_isupper(UChar32 c)
Determines whether the specified code point has the general category &quot;Lu&quot; (uppercase letter)...
String property Lowercase_Mapping.
Definition: uchar.h:710
char16_t UChar
The base type for UTF-16 code units and pointers.
Definition: umachine.h:418
#define U_CDECL_END
This is used to end a declaration of a library private ICU C API.
Definition: umachine.h:86
Enumerated property NFKC_Quick_Check.
Definition: uchar.h:595
Not a paired bracket.
Definition: uchar.h:1051
U_CAPI UBool u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which)
Returns true if the property is true for the string.
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:195
U_CAPI const USet * u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode)
Returns a frozen USet for a binary property.
Enumerated property Hangul_Syllable_Type, new in Unicode 4.
Definition: uchar.h:583
Binary property Dash.
Definition: uchar.h:224
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:397
Cf.
Definition: uchar.h:819
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition: uchar.h:343
Binary property of strings RGI_Emoji_ZWJ_Sequence.
Definition: uchar.h:528
One more than the highest normal UHangulSyllableType value.
Definition: uchar.h:2492
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:1071
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:624
Binary property Deprecated (new in Unicode 3.2).
Definition: uchar.h:231
Binary property Bidi_Control.
Definition: uchar.h:217
Binary property XID_Continue.
Definition: uchar.h:325
Same as UBLOCK_PRIVATE_USE.
Definition: uchar.h:1327
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415
Binary property Uppercase.
Definition: uchar.h:317
Binary property Changes_When_Uppercased.
Definition: uchar.h:425
UJoiningGroup
Joining Group constants.
Definition: uchar.h:2074
Binary property Cased.
Definition: uchar.h:419
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition: uset.h:50
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:1932
U_CAPI UCharDirection u_charDirection(UChar32 c)
Returns the bidirectional category value for the code point, which is used in the Unicode bidirection...
One more than the highest normal UDecompositionType value.
Definition: uchar.h:2034
Cs.
Definition: uchar.h:823
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:1912
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:609
U_CAPI int8_t u_charType(UChar32 c)
Returns the general category value for the code point.
U_CAPI UBool u_hasBinaryProperty(UChar32 c, UProperty which)
Check a binary Unicode property for a code point.
Non-category for unassigned and non-character code points.
Definition: uchar.h:785
First constant for Unicode properties with unusual value types.
Definition: uchar.h:757
U_CAPI int32_t u_charDigitValue(UChar32 c)
Returns the decimal digit value of a decimal digit character.
U_CAPI int32_t u_getIntPropertyMaxValue(UProperty which)
Get the maximum value for an enumerated/integer/binary Unicode property.
UWordBreakValues
Word Break constants.
Definition: uchar.h:2259
Binary property of strings RGI_Emoji_Modifier_Sequence.
Definition: uchar.h:507
U_CAPI UBool u_isJavaIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in a Java identifier...
Enumerated property Joining_Type.
Definition: uchar.h:571
One more than the last constant for double Unicode properties.
Definition: uchar.h:689
Binary property Emoji_Modifier.
Definition: uchar.h:454
Basic definitions for ICU, for both C and C++ APIs.
U_CAPI UBool u_isWhitespace(UChar32 c)
Determines if the specified code point is a whitespace character according to Java/ICU.
U_CAPI UBool u_isJavaSpaceChar(UChar32 c)
Determine if the specified code point is a space character according to Java.
String property Bidi_Paired_Bracket (new in Unicode 6.3).
Definition: uchar.h:741
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:602
Binary property ASCII_Hex_Digit.
Definition: uchar.h:213
UJoiningType
Joining Type constants.
Definition: uchar.h:2044
Binary property Soft_Dotted (new in Unicode 3.2).
Definition: uchar.h:306
One more than the last constant for string Unicode properties.
Definition: uchar.h:747
Binary property Emoji_Modifier_Base.
Definition: uchar.h:461
Binary Property Segment_Starter.
Definition: uchar.h:382
U_CAPI UChar32 u_forDigit(int32_t digit, int8_t radix)
Determines the character representation for a specific digit in the specified radix.
Binary property ID_Start.
Definition: uchar.h:268
Corrected name from NameAliases.txt.
Definition: uchar.h:1961
Binary property Changes_When_Titlecased.
Definition: uchar.h:427
Enumerated property Line_Break.
Definition: uchar.h:574
One more than the highest normal UJoiningType value.
Definition: uchar.h:2064
U_CAPI UBool u_isalpha(UChar32 c)
Determines whether the specified code point is a letter character.
U_CAPI const char * u_getPropertyValueName(UProperty property, int32_t value, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property value, as given in the Unicode database file PropertyVal...
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:614
U_CAPI UBool u_iscntrl(UChar32 c)
Determines whether the specified code point is a control character (as defined by this function)...
Deprecated string property ISO_Comment.
Definition: uchar.h:706
Binary property of strings RGI_Emoji_Flag_Sequence.
Definition: uchar.h:514
U_CAPI UBool u_isIDPart(UChar32 c)
Determines if the specified character is permissible in an identifier according to Java...
U_CAPI UBool u_islower(UChar32 c)
Determines whether the specified code point has the general category &quot;Ll&quot; (lowercase letter)...
Binary property Default_Ignorable_Code_Point (new in Unicode 3.2).
Definition: uchar.h:228
One more than the highest normal UBlockCode value.
Definition: uchar.h:1895
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:1956
int32_t u_getISOComment(UChar32 c, char *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Returns an empty string.
One higher than the last enum UCharCategory constant.
Definition: uchar.h:853
Binary property of strings RGI_Emoji_Tag_Sequence.
Definition: uchar.h:521
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2312
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition: uchar.h:387
One more than the last constant for Unicode properties with unusual value types.
Definition: uchar.h:763
U_CAPI UBool u_isprint(UChar32 c)
Determines whether the specified code point is a printable character.
UBidiPairedBracketType
Bidi Paired Bracket Type constants.
Definition: uchar.h:1043
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:1967
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition: uchar.h:2549
U_CAPI UBool u_isdigit(UChar32 c)
Determines whether the specified code point is a digit character according to Java.
String property Age.
Definition: uchar.h:694
int8_t UBool
The ICU boolean type, a signed-byte integer.
Definition: umachine.h:269
U_CAPI UChar32 u_charFromName(UCharNameChoice nameChoice, const char *name, UErrorCode *pErrorCode)
Find a Unicode character by its name and return its code point value.
Enumerated property NFKD_Quick_Check.
Definition: uchar.h:589
U_CAPI UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.