ICU 75.1  75.1
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 <stdbool.h>
29 #include "unicode/utypes.h"
30 #include "unicode/stringoptions.h"
31 #include "unicode/ucpmap.h"
32 
33 #if !defined(USET_DEFINED) && !defined(U_IN_DOXYGEN)
34 
35 #define USET_DEFINED
36 
45 typedef struct USet USet;
46 
47 #endif
48 
49 
51 
52 /*==========================================================================*/
53 /* Unicode version number */
54 /*==========================================================================*/
64 #define U_UNICODE_VERSION "15.1"
65 
158 #define UCHAR_MIN_VALUE 0
159 
168 #define UCHAR_MAX_VALUE 0x10ffff
169 
174 #define U_MASK(x) ((uint32_t)1<<(x))
175 
196 typedef enum UProperty {
197  /*
198  * Note: UProperty constants are parsed by preparseucd.py.
199  * It matches lines like
200  * UCHAR_<Unicode property name>=<integer>,
201  */
202 
203  /* Note: Place UCHAR_ALPHABETIC before UCHAR_BINARY_START so that
204  debuggers display UCHAR_ALPHABETIC as the symbolic name for 0,
205  rather than UCHAR_BINARY_START. Likewise for other *_START
206  identifiers. */
207 
536 #ifndef U_HIDE_DRAFT_API
556 #endif // U_HIDE_DRAFT_API
557 #ifndef U_HIDE_DEPRECATED_API
563 #endif // U_HIDE_DEPRECATED_API
564 
572  UCHAR_BLOCK=0x1001,
600  UCHAR_SCRIPT=0x100A,
672 #ifndef U_HIDE_DRAFT_API
680 #endif // U_HIDE_DRAFT_API
681 #ifndef U_HIDE_DEPRECATED_API
687 #endif // U_HIDE_DEPRECATED_API
688 
700 #ifndef U_HIDE_DEPRECATED_API
706 #endif // U_HIDE_DEPRECATED_API
707 
713 #ifndef U_HIDE_DEPRECATED_API
719 #endif // U_HIDE_DEPRECATED_API
720 
723  UCHAR_AGE=0x4000,
732 #ifndef U_HIDE_DEPRECATED_API
736 #endif /* U_HIDE_DEPRECATED_API */
742  UCHAR_NAME=0x4005,
758 #ifndef U_HIDE_DEPRECATED_API
764 #endif /* U_HIDE_DEPRECATED_API */
771 #ifndef U_HIDE_DEPRECATED_API
777 #endif // U_HIDE_DEPRECATED_API
778 
787 #ifndef U_HIDE_DRAFT_API
802 #endif // U_HIDE_DRAFT_API
803 #ifndef U_HIDE_DEPRECATED_API
809 #endif // U_HIDE_DEPRECATED_API
810 
812  UCHAR_INVALID_CODE = -1
814 
820 typedef enum UCharCategory
821 {
822  /*
823  * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
824  * It matches pairs of lines like
825  * / ** <Unicode 2-letter General_Category value> comment... * /
826  * U_<[A-Z_]+> = <integer>,
827  */
828 
900 
915 #define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES)
916 
918 #define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER)
920 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
922 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
924 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
926 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
927 
929 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
931 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
933 #define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK)
934 
936 #define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER)
938 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
940 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
941 
943 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
945 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
947 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
948 
950 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
952 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
954 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
956 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
957 
959 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
961 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
963 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
965 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
967 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
968 
970 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
972 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
974 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
976 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
977 
979 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
981 #define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION)
982 
983 
985 #define U_GC_L_MASK \
986  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
987 
989 #define U_GC_LC_MASK \
990  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
991 
993 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
994 
996 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
997 
999 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
1000 
1002 #define U_GC_C_MASK \
1003  (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
1004 
1006 #define U_GC_P_MASK \
1007  (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
1008  U_GC_PI_MASK|U_GC_PF_MASK)
1009 
1011 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
1012 
1017 typedef enum UCharDirection {
1018  /*
1019  * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
1020  * It matches pairs of lines like
1021  * / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
1022  * U_<[A-Z_]+> = <integer>,
1023  */
1024 
1071 #ifndef U_HIDE_DEPRECATED_API
1079 #endif // U_HIDE_DEPRECATED_API
1081 
1089  /*
1090  * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
1091  * It matches lines like
1092  * U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
1093  */
1094 
1101 #ifndef U_HIDE_DEPRECATED_API
1108  U_BPT_COUNT /* 3 */
1109 #endif // U_HIDE_DEPRECATED_API
1111 
1117  /*
1118  * Note: UBlockCode constants are parsed by preparseucd.py.
1119  * It matches lines like
1120  * UBLOCK_<Unicode Block value name> = <integer>,
1121  */
1122 
1124  UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
1125 
1127  UBLOCK_BASIC_LATIN = 1, /*[0000]*/
1128 
1131 
1134 
1137 
1139  UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
1140 
1143 
1146 
1151  UBLOCK_GREEK =8, /*[0370]*/
1152 
1154  UBLOCK_CYRILLIC =9, /*[0400]*/
1155 
1157  UBLOCK_ARMENIAN =10, /*[0530]*/
1158 
1160  UBLOCK_HEBREW =11, /*[0590]*/
1161 
1163  UBLOCK_ARABIC =12, /*[0600]*/
1164 
1166  UBLOCK_SYRIAC =13, /*[0700]*/
1167 
1169  UBLOCK_THAANA =14, /*[0780]*/
1170 
1172  UBLOCK_DEVANAGARI =15, /*[0900]*/
1173 
1175  UBLOCK_BENGALI =16, /*[0980]*/
1176 
1178  UBLOCK_GURMUKHI =17, /*[0A00]*/
1179 
1181  UBLOCK_GUJARATI =18, /*[0A80]*/
1182 
1184  UBLOCK_ORIYA =19, /*[0B00]*/
1185 
1187  UBLOCK_TAMIL =20, /*[0B80]*/
1188 
1190  UBLOCK_TELUGU =21, /*[0C00]*/
1191 
1193  UBLOCK_KANNADA =22, /*[0C80]*/
1194 
1196  UBLOCK_MALAYALAM =23, /*[0D00]*/
1197 
1199  UBLOCK_SINHALA =24, /*[0D80]*/
1200 
1202  UBLOCK_THAI =25, /*[0E00]*/
1203 
1205  UBLOCK_LAO =26, /*[0E80]*/
1206 
1208  UBLOCK_TIBETAN =27, /*[0F00]*/
1209 
1211  UBLOCK_MYANMAR =28, /*[1000]*/
1212 
1214  UBLOCK_GEORGIAN =29, /*[10A0]*/
1215 
1217  UBLOCK_HANGUL_JAMO =30, /*[1100]*/
1218 
1220  UBLOCK_ETHIOPIC =31, /*[1200]*/
1221 
1223  UBLOCK_CHEROKEE =32, /*[13A0]*/
1224 
1227 
1229  UBLOCK_OGHAM =34, /*[1680]*/
1230 
1232  UBLOCK_RUNIC =35, /*[16A0]*/
1233 
1235  UBLOCK_KHMER =36, /*[1780]*/
1236 
1238  UBLOCK_MONGOLIAN =37, /*[1800]*/
1239 
1242 
1244  UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
1245 
1248 
1251 
1253  UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
1254 
1260 
1263 
1265  UBLOCK_NUMBER_FORMS =45, /*[2150]*/
1266 
1268  UBLOCK_ARROWS =46, /*[2190]*/
1269 
1272 
1275 
1277  UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
1278 
1281 
1284 
1286  UBLOCK_BOX_DRAWING =52, /*[2500]*/
1287 
1289  UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
1290 
1292  UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
1293 
1296 
1298  UBLOCK_DINGBATS =56, /*[2700]*/
1299 
1301  UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
1302 
1305 
1307  UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
1308 
1311 
1314 
1316  UBLOCK_HIRAGANA =62, /*[3040]*/
1317 
1319  UBLOCK_KATAKANA =63, /*[30A0]*/
1320 
1322  UBLOCK_BOPOMOFO =64, /*[3100]*/
1323 
1326 
1328  UBLOCK_KANBUN =66, /*[3190]*/
1329 
1332 
1335 
1338 
1341 
1344 
1346  UBLOCK_YI_SYLLABLES =72, /*[A000]*/
1347 
1349  UBLOCK_YI_RADICALS =73, /*[A490]*/
1350 
1352  UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
1353 
1355  UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
1356 
1359 
1361  UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
1362 
1372  UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
1383 
1386 
1389 
1392 
1395 
1398 
1401 
1404 
1406  UBLOCK_SPECIALS =86, /*[FFF0]*/
1407 
1410 
1411  /* New blocks in Unicode 3.1 */
1412 
1414  UBLOCK_OLD_ITALIC = 88, /*[10300]*/
1416  UBLOCK_GOTHIC = 89, /*[10330]*/
1418  UBLOCK_DESERET = 90, /*[10400]*/
1422  UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
1430  UBLOCK_TAGS = 96, /*[E0000]*/
1431 
1432  /* New blocks in Unicode 3.2 */
1433 
1442  UBLOCK_TAGALOG = 98, /*[1700]*/
1444  UBLOCK_HANUNOO = 99, /*[1720]*/
1446  UBLOCK_BUHID = 100, /*[1740]*/
1448  UBLOCK_TAGBANWA = 101, /*[1760]*/
1462  UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
1467 
1468  /* New blocks in Unicode 4 */
1469 
1471  UBLOCK_LIMBU = 111, /*[1900]*/
1473  UBLOCK_TAI_LE = 112, /*[1950]*/
1475  UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
1477  UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
1483  UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
1485  UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
1487  UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
1489  UBLOCK_UGARITIC = 120, /*[10380]*/
1491  UBLOCK_SHAVIAN = 121, /*[10450]*/
1493  UBLOCK_OSMANYA = 122, /*[10480]*/
1495  UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
1497  UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
1500 
1501  /* New blocks in Unicode 4.1 */
1502 
1506  UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
1508  UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
1510  UBLOCK_BUGINESE = 129, /*[1A00]*/
1512  UBLOCK_CJK_STROKES = 130, /*[31C0]*/
1516  UBLOCK_COPTIC = 132, /*[2C80]*/
1518  UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
1520  UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
1522  UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
1524  UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
1526  UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
1530  UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
1532  UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
1538  UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
1540  UBLOCK_TIFINAGH = 144, /*[2D30]*/
1542  UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
1543 
1544  /* New blocks in Unicode 5.0 */
1545 
1547  UBLOCK_NKO = 146, /*[07C0]*/
1549  UBLOCK_BALINESE = 147, /*[1B00]*/
1551  UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
1553  UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
1555  UBLOCK_PHAGS_PA = 150, /*[A840]*/
1557  UBLOCK_PHOENICIAN = 151, /*[10900]*/
1559  UBLOCK_CUNEIFORM = 152, /*[12000]*/
1563  UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
1564 
1565  /* New blocks in Unicode 5.1 */
1566 
1568  UBLOCK_SUNDANESE = 155, /*[1B80]*/
1570  UBLOCK_LEPCHA = 156, /*[1C00]*/
1572  UBLOCK_OL_CHIKI = 157, /*[1C50]*/
1574  UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
1576  UBLOCK_VAI = 159, /*[A500]*/
1578  UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
1580  UBLOCK_SAURASHTRA = 161, /*[A880]*/
1582  UBLOCK_KAYAH_LI = 162, /*[A900]*/
1584  UBLOCK_REJANG = 163, /*[A930]*/
1586  UBLOCK_CHAM = 164, /*[AA00]*/
1588  UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
1590  UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
1592  UBLOCK_LYCIAN = 167, /*[10280]*/
1594  UBLOCK_CARIAN = 168, /*[102A0]*/
1596  UBLOCK_LYDIAN = 169, /*[10920]*/
1598  UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
1600  UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
1601 
1602  /* New blocks in Unicode 5.2 */
1603 
1605  UBLOCK_SAMARITAN = 172, /*[0800]*/
1609  UBLOCK_TAI_THAM = 174, /*[1A20]*/
1611  UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
1613  UBLOCK_LISU = 176, /*[A4D0]*/
1615  UBLOCK_BAMUM = 177, /*[A6A0]*/
1619  UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
1623  UBLOCK_JAVANESE = 181, /*[A980]*/
1625  UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
1627  UBLOCK_TAI_VIET = 183, /*[AA80]*/
1629  UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
1633  UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
1635  UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
1637  UBLOCK_AVESTAN = 188, /*[10B00]*/
1641  UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
1643  UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
1645  UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
1647  UBLOCK_KAITHI = 193, /*[11080]*/
1649  UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
1656 
1657  /* New blocks in Unicode 6.0 */
1658 
1660  UBLOCK_MANDAIC = 198, /*[0840]*/
1662  UBLOCK_BATAK = 199, /*[1BC0]*/
1664  UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
1666  UBLOCK_BRAHMI = 201, /*[11000]*/
1668  UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
1670  UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
1672  UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
1676  UBLOCK_EMOTICONS = 206, /*[1F600]*/
1680  UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
1683 
1684  /* New blocks in Unicode 6.1 */
1685 
1687  UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
1691  UBLOCK_CHAKMA = 212, /*[11100]*/
1695  UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
1697  UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
1699  UBLOCK_MIAO = 216, /*[16F00]*/
1701  UBLOCK_SHARADA = 217, /*[11180]*/
1703  UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
1707  UBLOCK_TAKRI = 220, /*[11680]*/
1708 
1709  /* New blocks in Unicode 7.0 */
1710 
1712  UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
1714  UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
1716  UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
1720  UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
1722  UBLOCK_ELBASAN = 226, /*[10500]*/
1726  UBLOCK_GRANTHA = 228, /*[11300]*/
1728  UBLOCK_KHOJKI = 229, /*[11200]*/
1730  UBLOCK_KHUDAWADI = 230, /*[112B0]*/
1732  UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
1734  UBLOCK_LINEAR_A = 232, /*[10600]*/
1736  UBLOCK_MAHAJANI = 233, /*[11150]*/
1738  UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
1740  UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
1742  UBLOCK_MODI = 236, /*[11600]*/
1744  UBLOCK_MRO = 237, /*[16A40]*/
1746  UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
1748  UBLOCK_NABATAEAN = 239, /*[10880]*/
1750  UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
1752  UBLOCK_OLD_PERMIC = 241, /*[10350]*/
1754  UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
1756  UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
1758  UBLOCK_PALMYRENE = 244, /*[10860]*/
1760  UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
1762  UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
1766  UBLOCK_SIDDHAM = 248, /*[11580]*/
1770  UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
1772  UBLOCK_TIRHUTA = 251, /*[11480]*/
1774  UBLOCK_WARANG_CITI = 252, /*[118A0]*/
1775 
1776  /* New blocks in Unicode 8.0 */
1777 
1779  UBLOCK_AHOM = 253, /*[11700]*/
1781  UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
1783  UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
1789  UBLOCK_HATRAN = 258, /*[108E0]*/
1791  UBLOCK_MULTANI = 259, /*[11280]*/
1793  UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
1797  UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/
1798 
1799  /* New blocks in Unicode 9.0 */
1800 
1802  UBLOCK_ADLAM = 263, /*[1E900]*/
1804  UBLOCK_BHAIKSUKI = 264, /*[11C00]*/
1806  UBLOCK_CYRILLIC_EXTENDED_C = 265, /*[1C80]*/
1808  UBLOCK_GLAGOLITIC_SUPPLEMENT = 266, /*[1E000]*/
1812  UBLOCK_MARCHEN = 268, /*[11C70]*/
1814  UBLOCK_MONGOLIAN_SUPPLEMENT = 269, /*[11660]*/
1816  UBLOCK_NEWA = 270, /*[11400]*/
1818  UBLOCK_OSAGE = 271, /*[104B0]*/
1820  UBLOCK_TANGUT = 272, /*[17000]*/
1822  UBLOCK_TANGUT_COMPONENTS = 273, /*[18800]*/
1823 
1824  // New blocks in Unicode 10.0
1825 
1829  UBLOCK_KANA_EXTENDED_A = 275, /*[1B100]*/
1831  UBLOCK_MASARAM_GONDI = 276, /*[11D00]*/
1833  UBLOCK_NUSHU = 277, /*[1B170]*/
1835  UBLOCK_SOYOMBO = 278, /*[11A50]*/
1837  UBLOCK_SYRIAC_SUPPLEMENT = 279, /*[0860]*/
1839  UBLOCK_ZANABAZAR_SQUARE = 280, /*[11A00]*/
1840 
1841  // New blocks in Unicode 11.0
1842 
1844  UBLOCK_CHESS_SYMBOLS = 281, /*[1FA00]*/
1846  UBLOCK_DOGRA = 282, /*[11800]*/
1848  UBLOCK_GEORGIAN_EXTENDED = 283, /*[1C90]*/
1850  UBLOCK_GUNJALA_GONDI = 284, /*[11D60]*/
1852  UBLOCK_HANIFI_ROHINGYA = 285, /*[10D00]*/
1854  UBLOCK_INDIC_SIYAQ_NUMBERS = 286, /*[1EC70]*/
1856  UBLOCK_MAKASAR = 287, /*[11EE0]*/
1858  UBLOCK_MAYAN_NUMERALS = 288, /*[1D2E0]*/
1860  UBLOCK_MEDEFAIDRIN = 289, /*[16E40]*/
1862  UBLOCK_OLD_SOGDIAN = 290, /*[10F00]*/
1864  UBLOCK_SOGDIAN = 291, /*[10F30]*/
1865 
1866  // New blocks in Unicode 12.0
1867 
1871  UBLOCK_ELYMAIC = 293, /*[10FE0]*/
1873  UBLOCK_NANDINAGARI = 294, /*[119A0]*/
1877  UBLOCK_OTTOMAN_SIYAQ_NUMBERS = 296, /*[1ED00]*/
1879  UBLOCK_SMALL_KANA_EXTENSION = 297, /*[1B130]*/
1883  UBLOCK_TAMIL_SUPPLEMENT = 299, /*[11FC0]*/
1885  UBLOCK_WANCHO = 300, /*[1E2C0]*/
1886 
1887  // New blocks in Unicode 13.0
1888 
1890  UBLOCK_CHORASMIAN = 301, /*[10FB0]*/
1894  UBLOCK_DIVES_AKURU = 303, /*[11900]*/
1896  UBLOCK_KHITAN_SMALL_SCRIPT = 304, /*[18B00]*/
1898  UBLOCK_LISU_SUPPLEMENT = 305, /*[11FB0]*/
1902  UBLOCK_TANGUT_SUPPLEMENT = 307, /*[18D00]*/
1904  UBLOCK_YEZIDI = 308, /*[10E80]*/
1905 
1906  // New blocks in Unicode 14.0
1907 
1909  UBLOCK_ARABIC_EXTENDED_B = 309, /*[0870]*/
1911  UBLOCK_CYPRO_MINOAN = 310, /*[12F90]*/
1913  UBLOCK_ETHIOPIC_EXTENDED_B = 311, /*[1E7E0]*/
1915  UBLOCK_KANA_EXTENDED_B = 312, /*[1AFF0]*/
1917  UBLOCK_LATIN_EXTENDED_F = 313, /*[10780]*/
1919  UBLOCK_LATIN_EXTENDED_G = 314, /*[1DF00]*/
1921  UBLOCK_OLD_UYGHUR = 315, /*[10F70]*/
1923  UBLOCK_TANGSA = 316, /*[16A70]*/
1925  UBLOCK_TOTO = 317, /*[1E290]*/
1929  UBLOCK_VITHKUQI = 319, /*[10570]*/
1932 
1933  // New blocks in Unicode 15.0
1934 
1936  UBLOCK_ARABIC_EXTENDED_C = 321, /*[10EC0]*/
1940  UBLOCK_CYRILLIC_EXTENDED_D = 323, /*[1E030]*/
1942  UBLOCK_DEVANAGARI_EXTENDED_A = 324, /*[11B00]*/
1944  UBLOCK_KAKTOVIK_NUMERALS = 325, /*[1D2C0]*/
1946  UBLOCK_KAWI = 326, /*[11F00]*/
1948  UBLOCK_NAG_MUNDARI = 327, /*[1E4D0]*/
1949 
1950  // New block in Unicode 15.1
1951 
1954 
1955 #ifndef U_HIDE_DEPRECATED_API
1963 #endif // U_HIDE_DEPRECATED_API
1964 
1967 };
1968 
1970 typedef enum UBlockCode UBlockCode;
1971 
1979 typedef enum UEastAsianWidth {
1980  /*
1981  * Note: UEastAsianWidth constants are parsed by preparseucd.py.
1982  * It matches lines like
1983  * U_EA_<Unicode East_Asian_Width value name>
1984  */
1985 
1986  U_EA_NEUTRAL, /*[N]*/
1987  U_EA_AMBIGUOUS, /*[A]*/
1988  U_EA_HALFWIDTH, /*[H]*/
1989  U_EA_FULLWIDTH, /*[F]*/
1990  U_EA_NARROW, /*[Na]*/
1991  U_EA_WIDE, /*[W]*/
1992 #ifndef U_HIDE_DEPRECATED_API
1999  U_EA_COUNT
2000 #endif // U_HIDE_DEPRECATED_API
2002 
2014 typedef enum UCharNameChoice {
2017 #ifndef U_HIDE_DEPRECATED_API
2024 #endif /* U_HIDE_DEPRECATED_API */
2029 #ifndef U_HIDE_DEPRECATED_API
2035 #endif // U_HIDE_DEPRECATED_API
2037 
2051 typedef enum UPropertyNameChoice {
2052  U_SHORT_PROPERTY_NAME,
2053  U_LONG_PROPERTY_NAME,
2054 #ifndef U_HIDE_DEPRECATED_API
2060 #endif // U_HIDE_DEPRECATED_API
2062 
2069 typedef enum UDecompositionType {
2070  /*
2071  * Note: UDecompositionType constants are parsed by preparseucd.py.
2072  * It matches lines like
2073  * U_DT_<Unicode Decomposition_Type value name>
2074  */
2075 
2076  U_DT_NONE, /*[none]*/
2077  U_DT_CANONICAL, /*[can]*/
2078  U_DT_COMPAT, /*[com]*/
2079  U_DT_CIRCLE, /*[enc]*/
2080  U_DT_FINAL, /*[fin]*/
2081  U_DT_FONT, /*[font]*/
2082  U_DT_FRACTION, /*[fra]*/
2083  U_DT_INITIAL, /*[init]*/
2084  U_DT_ISOLATED, /*[iso]*/
2085  U_DT_MEDIAL, /*[med]*/
2086  U_DT_NARROW, /*[nar]*/
2087  U_DT_NOBREAK, /*[nb]*/
2088  U_DT_SMALL, /*[sml]*/
2089  U_DT_SQUARE, /*[sqr]*/
2090  U_DT_SUB, /*[sub]*/
2091  U_DT_SUPER, /*[sup]*/
2092  U_DT_VERTICAL, /*[vert]*/
2093  U_DT_WIDE, /*[wide]*/
2094 #ifndef U_HIDE_DEPRECATED_API
2101  U_DT_COUNT /* 18 */
2102 #endif // U_HIDE_DEPRECATED_API
2104 
2111 typedef enum UJoiningType {
2112  /*
2113  * Note: UJoiningType constants are parsed by preparseucd.py.
2114  * It matches lines like
2115  * U_JT_<Unicode Joining_Type value name>
2116  */
2117 
2118  U_JT_NON_JOINING, /*[U]*/
2119  U_JT_JOIN_CAUSING, /*[C]*/
2120  U_JT_DUAL_JOINING, /*[D]*/
2121  U_JT_LEFT_JOINING, /*[L]*/
2122  U_JT_RIGHT_JOINING, /*[R]*/
2123  U_JT_TRANSPARENT, /*[T]*/
2124 #ifndef U_HIDE_DEPRECATED_API
2131  U_JT_COUNT /* 6 */
2132 #endif // U_HIDE_DEPRECATED_API
2134 
2141 typedef enum UJoiningGroup {
2142  /*
2143  * Note: UJoiningGroup constants are parsed by preparseucd.py.
2144  * It matches lines like
2145  * U_JG_<Unicode Joining_Group value name>
2146  */
2147 
2148  U_JG_NO_JOINING_GROUP,
2149  U_JG_AIN,
2150  U_JG_ALAPH,
2151  U_JG_ALEF,
2152  U_JG_BEH,
2153  U_JG_BETH,
2154  U_JG_DAL,
2155  U_JG_DALATH_RISH,
2156  U_JG_E,
2157  U_JG_FEH,
2158  U_JG_FINAL_SEMKATH,
2159  U_JG_GAF,
2160  U_JG_GAMAL,
2161  U_JG_HAH,
2163  U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
2164  U_JG_HE,
2165  U_JG_HEH,
2166  U_JG_HEH_GOAL,
2167  U_JG_HETH,
2168  U_JG_KAF,
2169  U_JG_KAPH,
2170  U_JG_KNOTTED_HEH,
2171  U_JG_LAM,
2172  U_JG_LAMADH,
2173  U_JG_MEEM,
2174  U_JG_MIM,
2175  U_JG_NOON,
2176  U_JG_NUN,
2177  U_JG_PE,
2178  U_JG_QAF,
2179  U_JG_QAPH,
2180  U_JG_REH,
2181  U_JG_REVERSED_PE,
2182  U_JG_SAD,
2183  U_JG_SADHE,
2184  U_JG_SEEN,
2185  U_JG_SEMKATH,
2186  U_JG_SHIN,
2187  U_JG_SWASH_KAF,
2188  U_JG_SYRIAC_WAW,
2189  U_JG_TAH,
2190  U_JG_TAW,
2191  U_JG_TEH_MARBUTA,
2192  U_JG_TETH,
2193  U_JG_WAW,
2194  U_JG_YEH,
2195  U_JG_YEH_BARREE,
2196  U_JG_YEH_WITH_TAIL,
2197  U_JG_YUDH,
2198  U_JG_YUDH_HE,
2199  U_JG_ZAIN,
2257 #ifndef U_HIDE_DEPRECATED_API
2264  U_JG_COUNT
2265 #endif // U_HIDE_DEPRECATED_API
2267 
2275  /*
2276  * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2277  * It matches lines like
2278  * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2279  */
2280 
2281  U_GCB_OTHER = 0, /*[XX]*/
2282  U_GCB_CONTROL = 1, /*[CN]*/
2283  U_GCB_CR = 2, /*[CR]*/
2284  U_GCB_EXTEND = 3, /*[EX]*/
2285  U_GCB_L = 4, /*[L]*/
2286  U_GCB_LF = 5, /*[LF]*/
2287  U_GCB_LV = 6, /*[LV]*/
2288  U_GCB_LVT = 7, /*[LVT]*/
2289  U_GCB_T = 8, /*[T]*/
2290  U_GCB_V = 9, /*[V]*/
2292  U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2294  U_GCB_PREPEND = 11, /*[PP]*/
2296  U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2298  U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2300  U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2302  U_GCB_E_MODIFIER = 15, /*[EM]*/
2304  U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2306  U_GCB_ZWJ = 17, /*[ZWJ]*/
2307 
2308 #ifndef U_HIDE_DEPRECATED_API
2315  U_GCB_COUNT = 18
2316 #endif // U_HIDE_DEPRECATED_API
2318 
2326 typedef enum UWordBreakValues {
2327  /*
2328  * Note: UWordBreakValues constants are parsed by preparseucd.py.
2329  * It matches lines like
2330  * U_WB_<Unicode Word_Break value name>
2331  */
2332 
2333  U_WB_OTHER = 0, /*[XX]*/
2334  U_WB_ALETTER = 1, /*[LE]*/
2335  U_WB_FORMAT = 2, /*[FO]*/
2336  U_WB_KATAKANA = 3, /*[KA]*/
2337  U_WB_MIDLETTER = 4, /*[ML]*/
2338  U_WB_MIDNUM = 5, /*[MN]*/
2339  U_WB_NUMERIC = 6, /*[NU]*/
2340  U_WB_EXTENDNUMLET = 7, /*[EX]*/
2342  U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2344  U_WB_EXTEND = 9, /*[Extend]*/
2346  U_WB_LF = 10, /*[LF]*/
2348  U_WB_MIDNUMLET =11, /*[MB]*/
2350  U_WB_NEWLINE =12, /*[NL]*/
2352  U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2354  U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2356  U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2358  U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2360  U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2362  U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2364  U_WB_E_MODIFIER = 19, /*[EM]*/
2366  U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2368  U_WB_ZWJ = 21, /*[ZWJ]*/
2370  U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/
2371 
2372 #ifndef U_HIDE_DEPRECATED_API
2379  U_WB_COUNT = 23
2380 #endif // U_HIDE_DEPRECATED_API
2382 
2389 typedef enum USentenceBreak {
2390  /*
2391  * Note: USentenceBreak constants are parsed by preparseucd.py.
2392  * It matches lines like
2393  * U_SB_<Unicode Sentence_Break value name>
2394  */
2395 
2396  U_SB_OTHER = 0, /*[XX]*/
2397  U_SB_ATERM = 1, /*[AT]*/
2398  U_SB_CLOSE = 2, /*[CL]*/
2399  U_SB_FORMAT = 3, /*[FO]*/
2400  U_SB_LOWER = 4, /*[LO]*/
2401  U_SB_NUMERIC = 5, /*[NU]*/
2402  U_SB_OLETTER = 6, /*[LE]*/
2403  U_SB_SEP = 7, /*[SE]*/
2404  U_SB_SP = 8, /*[SP]*/
2405  U_SB_STERM = 9, /*[ST]*/
2406  U_SB_UPPER = 10, /*[UP]*/
2407  U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2408  U_SB_EXTEND = 12, /*[EX]*/
2409  U_SB_LF = 13, /*[LF]*/
2410  U_SB_SCONTINUE = 14, /*[SC]*/
2411 #ifndef U_HIDE_DEPRECATED_API
2418  U_SB_COUNT = 15
2419 #endif // U_HIDE_DEPRECATED_API
2421 
2428 typedef enum ULineBreak {
2429  /*
2430  * Note: ULineBreak constants are parsed by preparseucd.py.
2431  * It matches lines like
2432  * U_LB_<Unicode Line_Break value name>
2433  */
2434 
2435  U_LB_UNKNOWN = 0, /*[XX]*/
2436  U_LB_AMBIGUOUS = 1, /*[AI]*/
2437  U_LB_ALPHABETIC = 2, /*[AL]*/
2438  U_LB_BREAK_BOTH = 3, /*[B2]*/
2439  U_LB_BREAK_AFTER = 4, /*[BA]*/
2440  U_LB_BREAK_BEFORE = 5, /*[BB]*/
2441  U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2442  U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2443  U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2444  U_LB_COMBINING_MARK = 9, /*[CM]*/
2445  U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2446  U_LB_EXCLAMATION = 11, /*[EX]*/
2447  U_LB_GLUE = 12, /*[GL]*/
2448  U_LB_HYPHEN = 13, /*[HY]*/
2449  U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2451  U_LB_INSEPARABLE = 15, /*[IN]*/
2452  U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2453  U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2454  U_LB_LINE_FEED = 17, /*[LF]*/
2455  U_LB_NONSTARTER = 18, /*[NS]*/
2456  U_LB_NUMERIC = 19, /*[NU]*/
2457  U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2458  U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2459  U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2460  U_LB_QUOTATION = 23, /*[QU]*/
2461  U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2462  U_LB_SURROGATE = 25, /*[SG]*/
2463  U_LB_SPACE = 26, /*[SP]*/
2464  U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2465  U_LB_ZWSPACE = 28, /*[ZW]*/
2467  U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2469  U_LB_WORD_JOINER = 30, /*[WJ]*/
2471  U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2473  U_LB_H3 = 32, /*[H3]*/
2475  U_LB_JL = 33, /*[JL]*/
2477  U_LB_JT = 34, /*[JT]*/
2479  U_LB_JV = 35, /*[JV]*/
2481  U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2483  U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2485  U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2487  U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2489  U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2491  U_LB_E_MODIFIER = 41, /*[EM]*/
2493  U_LB_ZWJ = 42, /*[ZWJ]*/
2495  U_LB_AKSARA = 43, /*[AK]*/
2497  U_LB_AKSARA_PREBASE = 44, /*[AP]*/
2499  U_LB_AKSARA_START = 45, /*[AS]*/
2501  U_LB_VIRAMA_FINAL = 46, /*[VF]*/
2503  U_LB_VIRAMA = 47, /*[VI]*/
2504 #ifndef U_HIDE_DEPRECATED_API
2511  U_LB_COUNT = 48
2512 #endif // U_HIDE_DEPRECATED_API
2514 
2521 typedef enum UNumericType {
2522  /*
2523  * Note: UNumericType constants are parsed by preparseucd.py.
2524  * It matches lines like
2525  * U_NT_<Unicode Numeric_Type value name>
2526  */
2527 
2528  U_NT_NONE, /*[None]*/
2529  U_NT_DECIMAL, /*[de]*/
2530  U_NT_DIGIT, /*[di]*/
2531  U_NT_NUMERIC, /*[nu]*/
2532 #ifndef U_HIDE_DEPRECATED_API
2539  U_NT_COUNT
2540 #endif // U_HIDE_DEPRECATED_API
2542 
2549 typedef enum UHangulSyllableType {
2550  /*
2551  * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2552  * It matches lines like
2553  * U_HST_<Unicode Hangul_Syllable_Type value name>
2554  */
2555 
2556  U_HST_NOT_APPLICABLE, /*[NA]*/
2557  U_HST_LEADING_JAMO, /*[L]*/
2558  U_HST_VOWEL_JAMO, /*[V]*/
2559  U_HST_TRAILING_JAMO, /*[T]*/
2560  U_HST_LV_SYLLABLE, /*[LV]*/
2561  U_HST_LVT_SYLLABLE, /*[LVT]*/
2562 #ifndef U_HIDE_DEPRECATED_API
2569  U_HST_COUNT
2570 #endif // U_HIDE_DEPRECATED_API
2572 
2580  /*
2581  * Note: UIndicPositionalCategory constants are parsed by preparseucd.py.
2582  * It matches lines like
2583  * U_INPC_<Unicode Indic_Positional_Category value name>
2584  */
2585 
2619 
2627  /*
2628  * Note: UIndicSyllabicCategory constants are parsed by preparseucd.py.
2629  * It matches lines like
2630  * U_INSC_<Unicode Indic_Syllabic_Category value name>
2631  */
2632 
2706 
2713 typedef enum UVerticalOrientation {
2714  /*
2715  * Note: UVerticalOrientation constants are parsed by preparseucd.py.
2716  * It matches lines like
2717  * U_VO_<Unicode Vertical_Orientation value name>
2718  */
2719 
2729 
2730 #ifndef U_HIDE_DRAFT_API
2738 typedef enum UIdentifierStatus {
2739  /*
2740  * Note: UIdentifierStatus constants are parsed by preparseucd.py.
2741  * It matches lines like
2742  * U_ID_STATUS_<Unicode Identifier_Status value name>
2743  */
2744 
2750 
2758 typedef enum UIdentifierType {
2759  /*
2760  * Note: UIdentifierType constants are parsed by preparseucd.py.
2761  * It matches lines like
2762  * U_ID_TYPE_<Unicode Identifier_Type value name>
2763  */
2764 
2790 #endif // U_HIDE_DRAFT_API
2791 
2819 U_CAPI UBool U_EXPORT2
2821 
2846 U_CAPI UBool U_EXPORT2
2847 u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which);
2848 
2864 U_CAPI const USet * U_EXPORT2
2866 
2879 U_CAPI UBool U_EXPORT2
2881 
2894 U_CAPI UBool U_EXPORT2
2896 
2909 U_CAPI UBool U_EXPORT2
2911 
2930 U_CAPI UBool U_EXPORT2
2932 
2971 U_CAPI int32_t U_EXPORT2
2973 
2992 U_CAPI int32_t U_EXPORT2
2994 
3021 U_CAPI int32_t U_EXPORT2
3023 
3039 U_CAPI const UCPMap * U_EXPORT2
3041 
3064 U_CAPI double U_EXPORT2
3066 
3074 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
3075 
3099 U_CAPI UBool U_EXPORT2
3101 
3126 U_CAPI UBool U_EXPORT2
3128 
3143 U_CAPI UBool U_EXPORT2
3145 
3164 U_CAPI UBool U_EXPORT2
3166 
3185 U_CAPI UBool U_EXPORT2
3187 
3206 U_CAPI UBool U_EXPORT2
3208 
3229 U_CAPI UBool U_EXPORT2
3231 
3245 U_CAPI UBool U_EXPORT2
3247 
3264 U_CAPI UBool U_EXPORT2
3266 
3292 U_CAPI UBool U_EXPORT2
3294 
3317 U_CAPI UBool U_EXPORT2
3319 
3338 U_CAPI UBool U_EXPORT2
3340 
3359 U_CAPI UBool U_EXPORT2
3361 
3399 U_CAPI UBool U_EXPORT2
3401 
3423 U_CAPI UBool U_EXPORT2
3425 
3438 U_CAPI UBool U_EXPORT2
3440 
3456 U_CAPI UBool U_EXPORT2
3458 
3476 U_CAPI UBool U_EXPORT2
3478 
3495 U_CAPI UCharDirection U_EXPORT2
3497 
3513 U_CAPI UBool U_EXPORT2
3515 
3535 U_CAPI UChar32 U_EXPORT2
3537 
3554 U_CAPI UChar32 U_EXPORT2
3556 
3568 U_CAPI int8_t U_EXPORT2
3570 
3584 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3585 
3603 typedef UBool U_CALLCONV
3604 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3605 
3625 U_CAPI void U_EXPORT2
3626 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3627 
3628 #if !UCONFIG_NO_NORMALIZATION
3629 
3637 U_CAPI uint8_t U_EXPORT2
3639 
3640 #endif
3641 
3665 U_CAPI int32_t U_EXPORT2
3667 
3677 U_CAPI UBlockCode U_EXPORT2
3679 
3712 U_CAPI int32_t U_EXPORT2
3714  char *buffer, int32_t bufferLength,
3715  UErrorCode *pErrorCode);
3716 
3717 #ifndef U_HIDE_DEPRECATED_API
3736 U_DEPRECATED int32_t U_EXPORT2
3738  char *dest, int32_t destCapacity,
3739  UErrorCode *pErrorCode);
3740 #endif /* U_HIDE_DEPRECATED_API */
3741 
3762 U_CAPI UChar32 U_EXPORT2
3764  const char *name,
3765  UErrorCode *pErrorCode);
3766 
3784 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3785  UChar32 code,
3786  UCharNameChoice nameChoice,
3787  const char *name,
3788  int32_t length);
3789 
3811 U_CAPI void U_EXPORT2
3813  UEnumCharNamesFn *fn,
3814  void *context,
3815  UCharNameChoice nameChoice,
3816  UErrorCode *pErrorCode);
3817 
3849 U_CAPI const char* U_EXPORT2
3851  UPropertyNameChoice nameChoice);
3852 
3872 U_CAPI UProperty U_EXPORT2
3873 u_getPropertyEnum(const char* alias);
3874 
3922 U_CAPI const char* U_EXPORT2
3924  int32_t value,
3925  UPropertyNameChoice nameChoice);
3926 
3958 U_CAPI int32_t U_EXPORT2
3960  const char* alias);
3961 
3976 U_CAPI UBool U_EXPORT2
3978 
3993 U_CAPI UBool U_EXPORT2
3995 
3996 #ifndef U_HIDE_DRAFT_API
4010 U_CAPI bool U_EXPORT2
4012 
4045 U_CAPI int32_t U_EXPORT2
4046 u_getIDTypes(UChar32 c, UIdentifierType *types, int32_t capacity, UErrorCode *pErrorCode);
4047 #endif // U_HIDE_DRAFT_API
4048 
4069 U_CAPI UBool U_EXPORT2
4071 
4088 U_CAPI UBool U_EXPORT2
4090 
4109 U_CAPI UBool U_EXPORT2
4111 
4134 U_CAPI UChar32 U_EXPORT2
4136 
4159 U_CAPI UChar32 U_EXPORT2
4161 
4184 U_CAPI UChar32 U_EXPORT2
4186 
4209 U_CAPI UChar32 U_EXPORT2
4210 u_foldCase(UChar32 c, uint32_t options);
4211 
4250 U_CAPI int32_t U_EXPORT2
4251 u_digit(UChar32 ch, int8_t radix);
4252 
4281 U_CAPI UChar32 U_EXPORT2
4282 u_forDigit(int32_t digit, int8_t radix);
4283 
4298 U_CAPI void U_EXPORT2
4299 u_charAge(UChar32 c, UVersionInfo versionArray);
4300 
4312 U_CAPI void U_EXPORT2
4314 
4315 #if !UCONFIG_NO_NORMALIZATION
4337 U_CAPI int32_t U_EXPORT2
4338 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
4339 
4340 #endif
4341 
4342 
4344 
4345 #endif /*_UCHAR*/
4346 /*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:834
C API: Bit set option bit constants for various string and character processing functions.
U_CAPI UBool u_isupper(UChar32 c)
Determines whether the specified code point has the general category "Lu" (uppercase letter).
U_CAPI UBool u_isUAlphabetic(UChar32 c)
Check if a code point has the Alphabetic Unicode property.
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:1116
@ UBLOCK_NANDINAGARI
Definition: uchar.h:1873
@ UBLOCK_MRO
Definition: uchar.h:1744
@ UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED
Definition: uchar.h:1607
@ UBLOCK_COUNT
One more than the highest normal UBlockCode value.
Definition: uchar.h:1962
@ UBLOCK_OPTICAL_CHARACTER_RECOGNITION
Definition: uchar.h:1280
@ UBLOCK_LATIN_EXTENDED_G
Definition: uchar.h:1919
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B
Definition: uchar.h:1426
@ UBLOCK_INSCRIPTIONAL_PAHLAVI
Definition: uchar.h:1641
@ UBLOCK_KHMER
Definition: uchar.h:1235
@ UBLOCK_BHAIKSUKI
Definition: uchar.h:1804
@ UBLOCK_DUPLOYAN
Definition: uchar.h:1720
@ UBLOCK_MEDEFAIDRIN
Definition: uchar.h:1860
@ UBLOCK_BALINESE
Definition: uchar.h:1549
@ UBLOCK_HEBREW
Definition: uchar.h:1160
@ UBLOCK_YIJING_HEXAGRAM_SYMBOLS
Definition: uchar.h:1481
@ UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS
Definition: uchar.h:1795
@ UBLOCK_CYRILLIC
Definition: uchar.h:1154
@ UBLOCK_ARABIC_EXTENDED_B
Definition: uchar.h:1909
@ UBLOCK_SPECIALS
Definition: uchar.h:1406
@ UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS
Definition: uchar.h:1409
@ UBLOCK_NEWA
Definition: uchar.h:1816
@ UBLOCK_ARROWS
Definition: uchar.h:1268
@ UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION
Definition: uchar.h:1313
@ UBLOCK_TAKRI
Definition: uchar.h:1707
@ UBLOCK_BOX_DRAWING
Definition: uchar.h:1286
@ UBLOCK_CURRENCY_SYMBOLS
Definition: uchar.h:1253
@ UBLOCK_MISCELLANEOUS_SYMBOLS
Definition: uchar.h:1295
@ UBLOCK_ANCIENT_GREEK_NUMBERS
Definition: uchar.h:1506
@ UBLOCK_OLD_PERMIC
Definition: uchar.h:1752
@ UBLOCK_CJK_STROKES
Definition: uchar.h:1512
@ UBLOCK_PALMYRENE
Definition: uchar.h:1758
@ UBLOCK_OGHAM
Definition: uchar.h:1229
@ UBLOCK_TANGSA
Definition: uchar.h:1923
@ UBLOCK_SUTTON_SIGNWRITING
Definition: uchar.h:1797
@ UBLOCK_ANATOLIAN_HIEROGLYPHS
Definition: uchar.h:1781
@ UBLOCK_SYMBOLS_FOR_LEGACY_COMPUTING
Definition: uchar.h:1900
@ UBLOCK_SUPPLEMENTAL_PUNCTUATION
Definition: uchar.h:1536
@ UBLOCK_MEROITIC_HIEROGLYPHS
Definition: uchar.h:1697
@ UBLOCK_EMOTICONS
Definition: uchar.h:1676
@ UBLOCK_INDIC_SIYAQ_NUMBERS
Definition: uchar.h:1854
@ UBLOCK_KAKTOVIK_NUMERALS
Definition: uchar.h:1944
@ UBLOCK_REJANG
Definition: uchar.h:1584
@ UBLOCK_TAGALOG
Definition: uchar.h:1442
@ UBLOCK_LISU
Definition: uchar.h:1613
@ UBLOCK_COPTIC_EPACT_NUMBERS
Definition: uchar.h:1716
@ UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS
Definition: uchar.h:1250
@ UBLOCK_TAGS
Definition: uchar.h:1430
@ UBLOCK_MIAO
Definition: uchar.h:1699
@ UBLOCK_TAMIL_SUPPLEMENT
Definition: uchar.h:1883
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C
Definition: uchar.h:1655
@ UBLOCK_LIMBU
Definition: uchar.h:1471
@ UBLOCK_LYDIAN
Definition: uchar.h:1596
@ UBLOCK_NYIAKENG_PUACHUE_HMONG
Definition: uchar.h:1875
@ UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT
Definition: uchar.h:1534
@ UBLOCK_TAI_THAM
Definition: uchar.h:1609
@ UBLOCK_MAHJONG_TILES
Definition: uchar.h:1598
@ UBLOCK_PLAYING_CARDS
Definition: uchar.h:1672
@ UBLOCK_ETHIOPIC_EXTENDED_B
Definition: uchar.h:1913
@ UBLOCK_TRANSPORT_AND_MAP_SYMBOLS
Definition: uchar.h:1678
@ UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS
Definition: uchar.h:1310
@ UBLOCK_BASIC_LATIN
Definition: uchar.h:1127
@ UBLOCK_GURMUKHI
Definition: uchar.h:1178
@ UBLOCK_SYMBOLS_AND_PICTOGRAPHS_EXTENDED_A
Definition: uchar.h:1881
@ UBLOCK_HATRAN
Definition: uchar.h:1789
@ UBLOCK_PRIVATE_USE_AREA
Same as UBLOCK_PRIVATE_USE.
Definition: uchar.h:1372
@ UBLOCK_ARMENIAN
Definition: uchar.h:1157
@ UBLOCK_HANGUL_JAMO
Definition: uchar.h:1217
@ UBLOCK_VERTICAL_FORMS
Definition: uchar.h:1542
@ UBLOCK_SINHALA_ARCHAIC_NUMBERS
Definition: uchar.h:1768
@ UBLOCK_SOGDIAN
Definition: uchar.h:1864
@ UBLOCK_TIRHUTA
Definition: uchar.h:1772
@ UBLOCK_ELBASAN
Definition: uchar.h:1722
@ UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS
Definition: uchar.h:1424
@ UBLOCK_KHITAN_SMALL_SCRIPT
Definition: uchar.h:1896
@ UBLOCK_NUMBER_FORMS
Definition: uchar.h:1265
@ UBLOCK_CHAM
Definition: uchar.h:1586
@ UBLOCK_TANGUT
Definition: uchar.h:1820
@ UBLOCK_TAMIL
Definition: uchar.h:1187
@ UBLOCK_BLOCK_ELEMENTS
Definition: uchar.h:1289
@ UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT
Definition: uchar.h:1651
@ UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION
Definition: uchar.h:1561
@ UBLOCK_LEPCHA
Definition: uchar.h:1570
@ UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT
Definition: uchar.h:1514
@ UBLOCK_VARIATION_SELECTORS_SUPPLEMENT
Definition: uchar.h:1499
@ UBLOCK_DOMINO_TILES
Definition: uchar.h:1600
@ UBLOCK_COMBINING_DIACRITICAL_MARKS_EXTENDED
Definition: uchar.h:1718
@ UBLOCK_EGYPTIAN_HIEROGLYPHS
Definition: uchar.h:1649
@ UBLOCK_IPA_EXTENSIONS
Definition: uchar.h:1139
@ UBLOCK_TANGUT_COMPONENTS
Definition: uchar.h:1822
@ UBLOCK_CHEROKEE_SUPPLEMENT
Definition: uchar.h:1783
@ UBLOCK_ELYMAIC
Definition: uchar.h:1871
@ UBLOCK_LATIN_EXTENDED_A
Definition: uchar.h:1133
@ UBLOCK_HANUNOO
Definition: uchar.h:1444
@ UBLOCK_LAO
Definition: uchar.h:1205
@ UBLOCK_SUNDANESE_SUPPLEMENT
Definition: uchar.h:1705
@ UBLOCK_CYRILLIC_EXTENDED_A
Definition: uchar.h:1574
@ UBLOCK_HANGUL_JAMO_EXTENDED_B
Definition: uchar.h:1631
@ UBLOCK_SAMARITAN
Definition: uchar.h:1605
@ UBLOCK_SUPPLEMENTAL_ARROWS_A
Definition: uchar.h:1452
@ UBLOCK_DEVANAGARI_EXTENDED_A
Definition: uchar.h:1942
@ UBLOCK_GLAGOLITIC
Definition: uchar.h:1524
@ UBLOCK_INSCRIPTIONAL_PARTHIAN
Definition: uchar.h:1639
@ UBLOCK_WARANG_CITI
Definition: uchar.h:1774
@ UBLOCK_MONGOLIAN
Definition: uchar.h:1238
@ UBLOCK_NKO
Definition: uchar.h:1547
@ UBLOCK_ARABIC_EXTENDED_A
Definition: uchar.h:1687
@ UBLOCK_LYCIAN
Definition: uchar.h:1592
@ UBLOCK_CYRILLIC_EXTENDED_C
Definition: uchar.h:1806
@ UBLOCK_KANA_EXTENDED_B
Definition: uchar.h:1915
@ UBLOCK_DINGBATS
Definition: uchar.h:1298
@ UBLOCK_OTTOMAN_SIYAQ_NUMBERS
Definition: uchar.h:1877
@ UBLOCK_LINEAR_B_SYLLABARY
Definition: uchar.h:1483
@ UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS
Definition: uchar.h:1334
@ UBLOCK_CHAKMA
Definition: uchar.h:1691
@ UBLOCK_SPACING_MODIFIER_LETTERS
Definition: uchar.h:1142
@ UBLOCK_MEROITIC_CURSIVE
Definition: uchar.h:1695
@ UBLOCK_CYRILLIC_EXTENDED_B
Definition: uchar.h:1578
@ UBLOCK_GENERAL_PUNCTUATION
Definition: uchar.h:1247
@ UBLOCK_MONGOLIAN_SUPPLEMENT
Definition: uchar.h:1814
@ UBLOCK_MISCELLANEOUS_TECHNICAL
Definition: uchar.h:1274
@ UBLOCK_TAI_XUAN_JING_SYMBOLS
Definition: uchar.h:1497
@ UBLOCK_NABATAEAN
Definition: uchar.h:1748
@ UBLOCK_TAGBANWA
Definition: uchar.h:1448
@ UBLOCK_CUNEIFORM
Definition: uchar.h:1559
@ UBLOCK_CONTROL_PICTURES
Definition: uchar.h:1277
@ UBLOCK_GREEK
Unicode 3.2 renames this block to "Greek and Coptic".
Definition: uchar.h:1151
@ UBLOCK_HIGH_SURROGATES
Definition: uchar.h:1355
@ UBLOCK_COUNTING_ROD_NUMERALS
Definition: uchar.h:1563
@ UBLOCK_AEGEAN_NUMBERS
Definition: uchar.h:1487
@ UBLOCK_KAWI
Definition: uchar.h:1946
@ UBLOCK_BENGALI
Definition: uchar.h:1175
@ UBLOCK_LATIN_EXTENDED_E
Definition: uchar.h:1732
@ UBLOCK_NAG_MUNDARI
Definition: uchar.h:1948
@ UBLOCK_LINEAR_B_IDEOGRAMS
Definition: uchar.h:1485
@ UBLOCK_PAU_CIN_HAU
Definition: uchar.h:1760
@ UBLOCK_BUHID
Definition: uchar.h:1446
@ UBLOCK_GUNJALA_GONDI
Definition: uchar.h:1850
@ UBLOCK_RUMI_NUMERAL_SYMBOLS
Definition: uchar.h:1645
@ UBLOCK_HIGH_PRIVATE_USE_SURROGATES
Definition: uchar.h:1358
@ UBLOCK_MEETEI_MAYEK_EXTENSIONS
Definition: uchar.h:1693
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_H
Definition: uchar.h:1938
@ UBLOCK_COMMON_INDIC_NUMBER_FORMS
Definition: uchar.h:1617
@ UBLOCK_ZANABAZAR_SQUARE
Definition: uchar.h:1839
@ UBLOCK_SUNDANESE
Definition: uchar.h:1568
@ UBLOCK_ARABIC_EXTENDED_C
Definition: uchar.h:1936
@ UBLOCK_CHESS_SYMBOLS
Definition: uchar.h:1844
@ UBLOCK_ENCLOSED_ALPHANUMERICS
Definition: uchar.h:1283
@ UBLOCK_COMBINING_HALF_MARKS
Definition: uchar.h:1394
@ UBLOCK_MANDAIC
Definition: uchar.h:1660
@ UBLOCK_OLD_UYGHUR
Definition: uchar.h:1921
@ UBLOCK_GLAGOLITIC_SUPPLEMENT
Definition: uchar.h:1808
@ UBLOCK_OSMANYA
Definition: uchar.h:1493
@ UBLOCK_DOGRA
Definition: uchar.h:1846
@ UBLOCK_IMPERIAL_ARAMAIC
Definition: uchar.h:1633
@ UBLOCK_BRAILLE_PATTERNS
Definition: uchar.h:1301
@ UBLOCK_OLD_HUNGARIAN
Definition: uchar.h:1793
@ UBLOCK_MATHEMATICAL_OPERATORS
Definition: uchar.h:1271
@ UBLOCK_KHOJKI
Definition: uchar.h:1728
@ UBLOCK_DIVES_AKURU
Definition: uchar.h:1894
@ UBLOCK_ORIYA
Definition: uchar.h:1184
@ UBLOCK_NO_BLOCK
New No_Block value in Unicode 4.
Definition: uchar.h:1124
@ UBLOCK_YI_RADICALS
Definition: uchar.h:1349
@ UBLOCK_BUGINESE
Definition: uchar.h:1510
@ UBLOCK_TANGUT_SUPPLEMENT
Definition: uchar.h:1902
@ UBLOCK_SMALL_FORM_VARIANTS
Definition: uchar.h:1400
@ UBLOCK_OLD_SOGDIAN
Definition: uchar.h:1862
@ UBLOCK_GEORGIAN_EXTENDED
Definition: uchar.h:1848
@ UBLOCK_TIFINAGH
Definition: uchar.h:1540
@ UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B
Definition: uchar.h:1456
@ UBLOCK_LATIN_EXTENDED_D
Definition: uchar.h:1553
@ UBLOCK_LATIN_EXTENDED_ADDITIONAL
Definition: uchar.h:1241
@ UBLOCK_MEETEI_MAYEK
Definition: uchar.h:1629
@ UBLOCK_EGYPTIAN_HIEROGLYPH_FORMAT_CONTROLS
Definition: uchar.h:1869
@ UBLOCK_MYANMAR
Definition: uchar.h:1211
@ UBLOCK_OL_CHIKI
Definition: uchar.h:1572
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D
Definition: uchar.h:1682
@ UBLOCK_SHORTHAND_FORMAT_CONTROLS
Definition: uchar.h:1764
@ UBLOCK_COMBINING_MARKS_FOR_SYMBOLS
Unicode 3.2 renames this block to "Combining Diacritical Marks for Symbols".
Definition: uchar.h:1259
@ UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS
Definition: uchar.h:1674
@ UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS
Definition: uchar.h:1226
@ UBLOCK_BOPOMOFO
Definition: uchar.h:1322
@ UBLOCK_KATAKANA_PHONETIC_EXTENSIONS
Definition: uchar.h:1460
@ UBLOCK_MARCHEN
Definition: uchar.h:1812
@ UBLOCK_MODI
Definition: uchar.h:1742
@ UBLOCK_YEZIDI
Definition: uchar.h:1904
@ UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS
Definition: uchar.h:1458
@ UBLOCK_CYPRO_MINOAN
Definition: uchar.h:1911
@ UBLOCK_MULTANI
Definition: uchar.h:1791
@ UBLOCK_ETHIOPIC_EXTENDED
Definition: uchar.h:1518
@ UBLOCK_AVESTAN
Definition: uchar.h:1637
@ UBLOCK_PRIVATE_USE
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1382
@ UBLOCK_GEORGIAN_SUPPLEMENT
Definition: uchar.h:1522
@ UBLOCK_COPTIC
Definition: uchar.h:1516
@ UBLOCK_HANGUL_COMPATIBILITY_JAMO
Definition: uchar.h:1325
@ UBLOCK_LOW_SURROGATES
Definition: uchar.h:1361
@ UBLOCK_SOYOMBO
Definition: uchar.h:1835
@ UBLOCK_ARABIC_SUPPLEMENT
Definition: uchar.h:1508
@ UBLOCK_HANGUL_SYLLABLES
Definition: uchar.h:1352
@ UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS
Definition: uchar.h:1479
@ UBLOCK_CJK_COMPATIBILITY
Definition: uchar.h:1337
@ UBLOCK_CARIAN
Definition: uchar.h:1594
@ UBLOCK_HANIFI_ROHINGYA
Definition: uchar.h:1852
@ UBLOCK_TIBETAN
Definition: uchar.h:1208
@ UBLOCK_BASSA_VAH
Definition: uchar.h:1712
@ UBLOCK_OLD_TURKIC
Definition: uchar.h:1643
@ UBLOCK_CJK_RADICALS_SUPPLEMENT
Definition: uchar.h:1304
@ UBLOCK_VITHKUQI
Definition: uchar.h:1929
@ UBLOCK_ADLAM
Definition: uchar.h:1802
@ UBLOCK_MASARAM_GONDI
Definition: uchar.h:1831
@ UBLOCK_ARABIC_PRESENTATION_FORMS_B
Definition: uchar.h:1403
@ UBLOCK_RUNIC
Definition: uchar.h:1232
@ UBLOCK_ARABIC_PRESENTATION_FORMS_A
Definition: uchar.h:1391
@ UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT
Definition: uchar.h:1653
@ UBLOCK_CHORASMIAN
Definition: uchar.h:1890
@ UBLOCK_BOPOMOFO_EXTENDED
Definition: uchar.h:1331
@ UBLOCK_IDEOGRAPHIC_SYMBOLS_AND_PUNCTUATION
Definition: uchar.h:1810
@ UBLOCK_LATIN_EXTENDED_F
Definition: uchar.h:1917
@ UBLOCK_KANGXI_RADICALS
Definition: uchar.h:1307
@ UBLOCK_SORA_SOMPENG
Definition: uchar.h:1703
@ UBLOCK_KANA_SUPPLEMENT
Definition: uchar.h:1670
@ UBLOCK_CHEROKEE
Definition: uchar.h:1223
@ UBLOCK_BRAHMI
Definition: uchar.h:1666
@ UBLOCK_SYLOTI_NAGRI
Definition: uchar.h:1538
@ UBLOCK_CYPRIOT_SYLLABARY
Definition: uchar.h:1495
@ UBLOCK_ETHIOPIC_SUPPLEMENT
Definition: uchar.h:1520
@ UBLOCK_KHAROSHTHI
Definition: uchar.h:1526
@ UBLOCK_OLD_ITALIC
Definition: uchar.h:1414
@ UBLOCK_OLD_SOUTH_ARABIAN
Definition: uchar.h:1635
@ UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B
Definition: uchar.h:1466
@ UBLOCK_MALAYALAM
Definition: uchar.h:1196
@ UBLOCK_KHMER_SYMBOLS
Definition: uchar.h:1475
@ UBLOCK_JAVANESE
Definition: uchar.h:1623
@ UBLOCK_BAMUM_SUPPLEMENT
Definition: uchar.h:1668
@ UBLOCK_CYRILLIC_SUPPLEMENT
Definition: uchar.h:1435
@ UBLOCK_MYANMAR_EXTENDED_B
Definition: uchar.h:1746
@ UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION
Definition: uchar.h:1504
@ UBLOCK_BATAK
Definition: uchar.h:1662
@ UBLOCK_SYRIAC_SUPPLEMENT
Definition: uchar.h:1837
@ UBLOCK_PAHAWH_HMONG
Definition: uchar.h:1756
@ UBLOCK_ORNAMENTAL_DINGBATS
Definition: uchar.h:1754
@ UBLOCK_SINHALA
Definition: uchar.h:1199
@ UBLOCK_AHOM
Definition: uchar.h:1779
@ UBLOCK_NEW_TAI_LUE
Definition: uchar.h:1530
@ UBLOCK_SIDDHAM
Definition: uchar.h:1766
@ UBLOCK_OLD_NORTH_ARABIAN
Definition: uchar.h:1750
@ UBLOCK_YI_SYLLABLES
Definition: uchar.h:1346
@ UBLOCK_TELUGU
Definition: uchar.h:1190
@ UBLOCK_DEVANAGARI_EXTENDED
Definition: uchar.h:1619
@ UBLOCK_SHARADA
Definition: uchar.h:1701
@ UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A
Definition: uchar.h:1464
@ UBLOCK_DESERET
Definition: uchar.h:1418
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_I
Definition: uchar.h:1953
@ UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS
Definition: uchar.h:1689
@ UBLOCK_SAURASHTRA
Definition: uchar.h:1580
@ UBLOCK_KANBUN
Definition: uchar.h:1328
@ UBLOCK_HIRAGANA
Definition: uchar.h:1316
@ UBLOCK_KANNADA
Definition: uchar.h:1193
@ UBLOCK_SUPPLEMENTAL_ARROWS_C
Definition: uchar.h:1770
@ UBLOCK_GEORGIAN
Definition: uchar.h:1214
@ UBLOCK_DEVANAGARI
Definition: uchar.h:1172
@ UBLOCK_LISU_SUPPLEMENT
Definition: uchar.h:1898
@ UBLOCK_KATAKANA
Definition: uchar.h:1319
@ UBLOCK_TAI_VIET
Definition: uchar.h:1627
@ UBLOCK_LATIN_EXTENDED_B
Definition: uchar.h:1136
@ UBLOCK_THAANA
Definition: uchar.h:1169
@ UBLOCK_MODIFIER_TONE_LETTERS
Definition: uchar.h:1528
@ UBLOCK_KHUDAWADI
Definition: uchar.h:1730
@ UBLOCK_LINEAR_A
Definition: uchar.h:1734
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS
Definition: uchar.h:1343
@ UBLOCK_GRANTHA
Definition: uchar.h:1726
@ UBLOCK_ETHIOPIC
Definition: uchar.h:1220
@ UBLOCK_SYRIAC
Definition: uchar.h:1166
@ UBLOCK_GREEK_EXTENDED
Definition: uchar.h:1244
@ UBLOCK_CJK_COMPATIBILITY_FORMS
Definition: uchar.h:1397
@ UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS
Definition: uchar.h:1385
@ UBLOCK_MAYAN_NUMERALS
Definition: uchar.h:1858
@ UBLOCK_KAYAH_LI
Definition: uchar.h:1582
@ UBLOCK_VEDIC_EXTENSIONS
Definition: uchar.h:1611
@ UBLOCK_CYRILLIC_EXTENDED_D
Definition: uchar.h:1940
@ UBLOCK_PHAISTOS_DISC
Definition: uchar.h:1590
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A
Definition: uchar.h:1340
@ UBLOCK_LATIN_1_SUPPLEMENT
Definition: uchar.h:1130
@ UBLOCK_PHOENICIAN
Definition: uchar.h:1557
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_G
Definition: uchar.h:1892
@ UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A
Definition: uchar.h:1450
@ UBLOCK_CAUCASIAN_ALBANIAN
Definition: uchar.h:1714
@ UBLOCK_ALCHEMICAL_SYMBOLS
Definition: uchar.h:1680
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_F
Definition: uchar.h:1827
@ UBLOCK_MYANMAR_EXTENDED_A
Definition: uchar.h:1625
@ UBLOCK_EARLY_DYNASTIC_CUNEIFORM
Definition: uchar.h:1787
@ UBLOCK_MENDE_KIKAKUI
Definition: uchar.h:1740
@ UBLOCK_VAI
Definition: uchar.h:1576
@ UBLOCK_PHONETIC_EXTENSIONS
Definition: uchar.h:1477
@ UBLOCK_PHAGS_PA
Definition: uchar.h:1555
@ UBLOCK_OLD_PERSIAN
Definition: uchar.h:1532
@ UBLOCK_KAITHI
Definition: uchar.h:1647
@ UBLOCK_GEOMETRIC_SHAPES
Definition: uchar.h:1292
@ UBLOCK_LATIN_EXTENDED_C
Definition: uchar.h:1551
@ UBLOCK_SUPPLEMENTAL_ARROWS_B
Definition: uchar.h:1454
@ UBLOCK_MAHAJANI
Definition: uchar.h:1736
@ UBLOCK_OSAGE
Definition: uchar.h:1818
@ UBLOCK_INVALID_CODE
Definition: uchar.h:1966
@ UBLOCK_TAI_LE
Definition: uchar.h:1473
@ UBLOCK_MAKASAR
Definition: uchar.h:1856
@ UBLOCK_GOTHIC
Definition: uchar.h:1416
@ UBLOCK_ETHIOPIC_EXTENDED_A
Definition: uchar.h:1664
@ UBLOCK_GEOMETRIC_SHAPES_EXTENDED
Definition: uchar.h:1724
@ UBLOCK_MANICHAEAN
Definition: uchar.h:1738
@ UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED_A
Definition: uchar.h:1927
@ UBLOCK_TOTO
Definition: uchar.h:1925
@ UBLOCK_ALPHABETIC_PRESENTATION_FORMS
Definition: uchar.h:1388
@ UBLOCK_MUSICAL_SYMBOLS
Definition: uchar.h:1422
@ UBLOCK_LETTERLIKE_SYMBOLS
Definition: uchar.h:1262
@ UBLOCK_ARABIC
Definition: uchar.h:1163
@ UBLOCK_ZNAMENNY_MUSICAL_NOTATION
Definition: uchar.h:1931
@ UBLOCK_NUSHU
Definition: uchar.h:1833
@ UBLOCK_CYRILLIC_SUPPLEMENTARY
Unicode 4.0.1 renames the "Cyrillic Supplementary" block to "Cyrillic Supplement".
Definition: uchar.h:1440
@ UBLOCK_HANGUL_JAMO_EXTENDED_A
Definition: uchar.h:1621
@ UBLOCK_BYZANTINE_MUSICAL_SYMBOLS
Definition: uchar.h:1420
@ UBLOCK_ANCIENT_SYMBOLS
Definition: uchar.h:1588
@ UBLOCK_UGARITIC
Definition: uchar.h:1489
@ UBLOCK_SMALL_KANA_EXTENSION
Definition: uchar.h:1879
@ UBLOCK_GUJARATI
Definition: uchar.h:1181
@ UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT
Definition: uchar.h:1428
@ UBLOCK_COMBINING_DIACRITICAL_MARKS
Definition: uchar.h:1145
@ UBLOCK_PSALTER_PAHLAVI
Definition: uchar.h:1762
@ UBLOCK_THAI
Definition: uchar.h:1202
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E
Definition: uchar.h:1785
@ UBLOCK_KANA_EXTENDED_A
Definition: uchar.h:1829
@ UBLOCK_VARIATION_SELECTORS
Definition: uchar.h:1462
@ UBLOCK_BAMUM
Definition: uchar.h:1615
@ UBLOCK_WANCHO
Definition: uchar.h:1885
@ UBLOCK_SHAVIAN
Definition: uchar.h:1491
U_CAPI UBool u_isblank(UChar32 c)
Determines whether the specified code point is a "blank" or "horizontal space", a character that visi...
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.
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:2014
@ U_CHAR_NAME_CHOICE_COUNT
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:2034
@ U_UNICODE_10_CHAR_NAME
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:2023
@ U_CHAR_NAME_ALIAS
Corrected name from NameAliases.txt.
Definition: uchar.h:2028
@ U_EXTENDED_CHAR_NAME
Standard or synthetic character name.
Definition: uchar.h:2026
@ U_UNICODE_CHAR_NAME
Unicode character name (Name property).
Definition: uchar.h:2016
U_CAPI UBool u_isUWhiteSpace(UChar32 c)
Check if a code point has the White_Space Unicode property.
UIdentifierType
Identifier Type constants.
Definition: uchar.h:2758
@ U_ID_TYPE_DEPRECATED
Definition: uchar.h:2768
@ U_ID_TYPE_LIMITED_USE
Definition: uchar.h:2784
@ U_ID_TYPE_DEFAULT_IGNORABLE
Definition: uchar.h:2770
@ U_ID_TYPE_RECOMMENDED
Definition: uchar.h:2788
@ U_ID_TYPE_NOT_CHARACTER
Definition: uchar.h:2766
@ U_ID_TYPE_UNCOMMON_USE
Definition: uchar.h:2782
@ U_ID_TYPE_NOT_NFKC
Definition: uchar.h:2772
@ U_ID_TYPE_NOT_XID
Definition: uchar.h:2774
@ U_ID_TYPE_TECHNICAL
Definition: uchar.h:2780
@ U_ID_TYPE_OBSOLETE
Definition: uchar.h:2778
@ U_ID_TYPE_INCLUSION
Definition: uchar.h:2786
@ U_ID_TYPE_EXCLUSION
Definition: uchar.h:2776
U_CAPI int32_t u_charName(UChar32 code, UCharNameChoice nameChoice, char *buffer, int32_t bufferLength, UErrorCode *pErrorCode)
Retrieve the name of a Unicode character.
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.
U_CAPI bool u_hasIDType(UChar32 c, UIdentifierType type)
Does the set of Identifier_Type values code point c contain the given type?
U_CAPI UBlockCode ublock_getCode(UChar32 c)
Returns the Unicode allocation block that contains the character.
U_CAPI UBool u_hasBinaryProperty(UChar32 c, UProperty which)
Check a binary Unicode property for a code point.
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 UChar32 u_toupper(UChar32 c)
The given character is mapped to its uppercase equivalent according to UnicodeData....
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...
UJoiningType
Joining Type constants.
Definition: uchar.h:2111
@ U_JT_COUNT
One more than the highest normal UJoiningType value.
Definition: uchar.h:2131
U_CAPI UBool u_isbase(UChar32 c)
Non-standard: Determines whether the specified code point is a base character.
UIdentifierStatus
Identifier Status constants.
Definition: uchar.h:2738
@ U_ID_STATUS_ALLOWED
Definition: uchar.h:2748
@ U_ID_STATUS_RESTRICTED
Definition: uchar.h:2746
U_CAPI UBool u_isdigit(UChar32 c)
Determines whether the specified code point is a digit character according to Java.
U_CAPI UBool u_isMirrored(UChar32 c)
Determines whether the code point has the Bidi_Mirrored property.
U_CAPI void u_charAge(UChar32 c, UVersionInfo versionArray)
Get the "age" of the code point.
U_CAPI UBool u_isspace(UChar32 c)
Determines if the specified character is a space character or not.
U_CAPI int32_t u_getIntPropertyMaxValue(UProperty which)
Get the maximum value for an enumerated/integer/binary Unicode property.
UIndicPositionalCategory
Indic Positional Category constants.
Definition: uchar.h:2579
@ U_INPC_BOTTOM
Definition: uchar.h:2589
@ U_INPC_VISUAL_ORDER_LEFT
Definition: uchar.h:2615
@ U_INPC_TOP_AND_BOTTOM_AND_RIGHT
Definition: uchar.h:2607
@ U_INPC_TOP_AND_BOTTOM_AND_LEFT
Definition: uchar.h:2617
@ U_INPC_TOP_AND_RIGHT
Definition: uchar.h:2613
@ U_INPC_BOTTOM_AND_LEFT
Definition: uchar.h:2591
@ U_INPC_LEFT_AND_RIGHT
Definition: uchar.h:2597
@ U_INPC_RIGHT
Definition: uchar.h:2601
@ U_INPC_TOP_AND_LEFT_AND_RIGHT
Definition: uchar.h:2611
@ U_INPC_LEFT
Definition: uchar.h:2595
@ U_INPC_NA
Definition: uchar.h:2587
@ U_INPC_TOP
Definition: uchar.h:2603
@ U_INPC_TOP_AND_LEFT
Definition: uchar.h:2609
@ U_INPC_OVERSTRUCK
Definition: uchar.h:2599
@ U_INPC_TOP_AND_BOTTOM
Definition: uchar.h:2605
@ U_INPC_BOTTOM_AND_RIGHT
Definition: uchar.h:2593
U_CAPI void u_getUnicodeVersion(UVersionInfo versionArray)
Gets the Unicode version information.
UCharDirection
This specifies the language directional property of a character set.
Definition: uchar.h:1017
@ U_SEGMENT_SEPARATOR
S.
Definition: uchar.h:1042
@ U_EUROPEAN_NUMBER_TERMINATOR
ET.
Definition: uchar.h:1034
@ U_BOUNDARY_NEUTRAL
BN.
Definition: uchar.h:1062
@ U_RIGHT_TO_LEFT_ARABIC
AL.
Definition: uchar.h:1052
@ U_RIGHT_TO_LEFT
R.
Definition: uchar.h:1028
@ U_POP_DIRECTIONAL_ISOLATE
PDI.
Definition: uchar.h:1070
@ U_COMMON_NUMBER_SEPARATOR
CS.
Definition: uchar.h:1038
@ U_LEFT_TO_RIGHT
L.
Definition: uchar.h:1026
@ U_DIR_NON_SPACING_MARK
NSM.
Definition: uchar.h:1060
@ U_FIRST_STRONG_ISOLATE
FSI.
Definition: uchar.h:1064
@ U_ARABIC_NUMBER
AN.
Definition: uchar.h:1036
@ U_POP_DIRECTIONAL_FORMAT
PDF.
Definition: uchar.h:1058
@ U_OTHER_NEUTRAL
ON.
Definition: uchar.h:1046
@ U_CHAR_DIRECTION_COUNT
One more than the highest UCharDirection value.
Definition: uchar.h:1078
@ U_WHITE_SPACE_NEUTRAL
WS.
Definition: uchar.h:1044
@ U_RIGHT_TO_LEFT_OVERRIDE
RLO.
Definition: uchar.h:1056
@ U_RIGHT_TO_LEFT_EMBEDDING
RLE.
Definition: uchar.h:1054
@ U_EUROPEAN_NUMBER_SEPARATOR
ES.
Definition: uchar.h:1032
@ U_LEFT_TO_RIGHT_ISOLATE
LRI.
Definition: uchar.h:1066
@ U_LEFT_TO_RIGHT_OVERRIDE
LRO.
Definition: uchar.h:1050
@ U_EUROPEAN_NUMBER
EN.
Definition: uchar.h:1030
@ U_BLOCK_SEPARATOR
B.
Definition: uchar.h:1040
@ U_LEFT_TO_RIGHT_EMBEDDING
LRE.
Definition: uchar.h:1048
@ U_RIGHT_TO_LEFT_ISOLATE
RLI.
Definition: uchar.h:1068
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:2051
@ U_PROPERTY_NAME_CHOICE_COUNT
One more than the highest normal UPropertyNameChoice value.
Definition: uchar.h:2059
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:3784
U_CAPI UBool u_isUUppercase(UChar32 c)
Check if a code point has the Uppercase Unicode property.
U_CAPI int8_t u_charType(UChar32 c)
Returns the general category value for the code point.
U_CAPI UChar32 u_forDigit(int32_t digit, int8_t radix)
Determines the character representation for a specific digit in the specified radix.
U_CAPI UBool u_isprint(UChar32 c)
Determines whether the specified code point is a printable character.
ULineBreak
Line Break constants.
Definition: uchar.h:2428
@ U_LB_E_MODIFIER
Definition: uchar.h:2491
@ U_LB_VIRAMA
Definition: uchar.h:2503
@ U_LB_WORD_JOINER
Definition: uchar.h:2469
@ U_LB_AKSARA
Definition: uchar.h:2495
@ U_LB_H3
Definition: uchar.h:2473
@ U_LB_H2
Definition: uchar.h:2471
@ U_LB_CONDITIONAL_JAPANESE_STARTER
Definition: uchar.h:2483
@ U_LB_NEXT_LINE
Definition: uchar.h:2467
@ U_LB_INSEPARABLE
Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2451
@ U_LB_E_BASE
Definition: uchar.h:2489
@ U_LB_JV
Definition: uchar.h:2479
@ U_LB_VIRAMA_FINAL
Definition: uchar.h:2501
@ U_LB_AKSARA_PREBASE
Definition: uchar.h:2497
@ U_LB_JL
Definition: uchar.h:2475
@ U_LB_CLOSE_PARENTHESIS
Definition: uchar.h:2481
@ U_LB_ZWJ
Definition: uchar.h:2493
@ U_LB_COUNT
One more than the highest normal ULineBreak value.
Definition: uchar.h:2511
@ U_LB_REGIONAL_INDICATOR
Definition: uchar.h:2487
@ U_LB_AKSARA_START
Definition: uchar.h:2499
@ U_LB_HEBREW_LETTER
Definition: uchar.h:2485
@ U_LB_JT
Definition: uchar.h:2477
int32_t u_getISOComment(UChar32 c, char *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Returns an empty string.
U_CAPI const UCPMap * u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode)
Returns an immutable UCPMap for an enumerated/catalog/int-valued property.
U_CAPI UBool u_islower(UChar32 c)
Determines whether the specified code point has the general category "Ll" (lowercase letter).
U_CAPI UBool u_isISOControl(UChar32 c)
Determines whether the specified code point is an ISO control code.
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:821
@ U_FORMAT_CHAR
Cf.
Definition: uchar.h:864
@ U_SPACE_SEPARATOR
Zs.
Definition: uchar.h:856
@ U_MODIFIER_SYMBOL
Sk.
Definition: uchar.h:884
@ U_GENERAL_OTHER_TYPES
Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNE...
Definition: uchar.h:832
@ U_TITLECASE_LETTER
Lt.
Definition: uchar.h:838
@ U_PRIVATE_USE_CHAR
Co.
Definition: uchar.h:866
@ U_OTHER_SYMBOL
So.
Definition: uchar.h:886
@ U_UPPERCASE_LETTER
Lu.
Definition: uchar.h:834
@ U_MODIFIER_LETTER
Lm.
Definition: uchar.h:840
@ U_PARAGRAPH_SEPARATOR
Zp.
Definition: uchar.h:860
@ U_OTHER_PUNCTUATION
Po.
Definition: uchar.h:878
@ U_ENCLOSING_MARK
Me.
Definition: uchar.h:846
@ U_INITIAL_PUNCTUATION
Pi.
Definition: uchar.h:888
@ U_UNASSIGNED
Non-category for unassigned and non-character code points.
Definition: uchar.h:830
@ U_CURRENCY_SYMBOL
Sc.
Definition: uchar.h:882
@ U_COMBINING_SPACING_MARK
Mc.
Definition: uchar.h:848
@ U_CONTROL_CHAR
Cc.
Definition: uchar.h:862
@ U_OTHER_LETTER
Lo.
Definition: uchar.h:842
@ U_NON_SPACING_MARK
Mn.
Definition: uchar.h:844
@ U_START_PUNCTUATION
Ps.
Definition: uchar.h:872
@ U_CONNECTOR_PUNCTUATION
Pc.
Definition: uchar.h:876
@ U_END_PUNCTUATION
Pe.
Definition: uchar.h:874
@ U_OTHER_NUMBER
No.
Definition: uchar.h:854
@ U_LINE_SEPARATOR
Zl.
Definition: uchar.h:858
@ U_DASH_PUNCTUATION
Pd.
Definition: uchar.h:870
@ U_MATH_SYMBOL
Sm.
Definition: uchar.h:880
@ U_CHAR_CATEGORY_COUNT
One higher than the last enum UCharCategory constant.
Definition: uchar.h:898
@ U_DECIMAL_DIGIT_NUMBER
Nd.
Definition: uchar.h:850
@ U_LOWERCASE_LETTER
Ll.
Definition: uchar.h:836
@ U_FINAL_PUNCTUATION
Pf.
Definition: uchar.h:890
@ U_LETTER_NUMBER
Nl.
Definition: uchar.h:852
@ U_SURROGATE
Cs.
Definition: uchar.h:868
U_CAPI UBool u_isWhitespace(UChar32 c)
Determines if the specified code point is a whitespace character according to Java/ICU.
UVerticalOrientation
Vertical Orientation constants.
Definition: uchar.h:2713
@ U_VO_UPRIGHT
Definition: uchar.h:2727
@ U_VO_TRANSFORMED_UPRIGHT
Definition: uchar.h:2725
@ U_VO_ROTATED
Definition: uchar.h:2721
@ U_VO_TRANSFORMED_ROTATED
Definition: uchar.h:2723
U_CAPI UBool u_isdefined(UChar32 c)
Determines whether the specified code point is "defined", which usually means that it is assigned a c...
UJoiningGroup
Joining Group constants.
Definition: uchar.h:2141
@ U_JG_FE
Definition: uchar.h:2200
@ U_JG_BURUSHASKI_YEH_BARREE
Definition: uchar.h:2203
@ U_JG_MANICHAEAN_KAPH
Definition: uchar.h:2216
@ U_JG_THIN_YEH
Definition: uchar.h:2254
@ U_JG_MANICHAEAN_SAMEKH
Definition: uchar.h:2225
@ U_JG_MANICHAEAN_ZAYIN
Definition: uchar.h:2233
@ U_JG_MALAYALAM_TTA
Definition: uchar.h:2249
@ U_JG_MANICHAEAN_AYIN
Definition: uchar.h:2208
@ U_JG_MANICHAEAN_ALEPH
Definition: uchar.h:2207
@ U_JG_FARSI_YEH
Definition: uchar.h:2204
@ U_JG_MANICHAEAN_DALETH
Definition: uchar.h:2210
@ U_JG_MANICHAEAN_FIVE
Definition: uchar.h:2212
@ U_JG_MANICHAEAN_SADHE
Definition: uchar.h:2224
@ U_JG_MANICHAEAN_TAW
Definition: uchar.h:2226
@ U_JG_MALAYALAM_LLLA
Definition: uchar.h:2242
@ U_JG_MALAYALAM_NNNA
Definition: uchar.h:2245
@ U_JG_MANICHAEAN_ONE
Definition: uchar.h:2220
@ U_JG_MALAYALAM_LLA
Definition: uchar.h:2241
@ U_JG_VERTICAL_TAIL
Definition: uchar.h:2255
@ U_JG_AFRICAN_FEH
Definition: uchar.h:2235
@ U_JG_HANIFI_ROHINGYA_PA
Definition: uchar.h:2252
@ U_JG_MALAYALAM_BHA
Definition: uchar.h:2239
@ U_JG_MANICHAEAN_TWENTY
Definition: uchar.h:2230
@ U_JG_MANICHAEAN_THAMEDH
Definition: uchar.h:2229
@ U_JG_MANICHAEAN_PE
Definition: uchar.h:2221
@ U_JG_ZHAIN
Definition: uchar.h:2202
@ U_JG_MANICHAEAN_TETH
Definition: uchar.h:2228
@ U_JG_MANICHAEAN_BETH
Definition: uchar.h:2209
@ U_JG_COUNT
One more than the highest normal UJoiningGroup value.
Definition: uchar.h:2264
@ U_JG_MANICHAEAN_TEN
Definition: uchar.h:2227
@ U_JG_MANICHAEAN_GIMEL
Definition: uchar.h:2213
@ U_JG_TEH_MARBUTA_GOAL
Definition: uchar.h:2162
@ U_JG_AFRICAN_NOON
Definition: uchar.h:2236
@ U_JG_MALAYALAM_NGA
Definition: uchar.h:2243
@ U_JG_MANICHAEAN_MEM
Definition: uchar.h:2218
@ U_JG_KHAPH
Definition: uchar.h:2201
@ U_JG_MALAYALAM_RA
Definition: uchar.h:2247
@ U_JG_MALAYALAM_NNA
Definition: uchar.h:2244
@ U_JG_MALAYALAM_JA
Definition: uchar.h:2240
@ U_JG_MALAYALAM_NYA
Definition: uchar.h:2246
@ U_JG_MALAYALAM_SSA
Definition: uchar.h:2248
@ U_JG_MANICHAEAN_WAW
Definition: uchar.h:2231
@ U_JG_AFRICAN_QAF
Definition: uchar.h:2237
@ U_JG_MANICHAEAN_DHAMEDH
Definition: uchar.h:2211
@ U_JG_STRAIGHT_WAW
Definition: uchar.h:2234
@ U_JG_ROHINGYA_YEH
Definition: uchar.h:2206
@ U_JG_MANICHAEAN_HETH
Definition: uchar.h:2214
@ U_JG_MANICHAEAN_LAMEDH
Definition: uchar.h:2217
@ U_JG_MANICHAEAN_HUNDRED
Definition: uchar.h:2215
@ U_JG_HANIFI_ROHINGYA_KINNA_YA
Definition: uchar.h:2251
@ U_JG_MANICHAEAN_NUN
Definition: uchar.h:2219
@ U_JG_NYA
Definition: uchar.h:2205
@ U_JG_MANICHAEAN_YODH
Definition: uchar.h:2232
@ U_JG_MANICHAEAN_RESH
Definition: uchar.h:2223
@ U_JG_MANICHAEAN_QOPH
Definition: uchar.h:2222
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2549
@ U_HST_COUNT
One more than the highest normal UHangulSyllableType value.
Definition: uchar.h:2569
U_CAPI UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
U_CAPI UBool u_isIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in an identifier accordin...
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....
USentenceBreak
Sentence Break constants.
Definition: uchar.h:2389
@ U_SB_COUNT
One more than the highest normal USentenceBreak value.
Definition: uchar.h:2418
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...
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:3604
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:1979
@ U_EA_COUNT
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:1999
U_CAPI UChar32 u_totitle(UChar32 c)
The given character is mapped to its titlecase equivalent according to UnicodeData....
U_CAPI UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
U_CAPI UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
U_CAPI UCharDirection u_charDirection(UChar32 c)
Returns the bidirectional category value for the code point, which is used in the Unicode bidirection...
U_CAPI UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.
U_CAPI UBool u_iscntrl(UChar32 c)
Determines whether the specified code point is a control character (as defined by this function).
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...
UGraphemeClusterBreak
Grapheme Cluster Break constants.
Definition: uchar.h:2274
@ U_GCB_SPACING_MARK
Definition: uchar.h:2292
@ U_GCB_E_BASE
Definition: uchar.h:2298
@ U_GCB_REGIONAL_INDICATOR
Definition: uchar.h:2296
@ U_GCB_GLUE_AFTER_ZWJ
Definition: uchar.h:2304
@ U_GCB_ZWJ
Definition: uchar.h:2306
@ U_GCB_PREPEND
Definition: uchar.h:2294
@ U_GCB_COUNT
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2315
@ U_GCB_E_MODIFIER
Definition: uchar.h:2302
@ U_GCB_E_BASE_GAZ
Definition: uchar.h:2300
U_CAPI const USet * u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode)
Returns a frozen USet for a binary property.
U_CAPI uint8_t u_getCombiningClass(UChar32 c)
Returns the combining class of the code point as specified in UnicodeData.txt.
U_CAPI void u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context)
Enumerate efficiently all code points with their Unicode general categories.
U_CAPI UBool u_isxdigit(UChar32 c)
Determines whether the specified code point is a hexadecimal digit.
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,...
U_CAPI UChar32 u_charMirror(UChar32 c)
Maps the specified character to a "mirror-image" character.
U_CAPI UBool u_isJavaSpaceChar(UChar32 c)
Determine if the specified code point is a space character according to Java.
U_CAPI UBool u_isIDPart(UChar32 c)
Determines if the specified character is permissible as a non-initial character of an identifier acco...
U_CAPI UChar32 u_foldCase(UChar32 c, uint32_t options)
The given character is mapped to its case folding equivalent according to UnicodeData....
U_CAPI UBool u_isgraph(UChar32 c)
Determines whether the specified code point is a "graphic" character (printable, excluding spaces).
U_CAPI int32_t u_charDigitValue(UChar32 c)
Returns the decimal digit value of a decimal digit character.
U_CAPI UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier,...
U_CAPI int32_t u_getIDTypes(UChar32 c, UIdentifierType *types, int32_t capacity, UErrorCode *pErrorCode)
Writes code point c's Identifier_Type as a list of UIdentifierType values to the output types array a...
U_CAPI UBool u_ispunct(UChar32 c)
Determines whether the specified code point is a punctuation character.
UNumericType
Numeric Type constants.
Definition: uchar.h:2521
@ U_NT_COUNT
One more than the highest normal UNumericType value.
Definition: uchar.h:2539
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:2069
@ U_DT_COUNT
One more than the highest normal UDecompositionType value.
Definition: uchar.h:2101
U_CAPI UBool u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which)
Returns true if the property is true for the string.
U_CAPI int32_t u_digit(UChar32 ch, int8_t radix)
Returns the decimal digit value of the code point in the specified radix.
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:196
@ UCHAR_GRAPHEME_CLUSTER_BREAK
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:634
@ UCHAR_BIDI_PAIRED_BRACKET
String property Bidi_Paired_Bracket (new in Unicode 6.3).
Definition: uchar.h:770
@ UCHAR_ALPHABETIC
Binary property Alphabetic.
Definition: uchar.h:210
@ UCHAR_POSIX_GRAPH
Binary property graph (a C/POSIX character class).
Definition: uchar.h:408
@ UCHAR_RGI_EMOJI_TAG_SEQUENCE
Binary property of strings RGI_Emoji_Tag_Sequence.
Definition: uchar.h:521
@ UCHAR_RGI_EMOJI
Binary property of strings RGI_Emoji.
Definition: uchar.h:535
@ UCHAR_BLOCK
Enumerated property Block.
Definition: uchar.h:572
@ UCHAR_INVALID_CODE
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:812
@ UCHAR_SEGMENT_STARTER
Binary Property Segment_Starter.
Definition: uchar.h:383
@ UCHAR_PREPENDED_CONCATENATION_MARK
Binary property Prepended_Concatenation_Mark.
Definition: uchar.h:479
@ UCHAR_REGIONAL_INDICATOR
Binary property Regional_Indicator.
Definition: uchar.h:474
@ UCHAR_OTHER_PROPERTY_START
First constant for Unicode properties with unusual value types.
Definition: uchar.h:786
@ UCHAR_S_TERM
Binary property STerm (new in Unicode 4.0.1).
Definition: uchar.h:338
@ UCHAR_WHITE_SPACE
Binary property White_Space.
Definition: uchar.h:322
@ UCHAR_CANONICAL_COMBINING_CLASS
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:575
@ UCHAR_SOFT_DOTTED
Binary property Soft_Dotted (new in Unicode 3.2).
Definition: uchar.h:307
@ UCHAR_GRAPHEME_LINK
Binary property Grapheme_Link (new in Unicode 3.2).
Definition: uchar.h:254
@ UCHAR_PATTERN_SYNTAX
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition: uchar.h:388
@ UCHAR_GRAPHEME_EXTEND
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition: uchar.h:251
@ UCHAR_CASED
Binary property Cased.
Definition: uchar.h:420
@ UCHAR_XID_START
Binary property XID_Start.
Definition: uchar.h:329
@ UCHAR_RGI_EMOJI_ZWJ_SEQUENCE
Binary property of strings RGI_Emoji_ZWJ_Sequence.
Definition: uchar.h:528
@ UCHAR_VERTICAL_ORIENTATION
Enumerated property Vertical_Orientation.
Definition: uchar.h:671
@ UCHAR_NFKC_INERT
Binary property NFKC_Inert.
Definition: uchar.h:372
@ UCHAR_INT_LIMIT
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:686
@ UCHAR_RGI_EMOJI_FLAG_SEQUENCE
Binary property of strings RGI_Emoji_Flag_Sequence.
Definition: uchar.h:514
@ UCHAR_OTHER_PROPERTY_LIMIT
One more than the last constant for Unicode properties with unusual value types.
Definition: uchar.h:808
@ UCHAR_PATTERN_WHITE_SPACE
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:393
@ UCHAR_VARIATION_SELECTOR
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition: uchar.h:344
@ UCHAR_NUMERIC_VALUE
Double property Numeric_Value.
Definition: uchar.h:710
@ UCHAR_DOUBLE_START
First constant for double Unicode properties.
Definition: uchar.h:712
@ UCHAR_HEX_DIGIT
Binary property Hex_Digit.
Definition: uchar.h:257
@ UCHAR_SIMPLE_TITLECASE_MAPPING
String property Simple_Titlecase_Mapping.
Definition: uchar.h:751
@ UCHAR_TRAIL_CANONICAL_COMBINING_CLASS
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:629
@ UCHAR_DIACRITIC
Binary property Diacritic.
Definition: uchar.h:235
@ UCHAR_XID_CONTINUE
Binary property XID_Continue.
Definition: uchar.h:326
@ UCHAR_NFKC_QUICK_CHECK
Enumerated property NFKC_Quick_Check.
Definition: uchar.h:615
@ UCHAR_HYPHEN
Binary property Hyphen.
Definition: uchar.h:260
@ UCHAR_RADICAL
Binary property Radical (new in Unicode 3.2).
Definition: uchar.h:302
@ UCHAR_BIDI_PAIRED_BRACKET_TYPE
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:649
@ UCHAR_ID_COMPAT_MATH_CONTINUE
Binary property ID_Compat_Math_Continue.
Definition: uchar.h:555
@ UCHAR_DASH
Binary property Dash.
Definition: uchar.h:225
@ UCHAR_ASCII_HEX_DIGIT
Binary property ASCII_Hex_Digit.
Definition: uchar.h:214
@ UCHAR_ID_CONTINUE
Binary property ID_Continue.
Definition: uchar.h:265
@ UCHAR_NFKD_INERT
Binary property NFKD_Inert.
Definition: uchar.h:358
@ UCHAR_BIDI_MIRRORING_GLYPH
String property Bidi_Mirroring_Glyph.
Definition: uchar.h:728
@ UCHAR_LEAD_CANONICAL_COMBINING_CLASS
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:622
@ UCHAR_UNICODE_1_NAME
String property Unicode_1_Name.
Definition: uchar.h:763
@ UCHAR_BIDI_MIRRORED
Binary property Bidi_Mirrored.
Definition: uchar.h:223
@ UCHAR_CASE_IGNORABLE
Binary property Case_Ignorable.
Definition: uchar.h:422
@ UCHAR_EAST_ASIAN_WIDTH
Enumerated property East_Asian_Width.
Definition: uchar.h:582
@ UCHAR_STRING_LIMIT
One more than the last constant for string Unicode properties.
Definition: uchar.h:776
@ UCHAR_IDENTIFIER_TYPE
Miscellaneous property Identifier_Type.
Definition: uchar.h:801
@ UCHAR_EMOJI_PRESENTATION
Binary property Emoji_Presentation.
Definition: uchar.h:448
@ UCHAR_GRAPHEME_BASE
Binary property Grapheme_Base (new in Unicode 3.2).
Definition: uchar.h:247
@ UCHAR_NFKD_QUICK_CHECK
Enumerated property NFKD_Quick_Check.
Definition: uchar.h:609
@ UCHAR_NAME
String property Name.
Definition: uchar.h:742
@ UCHAR_UPPERCASE
Binary property Uppercase.
Definition: uchar.h:318
@ UCHAR_CASE_FOLDING
String property Case_Folding.
Definition: uchar.h:731
@ UCHAR_INDIC_SYLLABIC_CATEGORY
Enumerated property Indic_Syllabic_Category.
Definition: uchar.h:664
@ UCHAR_MATH
Binary property Math.
Definition: uchar.h:292
@ UCHAR_NUMERIC_TYPE
Enumerated property Numeric_Type.
Definition: uchar.h:597
@ UCHAR_RGI_EMOJI_MODIFIER_SEQUENCE
Binary property of strings RGI_Emoji_Modifier_Sequence.
Definition: uchar.h:507
@ UCHAR_ISO_COMMENT
Deprecated string property ISO_Comment.
Definition: uchar.h:735
@ UCHAR_SCRIPT
Enumerated property Script.
Definition: uchar.h:600
@ UCHAR_CHANGES_WHEN_TITLECASED
Binary property Changes_When_Titlecased.
Definition: uchar.h:428
@ UCHAR_EXTENDED_PICTOGRAPHIC
Binary property Extended_Pictographic.
Definition: uchar.h:486
@ UCHAR_DEPRECATED
Binary property Deprecated (new in Unicode 3.2).
Definition: uchar.h:232
@ UCHAR_CHANGES_WHEN_UPPERCASED
Binary property Changes_When_Uppercased.
Definition: uchar.h:426
@ UCHAR_CHANGES_WHEN_CASEFOLDED
Binary property Changes_When_Casefolded.
Definition: uchar.h:430
@ UCHAR_INDIC_POSITIONAL_CATEGORY
Enumerated property Indic_Positional_Category.
Definition: uchar.h:657
@ UCHAR_DEFAULT_IGNORABLE_CODE_POINT
Binary property Default_Ignorable_Code_Point (new in Unicode 3.2).
Definition: uchar.h:229
@ UCHAR_GENERAL_CATEGORY
Enumerated property General_Category.
Definition: uchar.h:585
@ UCHAR_INT_START
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:569
@ UCHAR_EMOJI_MODIFIER
Binary property Emoji_Modifier.
Definition: uchar.h:455
@ UCHAR_NFD_QUICK_CHECK
Enumerated property NFD_Quick_Check.
Definition: uchar.h:606
@ UCHAR_IDS_BINARY_OPERATOR
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition: uchar.h:276
@ UCHAR_BINARY_START
First constant for binary Unicode properties.
Definition: uchar.h:212
@ UCHAR_IDS_UNARY_OPERATOR
Binary property IDS_Unary_Operator.
Definition: uchar.h:543
@ UCHAR_TERMINAL_PUNCTUATION
Binary property Terminal_Punctuation.
Definition: uchar.h:311
@ UCHAR_GENERAL_CATEGORY_MASK
Bitmask property General_Category_Mask.
Definition: uchar.h:697
@ UCHAR_MASK_START
First constant for bit-mask Unicode properties.
Definition: uchar.h:699
@ UCHAR_DECOMPOSITION_TYPE
Enumerated property Decomposition_Type.
Definition: uchar.h:578
@ UCHAR_TITLECASE_MAPPING
String property Titlecase_Mapping.
Definition: uchar.h:757
@ UCHAR_HANGUL_SYLLABLE_TYPE
Enumerated property Hangul_Syllable_Type, new in Unicode 4.
Definition: uchar.h:603
@ UCHAR_LINE_BREAK
Enumerated property Line_Break.
Definition: uchar.h:594
@ UCHAR_SIMPLE_UPPERCASE_MAPPING
String property Simple_Uppercase_Mapping.
Definition: uchar.h:754
@ UCHAR_POSIX_ALNUM
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:398
@ UCHAR_JOINING_TYPE
Enumerated property Joining_Type.
Definition: uchar.h:591
@ UCHAR_EMOJI_KEYCAP_SEQUENCE
Binary property of strings Emoji_Keycap_Sequence.
Definition: uchar.h:500
@ UCHAR_QUOTATION_MARK
Binary property Quotation_Mark.
Definition: uchar.h:298
@ UCHAR_NFC_INERT
Binary property NFC_Inert.
Definition: uchar.h:365
@ UCHAR_LOWERCASE_MAPPING
String property Lowercase_Mapping.
Definition: uchar.h:739
@ UCHAR_SIMPLE_CASE_FOLDING
String property Simple_Case_Folding.
Definition: uchar.h:745
@ UCHAR_JOIN_CONTROL
Binary property Join_Control.
Definition: uchar.h:283
@ UCHAR_NONCHARACTER_CODE_POINT
Binary property Noncharacter_Code_Point.
Definition: uchar.h:296
@ UCHAR_BIDI_CONTROL
Binary property Bidi_Control.
Definition: uchar.h:218
@ UCHAR_CHANGES_WHEN_LOWERCASED
Binary property Changes_When_Lowercased.
Definition: uchar.h:424
@ UCHAR_BINARY_LIMIT
One more than the last constant for binary Unicode properties.
Definition: uchar.h:562
@ UCHAR_IDS_TRINARY_OPERATOR
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition: uchar.h:280
@ UCHAR_ID_START
Binary property ID_Start.
Definition: uchar.h:269
@ UCHAR_AGE
String property Age.
Definition: uchar.h:723
@ UCHAR_WORD_BREAK
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:644
@ UCHAR_DOUBLE_LIMIT
One more than the last constant for double Unicode properties.
Definition: uchar.h:718
@ UCHAR_EMOJI_MODIFIER_BASE
Binary property Emoji_Modifier_Base.
Definition: uchar.h:462
@ UCHAR_EMOJI_COMPONENT
Binary property Emoji_Component.
Definition: uchar.h:469
@ UCHAR_POSIX_BLANK
Binary property blank (a C/POSIX character class).
Definition: uchar.h:403
@ UCHAR_SIMPLE_LOWERCASE_MAPPING
String property Simple_Lowercase_Mapping.
Definition: uchar.h:748
@ UCHAR_NFD_INERT
Binary property NFD_Inert.
Definition: uchar.h:351
@ UCHAR_POSIX_PRINT
Binary property print (a C/POSIX character class).
Definition: uchar.h:413
@ UCHAR_SENTENCE_BREAK
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:639
@ UCHAR_IDEOGRAPHIC
Binary property Ideographic.
Definition: uchar.h:272
@ UCHAR_ID_COMPAT_MATH_START
Binary property ID_Compat_Math_Start.
Definition: uchar.h:549
@ UCHAR_UNIFIED_IDEOGRAPH
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:315
@ UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED
Binary property Changes_When_NFKC_Casefolded.
Definition: uchar.h:434
@ UCHAR_NFC_QUICK_CHECK
Enumerated property NFC_Quick_Check.
Definition: uchar.h:612
@ UCHAR_CASE_SENSITIVE
Binary property Case_Sensitive.
Definition: uchar.h:333
@ UCHAR_UPPERCASE_MAPPING
String property Uppercase_Mapping.
Definition: uchar.h:767
@ UCHAR_BIDI_CLASS
Enumerated property Bidi_Class.
Definition: uchar.h:567
@ UCHAR_BASIC_EMOJI
Binary property of strings Basic_Emoji.
Definition: uchar.h:493
@ UCHAR_MASK_LIMIT
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:705
@ UCHAR_IDENTIFIER_STATUS
Enumerated property Identifier_Status.
Definition: uchar.h:679
@ UCHAR_JOINING_GROUP
Enumerated property Joining_Group.
Definition: uchar.h:588
@ UCHAR_LOGICAL_ORDER_EXCEPTION
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition: uchar.h:287
@ UCHAR_EXTENDER
Binary property Extender.
Definition: uchar.h:239
@ UCHAR_STRING_START
First constant for string Unicode properties.
Definition: uchar.h:725
@ UCHAR_SCRIPT_EXTENSIONS
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:784
@ UCHAR_FULL_COMPOSITION_EXCLUSION
Binary property Full_Composition_Exclusion.
Definition: uchar.h:243
@ UCHAR_EMOJI
Binary property Emoji.
Definition: uchar.h:441
@ UCHAR_LOWERCASE
Binary property Lowercase.
Definition: uchar.h:290
@ UCHAR_CHANGES_WHEN_CASEMAPPED
Binary property Changes_When_Casemapped.
Definition: uchar.h:432
@ UCHAR_POSIX_XDIGIT
Binary property xdigit (a C/POSIX character class).
Definition: uchar.h:418
U_CAPI int32_t u_getIntPropertyMinValue(UProperty which)
Get the minimum value for an enumerated/integer/binary Unicode property.
U_CAPI UBool u_isalnum(UChar32 c)
Determines whether the specified code point is an alphanumeric character (letter or digit) according ...
U_CAPI UBool u_isalpha(UChar32 c)
Determines whether the specified code point is a letter character.
U_CAPI UChar32 u_tolower(UChar32 c)
The given character is mapped to its lowercase equivalent according to UnicodeData....
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.
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition: uchar.h:2626
@ U_INSC_VISARGA
Definition: uchar.h:2698
@ U_INSC_BINDU
Definition: uchar.h:2638
@ U_INSC_INVISIBLE_STACKER
Definition: uchar.h:2672
@ U_INSC_SYLLABLE_MODIFIER
Definition: uchar.h:2690
@ U_INSC_VIRAMA
Definition: uchar.h:2696
@ U_INSC_MODIFYING_LETTER
Definition: uchar.h:2676
@ U_INSC_REGISTER_SHIFTER
Definition: uchar.h:2688
@ U_INSC_NUKTA
Definition: uchar.h:2680
@ U_INSC_NON_JOINER
Definition: uchar.h:2678
@ U_INSC_TONE_MARK
Definition: uchar.h:2694
@ U_INSC_GEMINATION_MARK
Definition: uchar.h:2670
@ U_INSC_CONSONANT_KILLER
Definition: uchar.h:2654
@ U_INSC_CONSONANT_HEAD_LETTER
Definition: uchar.h:2650
@ U_INSC_BRAHMI_JOINING_NUMBER
Definition: uchar.h:2640
@ U_INSC_CONSONANT_DEAD
Definition: uchar.h:2646
@ U_INSC_OTHER
Definition: uchar.h:2634
@ U_INSC_NUMBER
Definition: uchar.h:2682
@ U_INSC_TONE_LETTER
Definition: uchar.h:2692
@ U_INSC_JOINER
Definition: uchar.h:2674
@ U_INSC_CANTILLATION_MARK
Definition: uchar.h:2642
@ U_INSC_PURE_KILLER
Definition: uchar.h:2686
@ U_INSC_VOWEL_INDEPENDENT
Definition: uchar.h:2704
@ U_INSC_CONSONANT_MEDIAL
Definition: uchar.h:2656
@ U_INSC_CONSONANT_WITH_STACKER
Definition: uchar.h:2668
@ U_INSC_CONSONANT_PREFIXED
Definition: uchar.h:2662
@ U_INSC_CONSONANT
Definition: uchar.h:2644
@ U_INSC_CONSONANT_SUCCEEDING_REPHA
Definition: uchar.h:2666
@ U_INSC_CONSONANT_PLACEHOLDER
Definition: uchar.h:2658
@ U_INSC_VOWEL
Definition: uchar.h:2700
@ U_INSC_NUMBER_JOINER
Definition: uchar.h:2684
@ U_INSC_CONSONANT_PRECEDING_REPHA
Definition: uchar.h:2660
@ U_INSC_CONSONANT_SUBJOINED
Definition: uchar.h:2664
@ U_INSC_VOWEL_DEPENDENT
Definition: uchar.h:2702
@ U_INSC_CONSONANT_INITIAL_POSTFIXED
Definition: uchar.h:2652
@ U_INSC_AVAGRAHA
Definition: uchar.h:2636
@ U_INSC_CONSONANT_FINAL
Definition: uchar.h:2648
UWordBreakValues
Word Break constants.
Definition: uchar.h:2326
@ U_WB_E_MODIFIER
Definition: uchar.h:2364
@ U_WB_E_BASE_GAZ
Definition: uchar.h:2362
@ U_WB_REGIONAL_INDICATOR
Definition: uchar.h:2352
@ U_WB_DOUBLE_QUOTE
Definition: uchar.h:2358
@ U_WB_LF
Definition: uchar.h:2346
@ U_WB_HEBREW_LETTER
Definition: uchar.h:2354
@ U_WB_WSEGSPACE
Definition: uchar.h:2370
@ U_WB_COUNT
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2379
@ U_WB_GLUE_AFTER_ZWJ
Definition: uchar.h:2366
@ U_WB_NEWLINE
Definition: uchar.h:2350
@ U_WB_SINGLE_QUOTE
Definition: uchar.h:2356
@ U_WB_CR
Definition: uchar.h:2342
@ U_WB_E_BASE
Definition: uchar.h:2360
@ U_WB_ZWJ
Definition: uchar.h:2368
@ U_WB_EXTEND
Definition: uchar.h:2344
@ U_WB_MIDNUMLET
Definition: uchar.h:2348
U_CAPI UBool u_isJavaIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in a Java identifier.
UBidiPairedBracketType
Bidi Paired Bracket Type constants.
Definition: uchar.h:1088
@ U_BPT_CLOSE
Close paired bracket.
Definition: uchar.h:1100
@ U_BPT_COUNT
One more than the highest normal UBidiPairedBracketType value.
Definition: uchar.h:1108
@ U_BPT_NONE
Not a paired bracket.
Definition: uchar.h:1096
@ U_BPT_OPEN
Open paired bracket.
Definition: uchar.h:1098
C API: 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:30
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:427
#define U_DEPRECATED
This is used to declare a function as a deprecated public ICU C API
Definition: umachine.h:116
#define U_CDECL_END
This is used to end a declaration of a library private ICU C API.
Definition: umachine.h:86
int8_t UBool
The ICU boolean type, a signed-byte integer.
Definition: umachine.h:247
#define U_CAPI
This is used to declare a function as a public ICU C API.
Definition: umachine.h:110
char16_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:85
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition: uset.h:50
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