ICU 77.1  77.1
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
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 "16.0"
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 
555 #ifndef U_HIDE_DRAFT_API
556 
562 #endif // U_HIDE_DRAFT_API
563 #ifndef U_HIDE_DEPRECATED_API
564 
569 #endif // U_HIDE_DEPRECATED_API
570 
578  UCHAR_BLOCK=0x1001,
606  UCHAR_SCRIPT=0x100A,
685 #ifndef U_HIDE_DRAFT_API
686 
692 #endif // U_HIDE_DRAFT_API
693 #ifndef U_HIDE_DEPRECATED_API
694 
699 #endif // U_HIDE_DEPRECATED_API
700 
712 #ifndef U_HIDE_DEPRECATED_API
713 
718 #endif // U_HIDE_DEPRECATED_API
719 
725 #ifndef U_HIDE_DEPRECATED_API
726 
731 #endif // U_HIDE_DEPRECATED_API
732 
735  UCHAR_AGE=0x4000,
744 #ifndef U_HIDE_DEPRECATED_API
745 
748 #endif /* U_HIDE_DEPRECATED_API */
749 
754  UCHAR_NAME=0x4005,
770 #ifndef U_HIDE_DEPRECATED_API
771 
776 #endif /* U_HIDE_DEPRECATED_API */
777 
783 #ifndef U_HIDE_DEPRECATED_API
784 
789 #endif // U_HIDE_DEPRECATED_API
790 
813 #ifndef U_HIDE_DEPRECATED_API
814 
819 #endif // U_HIDE_DEPRECATED_API
820 
823 } UProperty;
824 
830 typedef enum UCharCategory
831 {
832  /*
833  * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
834  * It matches pairs of lines like
835  * / ** <Unicode 2-letter General_Category value> comment... * /
836  * U_<[A-Z_]+> = <integer>,
837  */
838 
909 } UCharCategory;
910 
925 #define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES)
926 
928 #define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER)
929 
930 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
931 
932 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
933 
934 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
935 
936 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
937 
939 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
940 
941 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
942 
943 #define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK)
944 
946 #define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER)
947 
948 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
949 
950 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
951 
953 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
954 
955 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
956 
957 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
958 
960 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
961 
962 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
963 
964 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
965 
966 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
967 
969 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
970 
971 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
972 
973 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
974 
975 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
976 
977 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
978 
980 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
981 
982 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
983 
984 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
985 
986 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
987 
989 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
990 
991 #define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION)
992 
993 
995 #define U_GC_L_MASK \
996  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
997 
999 #define U_GC_LC_MASK \
1000  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
1001 
1003 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
1004 
1006 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
1007 
1009 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
1010 
1012 #define U_GC_C_MASK \
1013  (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
1014 
1016 #define U_GC_P_MASK \
1017  (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
1018  U_GC_PI_MASK|U_GC_PF_MASK)
1019 
1021 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
1022 
1027 typedef enum UCharDirection {
1028  /*
1029  * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
1030  * It matches pairs of lines like
1031  * / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
1032  * U_<[A-Z_]+> = <integer>,
1033  */
1034 
1081 #ifndef U_HIDE_DEPRECATED_API
1082 
1089 #endif // U_HIDE_DEPRECATED_API
1090 } UCharDirection;
1091 
1099  /*
1100  * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
1101  * It matches lines like
1102  * U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
1103  */
1104 
1111 #ifndef U_HIDE_DEPRECATED_API
1112 
1118  U_BPT_COUNT /* 3 */
1119 #endif // U_HIDE_DEPRECATED_API
1121 
1127  /*
1128  * Note: UBlockCode constants are parsed by preparseucd.py.
1129  * It matches lines like
1130  * UBLOCK_<Unicode Block value name> = <integer>,
1131  */
1132 
1134  UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
1135 
1137  UBLOCK_BASIC_LATIN = 1, /*[0000]*/
1138 
1141 
1144 
1147 
1149  UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
1150 
1153 
1156 
1161  UBLOCK_GREEK =8, /*[0370]*/
1162 
1164  UBLOCK_CYRILLIC =9, /*[0400]*/
1165 
1167  UBLOCK_ARMENIAN =10, /*[0530]*/
1168 
1170  UBLOCK_HEBREW =11, /*[0590]*/
1171 
1173  UBLOCK_ARABIC =12, /*[0600]*/
1174 
1176  UBLOCK_SYRIAC =13, /*[0700]*/
1177 
1179  UBLOCK_THAANA =14, /*[0780]*/
1180 
1182  UBLOCK_DEVANAGARI =15, /*[0900]*/
1183 
1185  UBLOCK_BENGALI =16, /*[0980]*/
1186 
1188  UBLOCK_GURMUKHI =17, /*[0A00]*/
1189 
1191  UBLOCK_GUJARATI =18, /*[0A80]*/
1192 
1194  UBLOCK_ORIYA =19, /*[0B00]*/
1195 
1197  UBLOCK_TAMIL =20, /*[0B80]*/
1198 
1200  UBLOCK_TELUGU =21, /*[0C00]*/
1201 
1203  UBLOCK_KANNADA =22, /*[0C80]*/
1204 
1206  UBLOCK_MALAYALAM =23, /*[0D00]*/
1207 
1209  UBLOCK_SINHALA =24, /*[0D80]*/
1210 
1212  UBLOCK_THAI =25, /*[0E00]*/
1213 
1215  UBLOCK_LAO =26, /*[0E80]*/
1216 
1218  UBLOCK_TIBETAN =27, /*[0F00]*/
1219 
1221  UBLOCK_MYANMAR =28, /*[1000]*/
1222 
1224  UBLOCK_GEORGIAN =29, /*[10A0]*/
1225 
1227  UBLOCK_HANGUL_JAMO =30, /*[1100]*/
1228 
1230  UBLOCK_ETHIOPIC =31, /*[1200]*/
1231 
1233  UBLOCK_CHEROKEE =32, /*[13A0]*/
1234 
1237 
1239  UBLOCK_OGHAM =34, /*[1680]*/
1240 
1242  UBLOCK_RUNIC =35, /*[16A0]*/
1243 
1245  UBLOCK_KHMER =36, /*[1780]*/
1246 
1248  UBLOCK_MONGOLIAN =37, /*[1800]*/
1249 
1252 
1254  UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
1255 
1258 
1261 
1263  UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
1264 
1270 
1273 
1275  UBLOCK_NUMBER_FORMS =45, /*[2150]*/
1276 
1278  UBLOCK_ARROWS =46, /*[2190]*/
1279 
1282 
1285 
1287  UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
1288 
1291 
1294 
1296  UBLOCK_BOX_DRAWING =52, /*[2500]*/
1297 
1299  UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
1300 
1302  UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
1303 
1306 
1308  UBLOCK_DINGBATS =56, /*[2700]*/
1309 
1311  UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
1312 
1315 
1317  UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
1318 
1321 
1324 
1326  UBLOCK_HIRAGANA =62, /*[3040]*/
1327 
1329  UBLOCK_KATAKANA =63, /*[30A0]*/
1330 
1332  UBLOCK_BOPOMOFO =64, /*[3100]*/
1333 
1336 
1338  UBLOCK_KANBUN =66, /*[3190]*/
1339 
1342 
1345 
1348 
1351 
1354 
1356  UBLOCK_YI_SYLLABLES =72, /*[A000]*/
1357 
1359  UBLOCK_YI_RADICALS =73, /*[A490]*/
1360 
1362  UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
1363 
1365  UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
1366 
1369 
1371  UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
1372 
1382  UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
1393 
1396 
1399 
1402 
1405 
1408 
1411 
1414 
1416  UBLOCK_SPECIALS =86, /*[FFF0]*/
1417 
1420 
1421  /* New blocks in Unicode 3.1 */
1422 
1424  UBLOCK_OLD_ITALIC = 88, /*[10300]*/
1426  UBLOCK_GOTHIC = 89, /*[10330]*/
1428  UBLOCK_DESERET = 90, /*[10400]*/
1432  UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
1440  UBLOCK_TAGS = 96, /*[E0000]*/
1441 
1442  /* New blocks in Unicode 3.2 */
1443 
1452  UBLOCK_TAGALOG = 98, /*[1700]*/
1454  UBLOCK_HANUNOO = 99, /*[1720]*/
1456  UBLOCK_BUHID = 100, /*[1740]*/
1458  UBLOCK_TAGBANWA = 101, /*[1760]*/
1472  UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
1477 
1478  /* New blocks in Unicode 4 */
1479 
1481  UBLOCK_LIMBU = 111, /*[1900]*/
1483  UBLOCK_TAI_LE = 112, /*[1950]*/
1485  UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
1487  UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
1493  UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
1495  UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
1497  UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
1499  UBLOCK_UGARITIC = 120, /*[10380]*/
1501  UBLOCK_SHAVIAN = 121, /*[10450]*/
1503  UBLOCK_OSMANYA = 122, /*[10480]*/
1505  UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
1507  UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
1510 
1511  /* New blocks in Unicode 4.1 */
1512 
1516  UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
1518  UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
1520  UBLOCK_BUGINESE = 129, /*[1A00]*/
1522  UBLOCK_CJK_STROKES = 130, /*[31C0]*/
1526  UBLOCK_COPTIC = 132, /*[2C80]*/
1528  UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
1530  UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
1532  UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
1534  UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
1536  UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
1540  UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
1542  UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
1548  UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
1550  UBLOCK_TIFINAGH = 144, /*[2D30]*/
1552  UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
1553 
1554  /* New blocks in Unicode 5.0 */
1555 
1557  UBLOCK_NKO = 146, /*[07C0]*/
1559  UBLOCK_BALINESE = 147, /*[1B00]*/
1561  UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
1563  UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
1565  UBLOCK_PHAGS_PA = 150, /*[A840]*/
1567  UBLOCK_PHOENICIAN = 151, /*[10900]*/
1569  UBLOCK_CUNEIFORM = 152, /*[12000]*/
1573  UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
1574 
1575  /* New blocks in Unicode 5.1 */
1576 
1578  UBLOCK_SUNDANESE = 155, /*[1B80]*/
1580  UBLOCK_LEPCHA = 156, /*[1C00]*/
1582  UBLOCK_OL_CHIKI = 157, /*[1C50]*/
1584  UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
1586  UBLOCK_VAI = 159, /*[A500]*/
1588  UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
1590  UBLOCK_SAURASHTRA = 161, /*[A880]*/
1592  UBLOCK_KAYAH_LI = 162, /*[A900]*/
1594  UBLOCK_REJANG = 163, /*[A930]*/
1596  UBLOCK_CHAM = 164, /*[AA00]*/
1598  UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
1600  UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
1602  UBLOCK_LYCIAN = 167, /*[10280]*/
1604  UBLOCK_CARIAN = 168, /*[102A0]*/
1606  UBLOCK_LYDIAN = 169, /*[10920]*/
1608  UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
1610  UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
1611 
1612  /* New blocks in Unicode 5.2 */
1613 
1615  UBLOCK_SAMARITAN = 172, /*[0800]*/
1619  UBLOCK_TAI_THAM = 174, /*[1A20]*/
1621  UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
1623  UBLOCK_LISU = 176, /*[A4D0]*/
1625  UBLOCK_BAMUM = 177, /*[A6A0]*/
1629  UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
1633  UBLOCK_JAVANESE = 181, /*[A980]*/
1635  UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
1637  UBLOCK_TAI_VIET = 183, /*[AA80]*/
1639  UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
1643  UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
1645  UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
1647  UBLOCK_AVESTAN = 188, /*[10B00]*/
1651  UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
1653  UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
1655  UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
1657  UBLOCK_KAITHI = 193, /*[11080]*/
1659  UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
1666 
1667  /* New blocks in Unicode 6.0 */
1668 
1670  UBLOCK_MANDAIC = 198, /*[0840]*/
1672  UBLOCK_BATAK = 199, /*[1BC0]*/
1674  UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
1676  UBLOCK_BRAHMI = 201, /*[11000]*/
1678  UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
1680  UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
1682  UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
1686  UBLOCK_EMOTICONS = 206, /*[1F600]*/
1690  UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
1693 
1694  /* New blocks in Unicode 6.1 */
1695 
1697  UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
1701  UBLOCK_CHAKMA = 212, /*[11100]*/
1705  UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
1707  UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
1709  UBLOCK_MIAO = 216, /*[16F00]*/
1711  UBLOCK_SHARADA = 217, /*[11180]*/
1713  UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
1717  UBLOCK_TAKRI = 220, /*[11680]*/
1718 
1719  /* New blocks in Unicode 7.0 */
1720 
1722  UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
1724  UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
1726  UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
1730  UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
1732  UBLOCK_ELBASAN = 226, /*[10500]*/
1736  UBLOCK_GRANTHA = 228, /*[11300]*/
1738  UBLOCK_KHOJKI = 229, /*[11200]*/
1740  UBLOCK_KHUDAWADI = 230, /*[112B0]*/
1742  UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
1744  UBLOCK_LINEAR_A = 232, /*[10600]*/
1746  UBLOCK_MAHAJANI = 233, /*[11150]*/
1748  UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
1750  UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
1752  UBLOCK_MODI = 236, /*[11600]*/
1754  UBLOCK_MRO = 237, /*[16A40]*/
1756  UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
1758  UBLOCK_NABATAEAN = 239, /*[10880]*/
1760  UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
1762  UBLOCK_OLD_PERMIC = 241, /*[10350]*/
1764  UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
1766  UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
1768  UBLOCK_PALMYRENE = 244, /*[10860]*/
1770  UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
1772  UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
1776  UBLOCK_SIDDHAM = 248, /*[11580]*/
1780  UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
1782  UBLOCK_TIRHUTA = 251, /*[11480]*/
1784  UBLOCK_WARANG_CITI = 252, /*[118A0]*/
1785 
1786  /* New blocks in Unicode 8.0 */
1787 
1789  UBLOCK_AHOM = 253, /*[11700]*/
1791  UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
1793  UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
1799  UBLOCK_HATRAN = 258, /*[108E0]*/
1801  UBLOCK_MULTANI = 259, /*[11280]*/
1803  UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
1807  UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/
1808 
1809  /* New blocks in Unicode 9.0 */
1810 
1812  UBLOCK_ADLAM = 263, /*[1E900]*/
1814  UBLOCK_BHAIKSUKI = 264, /*[11C00]*/
1816  UBLOCK_CYRILLIC_EXTENDED_C = 265, /*[1C80]*/
1818  UBLOCK_GLAGOLITIC_SUPPLEMENT = 266, /*[1E000]*/
1822  UBLOCK_MARCHEN = 268, /*[11C70]*/
1824  UBLOCK_MONGOLIAN_SUPPLEMENT = 269, /*[11660]*/
1826  UBLOCK_NEWA = 270, /*[11400]*/
1828  UBLOCK_OSAGE = 271, /*[104B0]*/
1830  UBLOCK_TANGUT = 272, /*[17000]*/
1832  UBLOCK_TANGUT_COMPONENTS = 273, /*[18800]*/
1833 
1834  // New blocks in Unicode 10.0
1835 
1839  UBLOCK_KANA_EXTENDED_A = 275, /*[1B100]*/
1841  UBLOCK_MASARAM_GONDI = 276, /*[11D00]*/
1843  UBLOCK_NUSHU = 277, /*[1B170]*/
1845  UBLOCK_SOYOMBO = 278, /*[11A50]*/
1847  UBLOCK_SYRIAC_SUPPLEMENT = 279, /*[0860]*/
1849  UBLOCK_ZANABAZAR_SQUARE = 280, /*[11A00]*/
1850 
1851  // New blocks in Unicode 11.0
1852 
1854  UBLOCK_CHESS_SYMBOLS = 281, /*[1FA00]*/
1856  UBLOCK_DOGRA = 282, /*[11800]*/
1858  UBLOCK_GEORGIAN_EXTENDED = 283, /*[1C90]*/
1860  UBLOCK_GUNJALA_GONDI = 284, /*[11D60]*/
1862  UBLOCK_HANIFI_ROHINGYA = 285, /*[10D00]*/
1864  UBLOCK_INDIC_SIYAQ_NUMBERS = 286, /*[1EC70]*/
1866  UBLOCK_MAKASAR = 287, /*[11EE0]*/
1868  UBLOCK_MAYAN_NUMERALS = 288, /*[1D2E0]*/
1870  UBLOCK_MEDEFAIDRIN = 289, /*[16E40]*/
1872  UBLOCK_OLD_SOGDIAN = 290, /*[10F00]*/
1874  UBLOCK_SOGDIAN = 291, /*[10F30]*/
1875 
1876  // New blocks in Unicode 12.0
1877 
1881  UBLOCK_ELYMAIC = 293, /*[10FE0]*/
1883  UBLOCK_NANDINAGARI = 294, /*[119A0]*/
1887  UBLOCK_OTTOMAN_SIYAQ_NUMBERS = 296, /*[1ED00]*/
1889  UBLOCK_SMALL_KANA_EXTENSION = 297, /*[1B130]*/
1893  UBLOCK_TAMIL_SUPPLEMENT = 299, /*[11FC0]*/
1895  UBLOCK_WANCHO = 300, /*[1E2C0]*/
1896 
1897  // New blocks in Unicode 13.0
1898 
1900  UBLOCK_CHORASMIAN = 301, /*[10FB0]*/
1904  UBLOCK_DIVES_AKURU = 303, /*[11900]*/
1906  UBLOCK_KHITAN_SMALL_SCRIPT = 304, /*[18B00]*/
1908  UBLOCK_LISU_SUPPLEMENT = 305, /*[11FB0]*/
1912  UBLOCK_TANGUT_SUPPLEMENT = 307, /*[18D00]*/
1914  UBLOCK_YEZIDI = 308, /*[10E80]*/
1915 
1916  // New blocks in Unicode 14.0
1917 
1919  UBLOCK_ARABIC_EXTENDED_B = 309, /*[0870]*/
1921  UBLOCK_CYPRO_MINOAN = 310, /*[12F90]*/
1923  UBLOCK_ETHIOPIC_EXTENDED_B = 311, /*[1E7E0]*/
1925  UBLOCK_KANA_EXTENDED_B = 312, /*[1AFF0]*/
1927  UBLOCK_LATIN_EXTENDED_F = 313, /*[10780]*/
1929  UBLOCK_LATIN_EXTENDED_G = 314, /*[1DF00]*/
1931  UBLOCK_OLD_UYGHUR = 315, /*[10F70]*/
1933  UBLOCK_TANGSA = 316, /*[16A70]*/
1935  UBLOCK_TOTO = 317, /*[1E290]*/
1939  UBLOCK_VITHKUQI = 319, /*[10570]*/
1942 
1943  // New blocks in Unicode 15.0
1944 
1946  UBLOCK_ARABIC_EXTENDED_C = 321, /*[10EC0]*/
1950  UBLOCK_CYRILLIC_EXTENDED_D = 323, /*[1E030]*/
1952  UBLOCK_DEVANAGARI_EXTENDED_A = 324, /*[11B00]*/
1954  UBLOCK_KAKTOVIK_NUMERALS = 325, /*[1D2C0]*/
1956  UBLOCK_KAWI = 326, /*[11F00]*/
1958  UBLOCK_NAG_MUNDARI = 327, /*[1E4D0]*/
1959 
1960  // New block in Unicode 15.1
1961 
1964 
1965  // New blocks in Unicode 16.0
1966 
1970  UBLOCK_GARAY = 330, /*[10D40]*/
1972  UBLOCK_GURUNG_KHEMA = 331, /*[16100]*/
1974  UBLOCK_KIRAT_RAI = 332, /*[16D40]*/
1976  UBLOCK_MYANMAR_EXTENDED_C = 333, /*[116D0]*/
1978  UBLOCK_OL_ONAL = 334, /*[1E5D0]*/
1980  UBLOCK_SUNUWAR = 335, /*[11BC0]*/
1984  UBLOCK_TODHRI = 337, /*[105C0]*/
1986  UBLOCK_TULU_TIGALARI = 338, /*[11380]*/
1987 
1988 #ifndef U_HIDE_DEPRECATED_API
1989 
1996 #endif // U_HIDE_DEPRECATED_API
1997 
2000 };
2001 
2003 typedef enum UBlockCode UBlockCode;
2004 
2012 typedef enum UEastAsianWidth {
2013  /*
2014  * Note: UEastAsianWidth constants are parsed by preparseucd.py.
2015  * It matches lines like
2016  * U_EA_<Unicode East_Asian_Width value name>
2017  */
2018 
2019  U_EA_NEUTRAL, /*[N]*/
2020  U_EA_AMBIGUOUS, /*[A]*/
2021  U_EA_HALFWIDTH, /*[H]*/
2022  U_EA_FULLWIDTH, /*[F]*/
2023  U_EA_NARROW, /*[Na]*/
2024  U_EA_WIDE, /*[W]*/
2025 #ifndef U_HIDE_DEPRECATED_API
2026 
2033 #endif // U_HIDE_DEPRECATED_API
2034 } UEastAsianWidth;
2035 
2047 typedef enum UCharNameChoice {
2050 #ifndef U_HIDE_DEPRECATED_API
2051 
2057 #endif /* U_HIDE_DEPRECATED_API */
2058 
2062 #ifndef U_HIDE_DEPRECATED_API
2063 
2068 #endif // U_HIDE_DEPRECATED_API
2069 } UCharNameChoice;
2070 
2084 typedef enum UPropertyNameChoice {
2085  U_SHORT_PROPERTY_NAME,
2086  U_LONG_PROPERTY_NAME,
2087 #ifndef U_HIDE_DEPRECATED_API
2088 
2093 #endif // U_HIDE_DEPRECATED_API
2095 
2102 typedef enum UDecompositionType {
2103  /*
2104  * Note: UDecompositionType constants are parsed by preparseucd.py.
2105  * It matches lines like
2106  * U_DT_<Unicode Decomposition_Type value name>
2107  */
2108 
2109  U_DT_NONE, /*[none]*/
2110  U_DT_CANONICAL, /*[can]*/
2111  U_DT_COMPAT, /*[com]*/
2112  U_DT_CIRCLE, /*[enc]*/
2113  U_DT_FINAL, /*[fin]*/
2114  U_DT_FONT, /*[font]*/
2115  U_DT_FRACTION, /*[fra]*/
2116  U_DT_INITIAL, /*[init]*/
2117  U_DT_ISOLATED, /*[iso]*/
2118  U_DT_MEDIAL, /*[med]*/
2119  U_DT_NARROW, /*[nar]*/
2120  U_DT_NOBREAK, /*[nb]*/
2121  U_DT_SMALL, /*[sml]*/
2122  U_DT_SQUARE, /*[sqr]*/
2123  U_DT_SUB, /*[sub]*/
2124  U_DT_SUPER, /*[sup]*/
2125  U_DT_VERTICAL, /*[vert]*/
2126  U_DT_WIDE, /*[wide]*/
2127 #ifndef U_HIDE_DEPRECATED_API
2128 
2134  U_DT_COUNT /* 18 */
2135 #endif // U_HIDE_DEPRECATED_API
2137 
2144 typedef enum UJoiningType {
2145  /*
2146  * Note: UJoiningType constants are parsed by preparseucd.py.
2147  * It matches lines like
2148  * U_JT_<Unicode Joining_Type value name>
2149  */
2150 
2151  U_JT_NON_JOINING, /*[U]*/
2152  U_JT_JOIN_CAUSING, /*[C]*/
2153  U_JT_DUAL_JOINING, /*[D]*/
2154  U_JT_LEFT_JOINING, /*[L]*/
2155  U_JT_RIGHT_JOINING, /*[R]*/
2156  U_JT_TRANSPARENT, /*[T]*/
2157 #ifndef U_HIDE_DEPRECATED_API
2158 
2164  U_JT_COUNT /* 6 */
2165 #endif // U_HIDE_DEPRECATED_API
2166 } UJoiningType;
2167 
2174 typedef enum UJoiningGroup {
2175  /*
2176  * Note: UJoiningGroup constants are parsed by preparseucd.py.
2177  * It matches lines like
2178  * U_JG_<Unicode Joining_Group value name>
2179  */
2180 
2181  U_JG_NO_JOINING_GROUP,
2182  U_JG_AIN,
2183  U_JG_ALAPH,
2184  U_JG_ALEF,
2185  U_JG_BEH,
2186  U_JG_BETH,
2187  U_JG_DAL,
2188  U_JG_DALATH_RISH,
2189  U_JG_E,
2190  U_JG_FEH,
2191  U_JG_FINAL_SEMKATH,
2192  U_JG_GAF,
2193  U_JG_GAMAL,
2194  U_JG_HAH,
2196  U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
2197  U_JG_HE,
2198  U_JG_HEH,
2199  U_JG_HEH_GOAL,
2200  U_JG_HETH,
2201  U_JG_KAF,
2202  U_JG_KAPH,
2203  U_JG_KNOTTED_HEH,
2204  U_JG_LAM,
2205  U_JG_LAMADH,
2206  U_JG_MEEM,
2207  U_JG_MIM,
2208  U_JG_NOON,
2209  U_JG_NUN,
2210  U_JG_PE,
2211  U_JG_QAF,
2212  U_JG_QAPH,
2213  U_JG_REH,
2214  U_JG_REVERSED_PE,
2215  U_JG_SAD,
2216  U_JG_SADHE,
2217  U_JG_SEEN,
2218  U_JG_SEMKATH,
2219  U_JG_SHIN,
2220  U_JG_SWASH_KAF,
2221  U_JG_SYRIAC_WAW,
2222  U_JG_TAH,
2223  U_JG_TAW,
2224  U_JG_TEH_MARBUTA,
2225  U_JG_TETH,
2226  U_JG_WAW,
2227  U_JG_YEH,
2228  U_JG_YEH_BARREE,
2229  U_JG_YEH_WITH_TAIL,
2230  U_JG_YUDH,
2231  U_JG_YUDH_HE,
2232  U_JG_ZAIN,
2292 #ifndef U_HIDE_DEPRECATED_API
2293 
2300 #endif // U_HIDE_DEPRECATED_API
2301 } UJoiningGroup;
2302 
2310  /*
2311  * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2312  * It matches lines like
2313  * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2314  */
2315 
2316  U_GCB_OTHER = 0, /*[XX]*/
2317  U_GCB_CONTROL = 1, /*[CN]*/
2318  U_GCB_CR = 2, /*[CR]*/
2319  U_GCB_EXTEND = 3, /*[EX]*/
2320  U_GCB_L = 4, /*[L]*/
2321  U_GCB_LF = 5, /*[LF]*/
2322  U_GCB_LV = 6, /*[LV]*/
2323  U_GCB_LVT = 7, /*[LVT]*/
2324  U_GCB_T = 8, /*[T]*/
2325  U_GCB_V = 9, /*[V]*/
2327  U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2329  U_GCB_PREPEND = 11, /*[PP]*/
2331  U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2333  U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2335  U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2337  U_GCB_E_MODIFIER = 15, /*[EM]*/
2339  U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2341  U_GCB_ZWJ = 17, /*[ZWJ]*/
2342 
2343 #ifndef U_HIDE_DEPRECATED_API
2344 
2351 #endif // U_HIDE_DEPRECATED_API
2353 
2361 typedef enum UWordBreakValues {
2362  /*
2363  * Note: UWordBreakValues constants are parsed by preparseucd.py.
2364  * It matches lines like
2365  * U_WB_<Unicode Word_Break value name>
2366  */
2367 
2368  U_WB_OTHER = 0, /*[XX]*/
2369  U_WB_ALETTER = 1, /*[LE]*/
2370  U_WB_FORMAT = 2, /*[FO]*/
2371  U_WB_KATAKANA = 3, /*[KA]*/
2372  U_WB_MIDLETTER = 4, /*[ML]*/
2373  U_WB_MIDNUM = 5, /*[MN]*/
2374  U_WB_NUMERIC = 6, /*[NU]*/
2375  U_WB_EXTENDNUMLET = 7, /*[EX]*/
2377  U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2379  U_WB_EXTEND = 9, /*[Extend]*/
2381  U_WB_LF = 10, /*[LF]*/
2383  U_WB_MIDNUMLET =11, /*[MB]*/
2385  U_WB_NEWLINE =12, /*[NL]*/
2387  U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2389  U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2391  U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2393  U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2395  U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2397  U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2399  U_WB_E_MODIFIER = 19, /*[EM]*/
2401  U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2403  U_WB_ZWJ = 21, /*[ZWJ]*/
2405  U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/
2406 
2407 #ifndef U_HIDE_DEPRECATED_API
2408 
2415 #endif // U_HIDE_DEPRECATED_API
2417 
2424 typedef enum USentenceBreak {
2425  /*
2426  * Note: USentenceBreak constants are parsed by preparseucd.py.
2427  * It matches lines like
2428  * U_SB_<Unicode Sentence_Break value name>
2429  */
2430 
2431  U_SB_OTHER = 0, /*[XX]*/
2432  U_SB_ATERM = 1, /*[AT]*/
2433  U_SB_CLOSE = 2, /*[CL]*/
2434  U_SB_FORMAT = 3, /*[FO]*/
2435  U_SB_LOWER = 4, /*[LO]*/
2436  U_SB_NUMERIC = 5, /*[NU]*/
2437  U_SB_OLETTER = 6, /*[LE]*/
2438  U_SB_SEP = 7, /*[SE]*/
2439  U_SB_SP = 8, /*[SP]*/
2440  U_SB_STERM = 9, /*[ST]*/
2441  U_SB_UPPER = 10, /*[UP]*/
2442  U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2443  U_SB_EXTEND = 12, /*[EX]*/
2444  U_SB_LF = 13, /*[LF]*/
2445  U_SB_SCONTINUE = 14, /*[SC]*/
2446 #ifndef U_HIDE_DEPRECATED_API
2447 
2454 #endif // U_HIDE_DEPRECATED_API
2455 } USentenceBreak;
2456 
2463 typedef enum ULineBreak {
2464  /*
2465  * Note: ULineBreak constants are parsed by preparseucd.py.
2466  * It matches lines like
2467  * U_LB_<Unicode Line_Break value name>
2468  */
2469 
2470  U_LB_UNKNOWN = 0, /*[XX]*/
2471  U_LB_AMBIGUOUS = 1, /*[AI]*/
2472  U_LB_ALPHABETIC = 2, /*[AL]*/
2473  U_LB_BREAK_BOTH = 3, /*[B2]*/
2474  U_LB_BREAK_AFTER = 4, /*[BA]*/
2475  U_LB_BREAK_BEFORE = 5, /*[BB]*/
2476  U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2477  U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2478  U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2479  U_LB_COMBINING_MARK = 9, /*[CM]*/
2480  U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2481  U_LB_EXCLAMATION = 11, /*[EX]*/
2482  U_LB_GLUE = 12, /*[GL]*/
2483  U_LB_HYPHEN = 13, /*[HY]*/
2484  U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2486  U_LB_INSEPARABLE = 15, /*[IN]*/
2487  U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2488  U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2489  U_LB_LINE_FEED = 17, /*[LF]*/
2490  U_LB_NONSTARTER = 18, /*[NS]*/
2491  U_LB_NUMERIC = 19, /*[NU]*/
2492  U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2493  U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2494  U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2495  U_LB_QUOTATION = 23, /*[QU]*/
2496  U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2497  U_LB_SURROGATE = 25, /*[SG]*/
2498  U_LB_SPACE = 26, /*[SP]*/
2499  U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2500  U_LB_ZWSPACE = 28, /*[ZW]*/
2502  U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2504  U_LB_WORD_JOINER = 30, /*[WJ]*/
2506  U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2508  U_LB_H3 = 32, /*[H3]*/
2510  U_LB_JL = 33, /*[JL]*/
2512  U_LB_JT = 34, /*[JT]*/
2514  U_LB_JV = 35, /*[JV]*/
2516  U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2518  U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2520  U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2522  U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2524  U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2526  U_LB_E_MODIFIER = 41, /*[EM]*/
2528  U_LB_ZWJ = 42, /*[ZWJ]*/
2530  U_LB_AKSARA = 43, /*[AK]*/
2532  U_LB_AKSARA_PREBASE = 44, /*[AP]*/
2534  U_LB_AKSARA_START = 45, /*[AS]*/
2536  U_LB_VIRAMA_FINAL = 46, /*[VF]*/
2538  U_LB_VIRAMA = 47, /*[VI]*/
2539 #ifndef U_HIDE_DEPRECATED_API
2540 
2547 #endif // U_HIDE_DEPRECATED_API
2548 } ULineBreak;
2549 
2556 typedef enum UNumericType {
2557  /*
2558  * Note: UNumericType constants are parsed by preparseucd.py.
2559  * It matches lines like
2560  * U_NT_<Unicode Numeric_Type value name>
2561  */
2562 
2563  U_NT_NONE, /*[None]*/
2564  U_NT_DECIMAL, /*[de]*/
2565  U_NT_DIGIT, /*[di]*/
2566  U_NT_NUMERIC, /*[nu]*/
2567 #ifndef U_HIDE_DEPRECATED_API
2568 
2575 #endif // U_HIDE_DEPRECATED_API
2576 } UNumericType;
2577 
2584 typedef enum UHangulSyllableType {
2585  /*
2586  * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2587  * It matches lines like
2588  * U_HST_<Unicode Hangul_Syllable_Type value name>
2589  */
2590 
2591  U_HST_NOT_APPLICABLE, /*[NA]*/
2592  U_HST_LEADING_JAMO, /*[L]*/
2593  U_HST_VOWEL_JAMO, /*[V]*/
2594  U_HST_TRAILING_JAMO, /*[T]*/
2595  U_HST_LV_SYLLABLE, /*[LV]*/
2596  U_HST_LVT_SYLLABLE, /*[LVT]*/
2597 #ifndef U_HIDE_DEPRECATED_API
2598 
2605 #endif // U_HIDE_DEPRECATED_API
2607 
2615  /*
2616  * Note: UIndicPositionalCategory constants are parsed by preparseucd.py.
2617  * It matches lines like
2618  * U_INPC_<Unicode Indic_Positional_Category value name>
2619  */
2620 
2654 
2662  /*
2663  * Note: UIndicSyllabicCategory constants are parsed by preparseucd.py.
2664  * It matches lines like
2665  * U_INSC_<Unicode Indic_Syllabic_Category value name>
2666  */
2667 
2743 
2744 #ifndef U_HIDE_DRAFT_API
2745 
2751 typedef enum UIndicConjunctBreak {
2752  /*
2753  * Note: UIndicConjunctBreak constants are parsed by preparseucd.py.
2754  * It matches lines like
2755  * U_INCB_<Unicode Indic_Conjunct_Break value name>
2756  */
2757 
2767 #endif // U_HIDE_DRAFT_API
2768 
2775 typedef enum UVerticalOrientation {
2776  /*
2777  * Note: UVerticalOrientation constants are parsed by preparseucd.py.
2778  * It matches lines like
2779  * U_VO_<Unicode Vertical_Orientation value name>
2780  */
2781 
2791 
2799 typedef enum UIdentifierStatus {
2800  /*
2801  * Note: UIdentifierStatus constants are parsed by preparseucd.py.
2802  * It matches lines like
2803  * U_ID_STATUS_<Unicode Identifier_Status value name>
2804  */
2805 
2811 
2819 typedef enum UIdentifierType {
2820  /*
2821  * Note: UIdentifierType constants are parsed by preparseucd.py.
2822  * It matches lines like
2823  * U_ID_TYPE_<Unicode Identifier_Type value name>
2824  */
2825 
2850 } UIdentifierType;
2851 
2879 U_CAPI UBool U_EXPORT2
2881 
2906 U_CAPI UBool U_EXPORT2
2907 u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which);
2908 
2924 U_CAPI const USet * U_EXPORT2
2925 u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode);
2926 
2939 U_CAPI UBool U_EXPORT2
2941 
2954 U_CAPI UBool U_EXPORT2
2956 
2969 U_CAPI UBool U_EXPORT2
2971 
2990 U_CAPI UBool U_EXPORT2
2992 
3031 U_CAPI int32_t U_EXPORT2
3033 
3052 U_CAPI int32_t U_EXPORT2
3054 
3081 U_CAPI int32_t U_EXPORT2
3083 
3099 U_CAPI const UCPMap * U_EXPORT2
3100 u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode);
3101 
3124 U_CAPI double U_EXPORT2
3126 
3134 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
3135 
3159 U_CAPI UBool U_EXPORT2
3160 u_islower(UChar32 c);
3161 
3186 U_CAPI UBool U_EXPORT2
3187 u_isupper(UChar32 c);
3188 
3203 U_CAPI UBool U_EXPORT2
3204 u_istitle(UChar32 c);
3205 
3224 U_CAPI UBool U_EXPORT2
3225 u_isdigit(UChar32 c);
3226 
3245 U_CAPI UBool U_EXPORT2
3246 u_isalpha(UChar32 c);
3247 
3266 U_CAPI UBool U_EXPORT2
3267 u_isalnum(UChar32 c);
3268 
3289 U_CAPI UBool U_EXPORT2
3290 u_isxdigit(UChar32 c);
3291 
3305 U_CAPI UBool U_EXPORT2
3306 u_ispunct(UChar32 c);
3307 
3324 U_CAPI UBool U_EXPORT2
3325 u_isgraph(UChar32 c);
3326 
3352 U_CAPI UBool U_EXPORT2
3353 u_isblank(UChar32 c);
3354 
3377 U_CAPI UBool U_EXPORT2
3378 u_isdefined(UChar32 c);
3379 
3398 U_CAPI UBool U_EXPORT2
3399 u_isspace(UChar32 c);
3400 
3419 U_CAPI UBool U_EXPORT2
3421 
3459 U_CAPI UBool U_EXPORT2
3461 
3483 U_CAPI UBool U_EXPORT2
3484 u_iscntrl(UChar32 c);
3485 
3498 U_CAPI UBool U_EXPORT2
3500 
3516 U_CAPI UBool U_EXPORT2
3517 u_isprint(UChar32 c);
3518 
3536 U_CAPI UBool U_EXPORT2
3537 u_isbase(UChar32 c);
3538 
3555 U_CAPI UCharDirection U_EXPORT2
3557 
3573 U_CAPI UBool U_EXPORT2
3575 
3595 U_CAPI UChar32 U_EXPORT2
3597 
3614 U_CAPI UChar32 U_EXPORT2
3616 
3628 U_CAPI int8_t U_EXPORT2
3629 u_charType(UChar32 c);
3630 
3644 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3645 
3663 typedef UBool U_CALLCONV
3664 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3665 
3685 U_CAPI void U_EXPORT2
3686 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3687 
3688 #if !UCONFIG_NO_NORMALIZATION
3689 
3697 U_CAPI uint8_t U_EXPORT2
3699 
3700 #endif
3701 
3725 U_CAPI int32_t U_EXPORT2
3727 
3737 U_CAPI UBlockCode U_EXPORT2
3739 
3772 U_CAPI int32_t U_EXPORT2
3773 u_charName(UChar32 code, UCharNameChoice nameChoice,
3774  char *buffer, int32_t bufferLength,
3775  UErrorCode *pErrorCode);
3776 
3777 #ifndef U_HIDE_DEPRECATED_API
3778 
3796 U_DEPRECATED int32_t U_EXPORT2
3798  char *dest, int32_t destCapacity,
3799  UErrorCode *pErrorCode);
3800 #endif /* U_HIDE_DEPRECATED_API */
3801 
3822 U_CAPI UChar32 U_EXPORT2
3823 u_charFromName(UCharNameChoice nameChoice,
3824  const char *name,
3825  UErrorCode *pErrorCode);
3826 
3844 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3845  UChar32 code,
3846  UCharNameChoice nameChoice,
3847  const char *name,
3848  int32_t length);
3849 
3871 U_CAPI void U_EXPORT2
3872 u_enumCharNames(UChar32 start, UChar32 limit,
3873  UEnumCharNamesFn *fn,
3874  void *context,
3875  UCharNameChoice nameChoice,
3876  UErrorCode *pErrorCode);
3877 
3909 U_CAPI const char* U_EXPORT2
3910 u_getPropertyName(UProperty property,
3911  UPropertyNameChoice nameChoice);
3912 
3932 U_CAPI UProperty U_EXPORT2
3933 u_getPropertyEnum(const char* alias);
3934 
3982 U_CAPI const char* U_EXPORT2
3984  int32_t value,
3985  UPropertyNameChoice nameChoice);
3986 
4018 U_CAPI int32_t U_EXPORT2
4020  const char* alias);
4021 
4036 U_CAPI UBool U_EXPORT2
4037 u_isIDStart(UChar32 c);
4038 
4053 U_CAPI UBool U_EXPORT2
4054 u_isIDPart(UChar32 c);
4055 
4069 U_CAPI bool U_EXPORT2
4071 
4104 U_CAPI int32_t U_EXPORT2
4105 u_getIDTypes(UChar32 c, UIdentifierType *types, int32_t capacity, UErrorCode *pErrorCode);
4106 
4127 U_CAPI UBool U_EXPORT2
4129 
4146 U_CAPI UBool U_EXPORT2
4148 
4167 U_CAPI UBool U_EXPORT2
4169 
4192 U_CAPI UChar32 U_EXPORT2
4193 u_tolower(UChar32 c);
4194 
4217 U_CAPI UChar32 U_EXPORT2
4218 u_toupper(UChar32 c);
4219 
4242 U_CAPI UChar32 U_EXPORT2
4243 u_totitle(UChar32 c);
4244 
4267 U_CAPI UChar32 U_EXPORT2
4268 u_foldCase(UChar32 c, uint32_t options);
4269 
4308 U_CAPI int32_t U_EXPORT2
4309 u_digit(UChar32 ch, int8_t radix);
4310 
4339 U_CAPI UChar32 U_EXPORT2
4340 u_forDigit(int32_t digit, int8_t radix);
4341 
4356 U_CAPI void U_EXPORT2
4357 u_charAge(UChar32 c, UVersionInfo versionArray);
4358 
4370 U_CAPI void U_EXPORT2
4371 u_getUnicodeVersion(UVersionInfo versionArray);
4372 
4373 #if !UCONFIG_NO_NORMALIZATION
4374 
4395 U_CAPI int32_t U_EXPORT2
4396 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
4397 
4398 #endif
4399 
4400 
4402 
4403 #endif /*_UCHAR*/
4404 /*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:272
Unicode 4.0.1 renames the &quot;Cyrillic Supplementary&quot; block to &quot;Cyrillic Supplement&quot;.
Definition: uchar.h:1450
Enumerated property Indic_Positional_Category.
Definition: uchar.h:663
Binary property Changes_When_Lowercased.
Definition: uchar.h:424
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition: uchar.h:276
Binary property ID_Compat_Math_Continue.
Definition: uchar.h:554
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:422
Miscellaneous property Identifier_Type.
Definition: uchar.h:812
Enumerated property NFC_Quick_Check.
Definition: uchar.h:618
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:2299
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:796
UIndicConjunctBreak
Indic Conjunct Break constants.
Definition: uchar.h:2751
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1392
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:575
Binary property Modifier_Combining_Mark.
Definition: uchar.h:561
Binary property XID_Start.
Definition: uchar.h:329
Binary property Join_Control.
Definition: uchar.h:283
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition: uchar.h:287
struct UCPMap UCPMap
Abstract map from Unicode code points (U+0000..U+10FFFF) to integer values.
Definition: ucpmap.h:30
Binary property White_Space.
Definition: uchar.h:322
String property Titlecase_Mapping.
Definition: uchar.h:769
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:2092
U_CAPI UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
UIdentifierType
Identifier Type constants.
Definition: uchar.h:2819
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:698
Enumerated property Numeric_Type.
Definition: uchar.h:603
Binary property xdigit (a C/POSIX character class).
Definition: uchar.h:418
Binary property Alphabetic.
Definition: uchar.h:210
First constant for double Unicode properties.
Definition: uchar.h:724
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:3844
Binary property of strings RGI_Emoji.
Definition: uchar.h:535
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 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:2453
Binary property Emoji_Component.
Definition: uchar.h:469
Binary property of strings Emoji_Keycap_Sequence.
Definition: uchar.h:500
Binary property Emoji.
Definition: uchar.h:441
Cn &quot;Other, Not Assigned (no characters in [UnicodeData.txt] have this property)&quot; (same as U_UNASSIGNE...
Definition: uchar.h:842
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...
U_CAPI int32_t u_getIDTypes(UChar32 c, UIdentifierType *types, int32_t capacity, UErrorCode *pErrorCode)
Writes code point c&#39;s Identifier_Type as a list of UIdentifierType values to the output types array a...
String property Simple_Case_Folding.
Definition: uchar.h:757
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:365
Binary property graph (a C/POSIX character class).
Definition: uchar.h:408
String property Bidi_Mirroring_Glyph.
Definition: uchar.h:740
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:717
U_CAPI UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
Enumerated property Block.
Definition: uchar.h:578
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:822
Renamed from the misspelled &quot;inseperable&quot; in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2486
Binary property IDS_Unary_Operator.
Definition: uchar.h:542
Binary property Radical (new in Unicode 3.2).
Definition: uchar.h:302
UCharDirection
This specifies the language directional property of a character set.
Definition: uchar.h:1027
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition: uchar.h:280
Binary property Grapheme_Link (new in Unicode 3.2).
Definition: uchar.h:254
Enumerated property Decomposition_Type.
Definition: uchar.h:584
Binary property Regional_Indicator.
Definition: uchar.h:474
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:846
String property Case_Folding.
Definition: uchar.h:743
String property Name.
Definition: uchar.h:754
C API: This file defines an abstract map from Unicode code points to integer values.
String property Simple_Uppercase_Mapping.
Definition: uchar.h:766
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:573
Enumerated property General_Category.
Definition: uchar.h:591
Binary property Prepended_Concatenation_Mark.
Definition: uchar.h:479
U_CAPI UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
Sm.
Definition: uchar.h:890
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:775
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:2556
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:1118
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:393
Enumerated property Vertical_Orientation.
Definition: uchar.h:677
Close paired bracket.
Definition: uchar.h:1110
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:3664
Binary property Changes_When_Casefolded.
Definition: uchar.h:430
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:351
Binary property Diacritic.
Definition: uchar.h:235
Binary property Terminal_Punctuation.
Definition: uchar.h:311
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:2574
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:612
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.
Enumerated property Identifier_Status.
Definition: uchar.h:684
#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:338
Enumerated property Joining_Group.
Definition: uchar.h:594
Binary property ID_Continue.
Definition: uchar.h:265
Binary property Emoji_Presentation.
Definition: uchar.h:448
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:1108
Binary property blank (a C/POSIX character class).
Definition: uchar.h:403
Binary property Quotation_Mark.
Definition: uchar.h:298
#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:434
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:212
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:296
Binary property Hyphen.
Definition: uchar.h:260
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:588
ULineBreak
Line Break constants.
Definition: uchar.h:2463
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:1088
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:243
Bitmask property General_Category_Mask.
Definition: uchar.h:709
String property Simple_Titlecase_Mapping.
Definition: uchar.h:763
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2350
Unicode 3.2 renames this block to &quot;Combining Diacritical Marks for Symbols&quot;.
Definition: uchar.h:1269
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:2102
One more than the highest normal ULineBreak value.
Definition: uchar.h:2546
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2584
String property Simple_Lowercase_Mapping.
Definition: uchar.h:760
UIndicPositionalCategory
Indic Positional Category constants.
Definition: uchar.h:2614
Binary property print (a C/POSIX character class).
Definition: uchar.h:413
Binary property Case_Sensitive.
Definition: uchar.h:333
Standard or synthetic character name.
Definition: uchar.h:2059
Binary property Bidi_Mirrored.
Definition: uchar.h:223
UVerticalOrientation
Vertical Orientation constants.
Definition: uchar.h:2775
Binary property NFKC_Inert.
Definition: uchar.h:372
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:432
First constant for string Unicode properties.
Definition: uchar.h:737
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition: uchar.h:251
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:427
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:2309
New No_Block value in Unicode 4.
Definition: uchar.h:1134
Binary property Extender.
Definition: uchar.h:239
Double property Numeric_Value.
Definition: uchar.h:722
Binary property Math.
Definition: uchar.h:292
Unicode character name (Name property).
Definition: uchar.h:2049
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:247
Binary property NFKD_Inert.
Definition: uchar.h:358
U_CAPI UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier...
Enumerated property Indic_Conjunct_Break.
Definition: uchar.h:691
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:830
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:645
Binary property Lowercase.
Definition: uchar.h:290
First constant for bit-mask Unicode properties.
Definition: uchar.h:711
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:655
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:2424
Binary property Extended_Pictographic.
Definition: uchar.h:486
Enumerated property Indic_Syllabic_Category.
Definition: uchar.h:670
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:315
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:581
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:2047
One more than the last constant for binary Unicode properties.
Definition: uchar.h:568
Enumerated property Script.
Definition: uchar.h:606
Unicode 3.2 renames this block to &quot;Greek and Coptic&quot;.
Definition: uchar.h:1161
Binary property Hex_Digit.
Definition: uchar.h:257
String property Uppercase_Mapping.
Definition: uchar.h:779
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:2084
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:751
char16_t UChar
The base type for UTF-16 code units and pointers.
Definition: umachine.h:378
#define U_CDECL_END
This is used to end a declaration of a library private ICU C API.
Definition: umachine.h:86
Enumerated property NFKC_Quick_Check.
Definition: uchar.h:621
Not a paired bracket.
Definition: uchar.h:1106
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:196
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:609
Binary property Dash.
Definition: uchar.h:225
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:398
Cf.
Definition: uchar.h:874
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition: uchar.h:344
Binary property of strings RGI_Emoji_ZWJ_Sequence.
Definition: uchar.h:528
One more than the highest normal UHangulSyllableType value.
Definition: uchar.h:2604
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:1126
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:650
Binary property Deprecated (new in Unicode 3.2).
Definition: uchar.h:232
Binary property Bidi_Control.
Definition: uchar.h:218
Binary property XID_Continue.
Definition: uchar.h:326
Same as UBLOCK_PRIVATE_USE.
Definition: uchar.h:1382
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:430
Binary property Uppercase.
Definition: uchar.h:318
Binary property Changes_When_Uppercased.
Definition: uchar.h:426
UJoiningGroup
Joining Group constants.
Definition: uchar.h:2174
Binary property Cased.
Definition: uchar.h:420
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition: uset.h:54
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:2032
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:2134
Cs.
Definition: uchar.h:878
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:2012
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:635
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:840
First constant for Unicode properties with unusual value types.
Definition: uchar.h:798
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:2361
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:597
One more than the last constant for double Unicode properties.
Definition: uchar.h:730
Binary property Emoji_Modifier.
Definition: uchar.h:455
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:782
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:628
Binary property ASCII_Hex_Digit.
Definition: uchar.h:214
UJoiningType
Joining Type constants.
Definition: uchar.h:2144
Binary property Soft_Dotted (new in Unicode 3.2).
Definition: uchar.h:307
One more than the last constant for string Unicode properties.
Definition: uchar.h:788
Binary property Emoji_Modifier_Base.
Definition: uchar.h:462
Binary Property Segment_Starter.
Definition: uchar.h:383
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:269
Corrected name from NameAliases.txt.
Definition: uchar.h:2061
Binary property Changes_When_Titlecased.
Definition: uchar.h:428
Enumerated property Line_Break.
Definition: uchar.h:600
One more than the highest normal UJoiningType value.
Definition: uchar.h:2164
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...
UIdentifierStatus
Identifier Status constants.
Definition: uchar.h:2799
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:640
Binary property ID_Compat_Math_Start.
Definition: uchar.h:548
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:747
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 as a non-initial character of an identifier acco...
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:229
One more than the highest normal UBlockCode value.
Definition: uchar.h:1995
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:2056
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:908
Binary property of strings RGI_Emoji_Tag_Sequence.
Definition: uchar.h:521
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2414
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition: uchar.h:388
One more than the last constant for Unicode properties with unusual value types.
Definition: uchar.h:818
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:1098
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:2067
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition: uchar.h:2661
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:735
int8_t UBool
The ICU boolean type, a signed-byte integer.
Definition: umachine.h:247
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:615
U_CAPI UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.