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
537 
556 #endif // U_HIDE_DRAFT_API
557 #ifndef U_HIDE_DEPRECATED_API
558 
563 #endif // U_HIDE_DEPRECATED_API
564 
572  UCHAR_BLOCK=0x1001,
600  UCHAR_SCRIPT=0x100A,
672 #ifndef U_HIDE_DRAFT_API
673 
680 #endif // U_HIDE_DRAFT_API
681 #ifndef U_HIDE_DEPRECATED_API
682 
687 #endif // U_HIDE_DEPRECATED_API
688 
700 #ifndef U_HIDE_DEPRECATED_API
701 
706 #endif // U_HIDE_DEPRECATED_API
707 
713 #ifndef U_HIDE_DEPRECATED_API
714 
719 #endif // U_HIDE_DEPRECATED_API
720 
723  UCHAR_AGE=0x4000,
732 #ifndef U_HIDE_DEPRECATED_API
733 
736 #endif /* U_HIDE_DEPRECATED_API */
737 
742  UCHAR_NAME=0x4005,
758 #ifndef U_HIDE_DEPRECATED_API
759 
764 #endif /* U_HIDE_DEPRECATED_API */
765 
771 #ifndef U_HIDE_DEPRECATED_API
772 
777 #endif // U_HIDE_DEPRECATED_API
778 
787 #ifndef U_HIDE_DRAFT_API
788 
802 #endif // U_HIDE_DRAFT_API
803 #ifndef U_HIDE_DEPRECATED_API
804 
809 #endif // U_HIDE_DEPRECATED_API
810 
813 } UProperty;
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 
899 } UCharCategory;
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)
919 
920 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
921 
922 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
923 
924 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
925 
926 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
927 
929 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
930 
931 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
932 
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)
937 
938 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
939 
940 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
941 
943 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
944 
945 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
946 
947 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
948 
950 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
951 
952 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
953 
954 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
955 
956 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
957 
959 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
960 
961 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
962 
963 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
964 
965 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
966 
967 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
968 
970 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
971 
972 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
973 
974 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
975 
976 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
977 
979 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
980 
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
1072 
1079 #endif // U_HIDE_DEPRECATED_API
1080 } UCharDirection;
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
1102 
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
1956 
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
1993 
2000 #endif // U_HIDE_DEPRECATED_API
2001 } UEastAsianWidth;
2002 
2014 typedef enum UCharNameChoice {
2017 #ifndef U_HIDE_DEPRECATED_API
2018 
2024 #endif /* U_HIDE_DEPRECATED_API */
2025 
2029 #ifndef U_HIDE_DEPRECATED_API
2030 
2035 #endif // U_HIDE_DEPRECATED_API
2036 } UCharNameChoice;
2037 
2051 typedef enum UPropertyNameChoice {
2052  U_SHORT_PROPERTY_NAME,
2053  U_LONG_PROPERTY_NAME,
2054 #ifndef U_HIDE_DEPRECATED_API
2055 
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
2095 
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
2125 
2131  U_JT_COUNT /* 6 */
2132 #endif // U_HIDE_DEPRECATED_API
2133 } UJoiningType;
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
2258 
2265 #endif // U_HIDE_DEPRECATED_API
2266 } UJoiningGroup;
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
2309 
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
2373 
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
2412 
2419 #endif // U_HIDE_DEPRECATED_API
2420 } USentenceBreak;
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
2505 
2512 #endif // U_HIDE_DEPRECATED_API
2513 } ULineBreak;
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
2533 
2540 #endif // U_HIDE_DEPRECATED_API
2541 } UNumericType;
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
2563 
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
2731 
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 
2789 } UIdentifierType;
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
2865 u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode);
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
3040 u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode);
3041 
3064 U_CAPI double U_EXPORT2
3066 
3074 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
3075 
3099 U_CAPI UBool U_EXPORT2
3100 u_islower(UChar32 c);
3101 
3126 U_CAPI UBool U_EXPORT2
3127 u_isupper(UChar32 c);
3128 
3143 U_CAPI UBool U_EXPORT2
3144 u_istitle(UChar32 c);
3145 
3164 U_CAPI UBool U_EXPORT2
3165 u_isdigit(UChar32 c);
3166 
3185 U_CAPI UBool U_EXPORT2
3186 u_isalpha(UChar32 c);
3187 
3206 U_CAPI UBool U_EXPORT2
3207 u_isalnum(UChar32 c);
3208 
3229 U_CAPI UBool U_EXPORT2
3230 u_isxdigit(UChar32 c);
3231 
3245 U_CAPI UBool U_EXPORT2
3246 u_ispunct(UChar32 c);
3247 
3264 U_CAPI UBool U_EXPORT2
3265 u_isgraph(UChar32 c);
3266 
3292 U_CAPI UBool U_EXPORT2
3293 u_isblank(UChar32 c);
3294 
3317 U_CAPI UBool U_EXPORT2
3318 u_isdefined(UChar32 c);
3319 
3338 U_CAPI UBool U_EXPORT2
3339 u_isspace(UChar32 c);
3340 
3359 U_CAPI UBool U_EXPORT2
3361 
3399 U_CAPI UBool U_EXPORT2
3401 
3423 U_CAPI UBool U_EXPORT2
3424 u_iscntrl(UChar32 c);
3425 
3438 U_CAPI UBool U_EXPORT2
3440 
3456 U_CAPI UBool U_EXPORT2
3457 u_isprint(UChar32 c);
3458 
3476 U_CAPI UBool U_EXPORT2
3477 u_isbase(UChar32 c);
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
3569 u_charType(UChar32 c);
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
3713 u_charName(UChar32 code, UCharNameChoice nameChoice,
3714  char *buffer, int32_t bufferLength,
3715  UErrorCode *pErrorCode);
3716 
3717 #ifndef U_HIDE_DEPRECATED_API
3718 
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
3763 u_charFromName(UCharNameChoice nameChoice,
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
3812 u_enumCharNames(UChar32 start, UChar32 limit,
3813  UEnumCharNamesFn *fn,
3814  void *context,
3815  UCharNameChoice nameChoice,
3816  UErrorCode *pErrorCode);
3817 
3849 U_CAPI const char* U_EXPORT2
3850 u_getPropertyName(UProperty property,
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
3977 u_isIDStart(UChar32 c);
3978 
3993 U_CAPI UBool U_EXPORT2
3994 u_isIDPart(UChar32 c);
3995 
3996 #ifndef U_HIDE_DRAFT_API
3997 
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
4135 u_tolower(UChar32 c);
4136 
4159 U_CAPI UChar32 U_EXPORT2
4160 u_toupper(UChar32 c);
4161 
4184 U_CAPI UChar32 U_EXPORT2
4185 u_totitle(UChar32 c);
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
4313 u_getUnicodeVersion(UVersionInfo versionArray);
4314 
4315 #if !UCONFIG_NO_NORMALIZATION
4316 
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*/
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:1440
Enumerated property Indic_Positional_Category.
Definition: uchar.h:657
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:555
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:801
Enumerated property NFC_Quick_Check.
Definition: uchar.h:612
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:2264
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:784
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1382
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:569
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:757
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:2059
U_CAPI UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
UIdentifierType
Identifier Type constants.
Definition: uchar.h:2758
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:686
Enumerated property Numeric_Type.
Definition: uchar.h:597
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:712
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:3784
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:2418
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:832
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:745
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:728
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:705
U_CAPI UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
Enumerated property Block.
Definition: uchar.h:572
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:812
Renamed from the misspelled &quot;inseperable&quot; in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2451
Binary property IDS_Unary_Operator.
Definition: uchar.h:543
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:1017
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:578
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:834
String property Case_Folding.
Definition: uchar.h:731
String property Name.
Definition: uchar.h:742
C API: This file defines an abstract map from Unicode code points to integer values.
String property Simple_Uppercase_Mapping.
Definition: uchar.h:754
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:567
Enumerated property General_Category.
Definition: uchar.h:585
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:880
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:763
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:2521
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:1108
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:393
Enumerated property Vertical_Orientation.
Definition: uchar.h:671
Close paired bracket.
Definition: uchar.h:1100
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:3604
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:2539
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:606
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:679
#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:588
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:1098
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:582
ULineBreak
Line Break constants.
Definition: uchar.h:2428
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:1078
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:697
String property Simple_Titlecase_Mapping.
Definition: uchar.h:751
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2315
Unicode 3.2 renames this block to &quot;Combining Diacritical Marks for Symbols&quot;.
Definition: uchar.h:1259
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:2069
One more than the highest normal ULineBreak value.
Definition: uchar.h:2511
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2549
String property Simple_Lowercase_Mapping.
Definition: uchar.h:748
UIndicPositionalCategory
Indic Positional Category constants.
Definition: uchar.h:2579
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:2026
Binary property Bidi_Mirrored.
Definition: uchar.h:223
UVerticalOrientation
Vertical Orientation constants.
Definition: uchar.h:2713
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:725
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:2274
New No_Block value in Unicode 4.
Definition: uchar.h:1124
Binary property Extender.
Definition: uchar.h:239
Double property Numeric_Value.
Definition: uchar.h:710
Binary property Math.
Definition: uchar.h:292
Unicode character name (Name property).
Definition: uchar.h:2016
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...
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:820
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:639
Binary property Lowercase.
Definition: uchar.h:290
First constant for bit-mask Unicode properties.
Definition: uchar.h:699
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:649
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:2389
Binary property Extended_Pictographic.
Definition: uchar.h:486
Enumerated property Indic_Syllabic_Category.
Definition: uchar.h:664
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:315
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:575
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:2014
One more than the last constant for binary Unicode properties.
Definition: uchar.h:562
Enumerated property Script.
Definition: uchar.h:600
Unicode 3.2 renames this block to &quot;Greek and Coptic&quot;.
Definition: uchar.h:1151
Binary property Hex_Digit.
Definition: uchar.h:257
String property Uppercase_Mapping.
Definition: uchar.h:767
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:2051
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:739
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:615
Not a paired bracket.
Definition: uchar.h:1096
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:603
Binary property Dash.
Definition: uchar.h:225
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:398
Cf.
Definition: uchar.h:864
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:2569
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:1116
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:644
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:1372
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415
Binary property Uppercase.
Definition: uchar.h:318
Binary property Changes_When_Uppercased.
Definition: uchar.h:426
UJoiningGroup
Joining Group constants.
Definition: uchar.h:2141
Binary property Cased.
Definition: uchar.h:420
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition: uset.h:50
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:1999
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:2101
Cs.
Definition: uchar.h:868
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:1979
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:629
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:830
First constant for Unicode properties with unusual value types.
Definition: uchar.h:786
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:2326
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:591
One more than the last constant for double Unicode properties.
Definition: uchar.h:718
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:770
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:622
Binary property ASCII_Hex_Digit.
Definition: uchar.h:214
UJoiningType
Joining Type constants.
Definition: uchar.h:2111
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:776
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:2028
Binary property Changes_When_Titlecased.
Definition: uchar.h:428
Enumerated property Line_Break.
Definition: uchar.h:594
One more than the highest normal UJoiningType value.
Definition: uchar.h:2131
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:2738
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:634
Binary property ID_Compat_Math_Start.
Definition: uchar.h:549
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:735
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:1962
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:2023
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:898
Binary property of strings RGI_Emoji_Tag_Sequence.
Definition: uchar.h:521
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2379
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:808
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:1088
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:2034
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition: uchar.h:2626
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:723
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:609
U_CAPI UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.