ICU 73.2  73.2
uchar.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 **********************************************************************
5 * Copyright (C) 1997-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
8 *
9 * File UCHAR.H
10 *
11 * Modification History:
12 *
13 * Date Name Description
14 * 04/02/97 aliu Creation.
15 * 03/29/99 helena Updated for C APIs.
16 * 4/15/99 Madhu Updated for C Implementation and Javadoc
17 * 5/20/99 Madhu Added the function u_getVersion()
18 * 8/19/1999 srl Upgraded scripts to Unicode 3.0
19 * 8/27/1999 schererm UCharDirection constants: U_...
20 * 11/11/1999 weiv added u_isalnum(), cleaned comments
21 * 01/11/2000 helena Renamed u_getVersion to u_getUnicodeVersion().
22 ******************************************************************************
23 */
24 
25 #ifndef UCHAR_H
26 #define UCHAR_H
27 
28 #include "unicode/utypes.h"
29 #include "unicode/stringoptions.h"
30 #include "unicode/ucpmap.h"
31 
32 #if !defined(USET_DEFINED) && !defined(U_IN_DOXYGEN)
33 
34 #define USET_DEFINED
35 
44 typedef struct USet USet;
45 
46 #endif
47 
48 
50 
51 /*==========================================================================*/
52 /* Unicode version number */
53 /*==========================================================================*/
63 #define U_UNICODE_VERSION "15.0"
64 
157 #define UCHAR_MIN_VALUE 0
158 
167 #define UCHAR_MAX_VALUE 0x10ffff
168 
173 #define U_MASK(x) ((uint32_t)1<<(x))
174 
195 typedef enum UProperty {
196  /*
197  * Note: UProperty constants are parsed by preparseucd.py.
198  * It matches lines like
199  * UCHAR_<Unicode property name>=<integer>,
200  */
201 
202  /* Note: Place UCHAR_ALPHABETIC before UCHAR_BINARY_START so that
203  debuggers display UCHAR_ALPHABETIC as the symbolic name for 0,
204  rather than UCHAR_BINARY_START. Likewise for other *_START
205  identifiers. */
206 
535 #ifndef U_HIDE_DEPRECATED_API
536 
541 #endif // U_HIDE_DEPRECATED_API
542 
550  UCHAR_BLOCK=0x1001,
578  UCHAR_SCRIPT=0x100A,
650 #ifndef U_HIDE_DEPRECATED_API
651 
656 #endif // U_HIDE_DEPRECATED_API
657 
669 #ifndef U_HIDE_DEPRECATED_API
670 
675 #endif // U_HIDE_DEPRECATED_API
676 
682 #ifndef U_HIDE_DEPRECATED_API
683 
688 #endif // U_HIDE_DEPRECATED_API
689 
692  UCHAR_AGE=0x4000,
701 #ifndef U_HIDE_DEPRECATED_API
702 
705 #endif /* U_HIDE_DEPRECATED_API */
706 
711  UCHAR_NAME=0x4005,
727 #ifndef U_HIDE_DEPRECATED_API
728 
733 #endif /* U_HIDE_DEPRECATED_API */
734 
740 #ifndef U_HIDE_DEPRECATED_API
741 
746 #endif // U_HIDE_DEPRECATED_API
747 
756 #ifndef U_HIDE_DEPRECATED_API
757 
762 #endif // U_HIDE_DEPRECATED_API
763 
766 } UProperty;
767 
773 typedef enum UCharCategory
774 {
775  /*
776  * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
777  * It matches pairs of lines like
778  * / ** <Unicode 2-letter General_Category value> comment... * /
779  * U_<[A-Z_]+> = <integer>,
780  */
781 
852 } UCharCategory;
853 
868 #define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES)
869 
871 #define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER)
872 
873 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
874 
875 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
876 
877 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
878 
879 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
880 
882 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
883 
884 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
885 
886 #define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK)
887 
889 #define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER)
890 
891 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
892 
893 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
894 
896 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
897 
898 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
899 
900 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
901 
903 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
904 
905 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
906 
907 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
908 
909 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
910 
912 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
913 
914 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
915 
916 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
917 
918 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
919 
920 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
921 
923 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
924 
925 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
926 
927 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
928 
929 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
930 
932 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
933 
934 #define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION)
935 
936 
938 #define U_GC_L_MASK \
939  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
940 
942 #define U_GC_LC_MASK \
943  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
944 
946 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
947 
949 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
950 
952 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
953 
955 #define U_GC_C_MASK \
956  (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
957 
959 #define U_GC_P_MASK \
960  (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
961  U_GC_PI_MASK|U_GC_PF_MASK)
962 
964 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
965 
970 typedef enum UCharDirection {
971  /*
972  * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
973  * It matches pairs of lines like
974  * / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
975  * U_<[A-Z_]+> = <integer>,
976  */
977 
1024 #ifndef U_HIDE_DEPRECATED_API
1025 
1032 #endif // U_HIDE_DEPRECATED_API
1033 } UCharDirection;
1034 
1042  /*
1043  * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
1044  * It matches lines like
1045  * U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
1046  */
1047 
1054 #ifndef U_HIDE_DEPRECATED_API
1055 
1061  U_BPT_COUNT /* 3 */
1062 #endif // U_HIDE_DEPRECATED_API
1064 
1070  /*
1071  * Note: UBlockCode constants are parsed by preparseucd.py.
1072  * It matches lines like
1073  * UBLOCK_<Unicode Block value name> = <integer>,
1074  */
1075 
1077  UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
1078 
1080  UBLOCK_BASIC_LATIN = 1, /*[0000]*/
1081 
1084 
1087 
1090 
1092  UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
1093 
1096 
1099 
1104  UBLOCK_GREEK =8, /*[0370]*/
1105 
1107  UBLOCK_CYRILLIC =9, /*[0400]*/
1108 
1110  UBLOCK_ARMENIAN =10, /*[0530]*/
1111 
1113  UBLOCK_HEBREW =11, /*[0590]*/
1114 
1116  UBLOCK_ARABIC =12, /*[0600]*/
1117 
1119  UBLOCK_SYRIAC =13, /*[0700]*/
1120 
1122  UBLOCK_THAANA =14, /*[0780]*/
1123 
1125  UBLOCK_DEVANAGARI =15, /*[0900]*/
1126 
1128  UBLOCK_BENGALI =16, /*[0980]*/
1129 
1131  UBLOCK_GURMUKHI =17, /*[0A00]*/
1132 
1134  UBLOCK_GUJARATI =18, /*[0A80]*/
1135 
1137  UBLOCK_ORIYA =19, /*[0B00]*/
1138 
1140  UBLOCK_TAMIL =20, /*[0B80]*/
1141 
1143  UBLOCK_TELUGU =21, /*[0C00]*/
1144 
1146  UBLOCK_KANNADA =22, /*[0C80]*/
1147 
1149  UBLOCK_MALAYALAM =23, /*[0D00]*/
1150 
1152  UBLOCK_SINHALA =24, /*[0D80]*/
1153 
1155  UBLOCK_THAI =25, /*[0E00]*/
1156 
1158  UBLOCK_LAO =26, /*[0E80]*/
1159 
1161  UBLOCK_TIBETAN =27, /*[0F00]*/
1162 
1164  UBLOCK_MYANMAR =28, /*[1000]*/
1165 
1167  UBLOCK_GEORGIAN =29, /*[10A0]*/
1168 
1170  UBLOCK_HANGUL_JAMO =30, /*[1100]*/
1171 
1173  UBLOCK_ETHIOPIC =31, /*[1200]*/
1174 
1176  UBLOCK_CHEROKEE =32, /*[13A0]*/
1177 
1180 
1182  UBLOCK_OGHAM =34, /*[1680]*/
1183 
1185  UBLOCK_RUNIC =35, /*[16A0]*/
1186 
1188  UBLOCK_KHMER =36, /*[1780]*/
1189 
1191  UBLOCK_MONGOLIAN =37, /*[1800]*/
1192 
1195 
1197  UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
1198 
1201 
1204 
1206  UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
1207 
1213 
1216 
1218  UBLOCK_NUMBER_FORMS =45, /*[2150]*/
1219 
1221  UBLOCK_ARROWS =46, /*[2190]*/
1222 
1225 
1228 
1230  UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
1231 
1234 
1237 
1239  UBLOCK_BOX_DRAWING =52, /*[2500]*/
1240 
1242  UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
1243 
1245  UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
1246 
1249 
1251  UBLOCK_DINGBATS =56, /*[2700]*/
1252 
1254  UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
1255 
1258 
1260  UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
1261 
1264 
1267 
1269  UBLOCK_HIRAGANA =62, /*[3040]*/
1270 
1272  UBLOCK_KATAKANA =63, /*[30A0]*/
1273 
1275  UBLOCK_BOPOMOFO =64, /*[3100]*/
1276 
1279 
1281  UBLOCK_KANBUN =66, /*[3190]*/
1282 
1285 
1288 
1291 
1294 
1297 
1299  UBLOCK_YI_SYLLABLES =72, /*[A000]*/
1300 
1302  UBLOCK_YI_RADICALS =73, /*[A490]*/
1303 
1305  UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
1306 
1308  UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
1309 
1312 
1314  UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
1315 
1325  UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
1336 
1339 
1342 
1345 
1348 
1351 
1354 
1357 
1359  UBLOCK_SPECIALS =86, /*[FFF0]*/
1360 
1363 
1364  /* New blocks in Unicode 3.1 */
1365 
1367  UBLOCK_OLD_ITALIC = 88, /*[10300]*/
1369  UBLOCK_GOTHIC = 89, /*[10330]*/
1371  UBLOCK_DESERET = 90, /*[10400]*/
1375  UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
1383  UBLOCK_TAGS = 96, /*[E0000]*/
1384 
1385  /* New blocks in Unicode 3.2 */
1386 
1395  UBLOCK_TAGALOG = 98, /*[1700]*/
1397  UBLOCK_HANUNOO = 99, /*[1720]*/
1399  UBLOCK_BUHID = 100, /*[1740]*/
1401  UBLOCK_TAGBANWA = 101, /*[1760]*/
1415  UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
1420 
1421  /* New blocks in Unicode 4 */
1422 
1424  UBLOCK_LIMBU = 111, /*[1900]*/
1426  UBLOCK_TAI_LE = 112, /*[1950]*/
1428  UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
1430  UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
1436  UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
1438  UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
1440  UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
1442  UBLOCK_UGARITIC = 120, /*[10380]*/
1444  UBLOCK_SHAVIAN = 121, /*[10450]*/
1446  UBLOCK_OSMANYA = 122, /*[10480]*/
1448  UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
1450  UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
1453 
1454  /* New blocks in Unicode 4.1 */
1455 
1459  UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
1461  UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
1463  UBLOCK_BUGINESE = 129, /*[1A00]*/
1465  UBLOCK_CJK_STROKES = 130, /*[31C0]*/
1469  UBLOCK_COPTIC = 132, /*[2C80]*/
1471  UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
1473  UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
1475  UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
1477  UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
1479  UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
1483  UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
1485  UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
1491  UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
1493  UBLOCK_TIFINAGH = 144, /*[2D30]*/
1495  UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
1496 
1497  /* New blocks in Unicode 5.0 */
1498 
1500  UBLOCK_NKO = 146, /*[07C0]*/
1502  UBLOCK_BALINESE = 147, /*[1B00]*/
1504  UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
1506  UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
1508  UBLOCK_PHAGS_PA = 150, /*[A840]*/
1510  UBLOCK_PHOENICIAN = 151, /*[10900]*/
1512  UBLOCK_CUNEIFORM = 152, /*[12000]*/
1516  UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
1517 
1518  /* New blocks in Unicode 5.1 */
1519 
1521  UBLOCK_SUNDANESE = 155, /*[1B80]*/
1523  UBLOCK_LEPCHA = 156, /*[1C00]*/
1525  UBLOCK_OL_CHIKI = 157, /*[1C50]*/
1527  UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
1529  UBLOCK_VAI = 159, /*[A500]*/
1531  UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
1533  UBLOCK_SAURASHTRA = 161, /*[A880]*/
1535  UBLOCK_KAYAH_LI = 162, /*[A900]*/
1537  UBLOCK_REJANG = 163, /*[A930]*/
1539  UBLOCK_CHAM = 164, /*[AA00]*/
1541  UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
1543  UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
1545  UBLOCK_LYCIAN = 167, /*[10280]*/
1547  UBLOCK_CARIAN = 168, /*[102A0]*/
1549  UBLOCK_LYDIAN = 169, /*[10920]*/
1551  UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
1553  UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
1554 
1555  /* New blocks in Unicode 5.2 */
1556 
1558  UBLOCK_SAMARITAN = 172, /*[0800]*/
1562  UBLOCK_TAI_THAM = 174, /*[1A20]*/
1564  UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
1566  UBLOCK_LISU = 176, /*[A4D0]*/
1568  UBLOCK_BAMUM = 177, /*[A6A0]*/
1572  UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
1576  UBLOCK_JAVANESE = 181, /*[A980]*/
1578  UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
1580  UBLOCK_TAI_VIET = 183, /*[AA80]*/
1582  UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
1586  UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
1588  UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
1590  UBLOCK_AVESTAN = 188, /*[10B00]*/
1594  UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
1596  UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
1598  UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
1600  UBLOCK_KAITHI = 193, /*[11080]*/
1602  UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
1609 
1610  /* New blocks in Unicode 6.0 */
1611 
1613  UBLOCK_MANDAIC = 198, /*[0840]*/
1615  UBLOCK_BATAK = 199, /*[1BC0]*/
1617  UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
1619  UBLOCK_BRAHMI = 201, /*[11000]*/
1621  UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
1623  UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
1625  UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
1629  UBLOCK_EMOTICONS = 206, /*[1F600]*/
1633  UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
1636 
1637  /* New blocks in Unicode 6.1 */
1638 
1640  UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
1644  UBLOCK_CHAKMA = 212, /*[11100]*/
1648  UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
1650  UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
1652  UBLOCK_MIAO = 216, /*[16F00]*/
1654  UBLOCK_SHARADA = 217, /*[11180]*/
1656  UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
1660  UBLOCK_TAKRI = 220, /*[11680]*/
1661 
1662  /* New blocks in Unicode 7.0 */
1663 
1665  UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
1667  UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
1669  UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
1673  UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
1675  UBLOCK_ELBASAN = 226, /*[10500]*/
1679  UBLOCK_GRANTHA = 228, /*[11300]*/
1681  UBLOCK_KHOJKI = 229, /*[11200]*/
1683  UBLOCK_KHUDAWADI = 230, /*[112B0]*/
1685  UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
1687  UBLOCK_LINEAR_A = 232, /*[10600]*/
1689  UBLOCK_MAHAJANI = 233, /*[11150]*/
1691  UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
1693  UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
1695  UBLOCK_MODI = 236, /*[11600]*/
1697  UBLOCK_MRO = 237, /*[16A40]*/
1699  UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
1701  UBLOCK_NABATAEAN = 239, /*[10880]*/
1703  UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
1705  UBLOCK_OLD_PERMIC = 241, /*[10350]*/
1707  UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
1709  UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
1711  UBLOCK_PALMYRENE = 244, /*[10860]*/
1713  UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
1715  UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
1719  UBLOCK_SIDDHAM = 248, /*[11580]*/
1723  UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
1725  UBLOCK_TIRHUTA = 251, /*[11480]*/
1727  UBLOCK_WARANG_CITI = 252, /*[118A0]*/
1728 
1729  /* New blocks in Unicode 8.0 */
1730 
1732  UBLOCK_AHOM = 253, /*[11700]*/
1734  UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
1736  UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
1742  UBLOCK_HATRAN = 258, /*[108E0]*/
1744  UBLOCK_MULTANI = 259, /*[11280]*/
1746  UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
1750  UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/
1751 
1752  /* New blocks in Unicode 9.0 */
1753 
1755  UBLOCK_ADLAM = 263, /*[1E900]*/
1757  UBLOCK_BHAIKSUKI = 264, /*[11C00]*/
1759  UBLOCK_CYRILLIC_EXTENDED_C = 265, /*[1C80]*/
1761  UBLOCK_GLAGOLITIC_SUPPLEMENT = 266, /*[1E000]*/
1765  UBLOCK_MARCHEN = 268, /*[11C70]*/
1767  UBLOCK_MONGOLIAN_SUPPLEMENT = 269, /*[11660]*/
1769  UBLOCK_NEWA = 270, /*[11400]*/
1771  UBLOCK_OSAGE = 271, /*[104B0]*/
1773  UBLOCK_TANGUT = 272, /*[17000]*/
1775  UBLOCK_TANGUT_COMPONENTS = 273, /*[18800]*/
1776 
1777  // New blocks in Unicode 10.0
1778 
1782  UBLOCK_KANA_EXTENDED_A = 275, /*[1B100]*/
1784  UBLOCK_MASARAM_GONDI = 276, /*[11D00]*/
1786  UBLOCK_NUSHU = 277, /*[1B170]*/
1788  UBLOCK_SOYOMBO = 278, /*[11A50]*/
1790  UBLOCK_SYRIAC_SUPPLEMENT = 279, /*[0860]*/
1792  UBLOCK_ZANABAZAR_SQUARE = 280, /*[11A00]*/
1793 
1794  // New blocks in Unicode 11.0
1795 
1797  UBLOCK_CHESS_SYMBOLS = 281, /*[1FA00]*/
1799  UBLOCK_DOGRA = 282, /*[11800]*/
1801  UBLOCK_GEORGIAN_EXTENDED = 283, /*[1C90]*/
1803  UBLOCK_GUNJALA_GONDI = 284, /*[11D60]*/
1805  UBLOCK_HANIFI_ROHINGYA = 285, /*[10D00]*/
1807  UBLOCK_INDIC_SIYAQ_NUMBERS = 286, /*[1EC70]*/
1809  UBLOCK_MAKASAR = 287, /*[11EE0]*/
1811  UBLOCK_MAYAN_NUMERALS = 288, /*[1D2E0]*/
1813  UBLOCK_MEDEFAIDRIN = 289, /*[16E40]*/
1815  UBLOCK_OLD_SOGDIAN = 290, /*[10F00]*/
1817  UBLOCK_SOGDIAN = 291, /*[10F30]*/
1818 
1819  // New blocks in Unicode 12.0
1820 
1824  UBLOCK_ELYMAIC = 293, /*[10FE0]*/
1826  UBLOCK_NANDINAGARI = 294, /*[119A0]*/
1830  UBLOCK_OTTOMAN_SIYAQ_NUMBERS = 296, /*[1ED00]*/
1832  UBLOCK_SMALL_KANA_EXTENSION = 297, /*[1B130]*/
1836  UBLOCK_TAMIL_SUPPLEMENT = 299, /*[11FC0]*/
1838  UBLOCK_WANCHO = 300, /*[1E2C0]*/
1839 
1840  // New blocks in Unicode 13.0
1841 
1843  UBLOCK_CHORASMIAN = 301, /*[10FB0]*/
1847  UBLOCK_DIVES_AKURU = 303, /*[11900]*/
1849  UBLOCK_KHITAN_SMALL_SCRIPT = 304, /*[18B00]*/
1851  UBLOCK_LISU_SUPPLEMENT = 305, /*[11FB0]*/
1855  UBLOCK_TANGUT_SUPPLEMENT = 307, /*[18D00]*/
1857  UBLOCK_YEZIDI = 308, /*[10E80]*/
1858 
1859  // New blocks in Unicode 14.0
1860 
1862  UBLOCK_ARABIC_EXTENDED_B = 309, /*[0870]*/
1864  UBLOCK_CYPRO_MINOAN = 310, /*[12F90]*/
1866  UBLOCK_ETHIOPIC_EXTENDED_B = 311, /*[1E7E0]*/
1868  UBLOCK_KANA_EXTENDED_B = 312, /*[1AFF0]*/
1870  UBLOCK_LATIN_EXTENDED_F = 313, /*[10780]*/
1872  UBLOCK_LATIN_EXTENDED_G = 314, /*[1DF00]*/
1874  UBLOCK_OLD_UYGHUR = 315, /*[10F70]*/
1876  UBLOCK_TANGSA = 316, /*[16A70]*/
1878  UBLOCK_TOTO = 317, /*[1E290]*/
1882  UBLOCK_VITHKUQI = 319, /*[10570]*/
1885 
1886  // New blocks in Unicode 15.0
1887 
1889  UBLOCK_ARABIC_EXTENDED_C = 321, /*[10EC0]*/
1893  UBLOCK_CYRILLIC_EXTENDED_D = 323, /*[1E030]*/
1895  UBLOCK_DEVANAGARI_EXTENDED_A = 324, /*[11B00]*/
1897  UBLOCK_KAKTOVIK_NUMERALS = 325, /*[1D2C0]*/
1899  UBLOCK_KAWI = 326, /*[11F00]*/
1901  UBLOCK_NAG_MUNDARI = 327, /*[1E4D0]*/
1902 
1903 #ifndef U_HIDE_DEPRECATED_API
1904 
1911 #endif // U_HIDE_DEPRECATED_API
1912 
1915 };
1916 
1918 typedef enum UBlockCode UBlockCode;
1919 
1927 typedef enum UEastAsianWidth {
1928  /*
1929  * Note: UEastAsianWidth constants are parsed by preparseucd.py.
1930  * It matches lines like
1931  * U_EA_<Unicode East_Asian_Width value name>
1932  */
1933 
1934  U_EA_NEUTRAL, /*[N]*/
1935  U_EA_AMBIGUOUS, /*[A]*/
1936  U_EA_HALFWIDTH, /*[H]*/
1937  U_EA_FULLWIDTH, /*[F]*/
1938  U_EA_NARROW, /*[Na]*/
1939  U_EA_WIDE, /*[W]*/
1940 #ifndef U_HIDE_DEPRECATED_API
1941 
1948 #endif // U_HIDE_DEPRECATED_API
1949 } UEastAsianWidth;
1950 
1962 typedef enum UCharNameChoice {
1965 #ifndef U_HIDE_DEPRECATED_API
1966 
1972 #endif /* U_HIDE_DEPRECATED_API */
1973 
1977 #ifndef U_HIDE_DEPRECATED_API
1978 
1983 #endif // U_HIDE_DEPRECATED_API
1984 } UCharNameChoice;
1985 
1999 typedef enum UPropertyNameChoice {
2000  U_SHORT_PROPERTY_NAME,
2001  U_LONG_PROPERTY_NAME,
2002 #ifndef U_HIDE_DEPRECATED_API
2003 
2008 #endif // U_HIDE_DEPRECATED_API
2010 
2017 typedef enum UDecompositionType {
2018  /*
2019  * Note: UDecompositionType constants are parsed by preparseucd.py.
2020  * It matches lines like
2021  * U_DT_<Unicode Decomposition_Type value name>
2022  */
2023 
2024  U_DT_NONE, /*[none]*/
2025  U_DT_CANONICAL, /*[can]*/
2026  U_DT_COMPAT, /*[com]*/
2027  U_DT_CIRCLE, /*[enc]*/
2028  U_DT_FINAL, /*[fin]*/
2029  U_DT_FONT, /*[font]*/
2030  U_DT_FRACTION, /*[fra]*/
2031  U_DT_INITIAL, /*[init]*/
2032  U_DT_ISOLATED, /*[iso]*/
2033  U_DT_MEDIAL, /*[med]*/
2034  U_DT_NARROW, /*[nar]*/
2035  U_DT_NOBREAK, /*[nb]*/
2036  U_DT_SMALL, /*[sml]*/
2037  U_DT_SQUARE, /*[sqr]*/
2038  U_DT_SUB, /*[sub]*/
2039  U_DT_SUPER, /*[sup]*/
2040  U_DT_VERTICAL, /*[vert]*/
2041  U_DT_WIDE, /*[wide]*/
2042 #ifndef U_HIDE_DEPRECATED_API
2043 
2049  U_DT_COUNT /* 18 */
2050 #endif // U_HIDE_DEPRECATED_API
2052 
2059 typedef enum UJoiningType {
2060  /*
2061  * Note: UJoiningType constants are parsed by preparseucd.py.
2062  * It matches lines like
2063  * U_JT_<Unicode Joining_Type value name>
2064  */
2065 
2066  U_JT_NON_JOINING, /*[U]*/
2067  U_JT_JOIN_CAUSING, /*[C]*/
2068  U_JT_DUAL_JOINING, /*[D]*/
2069  U_JT_LEFT_JOINING, /*[L]*/
2070  U_JT_RIGHT_JOINING, /*[R]*/
2071  U_JT_TRANSPARENT, /*[T]*/
2072 #ifndef U_HIDE_DEPRECATED_API
2073 
2079  U_JT_COUNT /* 6 */
2080 #endif // U_HIDE_DEPRECATED_API
2081 } UJoiningType;
2082 
2089 typedef enum UJoiningGroup {
2090  /*
2091  * Note: UJoiningGroup constants are parsed by preparseucd.py.
2092  * It matches lines like
2093  * U_JG_<Unicode Joining_Group value name>
2094  */
2095 
2096  U_JG_NO_JOINING_GROUP,
2097  U_JG_AIN,
2098  U_JG_ALAPH,
2099  U_JG_ALEF,
2100  U_JG_BEH,
2101  U_JG_BETH,
2102  U_JG_DAL,
2103  U_JG_DALATH_RISH,
2104  U_JG_E,
2105  U_JG_FEH,
2106  U_JG_FINAL_SEMKATH,
2107  U_JG_GAF,
2108  U_JG_GAMAL,
2109  U_JG_HAH,
2111  U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
2112  U_JG_HE,
2113  U_JG_HEH,
2114  U_JG_HEH_GOAL,
2115  U_JG_HETH,
2116  U_JG_KAF,
2117  U_JG_KAPH,
2118  U_JG_KNOTTED_HEH,
2119  U_JG_LAM,
2120  U_JG_LAMADH,
2121  U_JG_MEEM,
2122  U_JG_MIM,
2123  U_JG_NOON,
2124  U_JG_NUN,
2125  U_JG_PE,
2126  U_JG_QAF,
2127  U_JG_QAPH,
2128  U_JG_REH,
2129  U_JG_REVERSED_PE,
2130  U_JG_SAD,
2131  U_JG_SADHE,
2132  U_JG_SEEN,
2133  U_JG_SEMKATH,
2134  U_JG_SHIN,
2135  U_JG_SWASH_KAF,
2136  U_JG_SYRIAC_WAW,
2137  U_JG_TAH,
2138  U_JG_TAW,
2139  U_JG_TEH_MARBUTA,
2140  U_JG_TETH,
2141  U_JG_WAW,
2142  U_JG_YEH,
2143  U_JG_YEH_BARREE,
2144  U_JG_YEH_WITH_TAIL,
2145  U_JG_YUDH,
2146  U_JG_YUDH_HE,
2147  U_JG_ZAIN,
2205 #ifndef U_HIDE_DEPRECATED_API
2206 
2213 #endif // U_HIDE_DEPRECATED_API
2214 } UJoiningGroup;
2215 
2223  /*
2224  * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2225  * It matches lines like
2226  * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2227  */
2228 
2229  U_GCB_OTHER = 0, /*[XX]*/
2230  U_GCB_CONTROL = 1, /*[CN]*/
2231  U_GCB_CR = 2, /*[CR]*/
2232  U_GCB_EXTEND = 3, /*[EX]*/
2233  U_GCB_L = 4, /*[L]*/
2234  U_GCB_LF = 5, /*[LF]*/
2235  U_GCB_LV = 6, /*[LV]*/
2236  U_GCB_LVT = 7, /*[LVT]*/
2237  U_GCB_T = 8, /*[T]*/
2238  U_GCB_V = 9, /*[V]*/
2240  U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2242  U_GCB_PREPEND = 11, /*[PP]*/
2244  U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2246  U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2248  U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2250  U_GCB_E_MODIFIER = 15, /*[EM]*/
2252  U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2254  U_GCB_ZWJ = 17, /*[ZWJ]*/
2255 
2256 #ifndef U_HIDE_DEPRECATED_API
2257 
2264 #endif // U_HIDE_DEPRECATED_API
2266 
2274 typedef enum UWordBreakValues {
2275  /*
2276  * Note: UWordBreakValues constants are parsed by preparseucd.py.
2277  * It matches lines like
2278  * U_WB_<Unicode Word_Break value name>
2279  */
2280 
2281  U_WB_OTHER = 0, /*[XX]*/
2282  U_WB_ALETTER = 1, /*[LE]*/
2283  U_WB_FORMAT = 2, /*[FO]*/
2284  U_WB_KATAKANA = 3, /*[KA]*/
2285  U_WB_MIDLETTER = 4, /*[ML]*/
2286  U_WB_MIDNUM = 5, /*[MN]*/
2287  U_WB_NUMERIC = 6, /*[NU]*/
2288  U_WB_EXTENDNUMLET = 7, /*[EX]*/
2290  U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2292  U_WB_EXTEND = 9, /*[Extend]*/
2294  U_WB_LF = 10, /*[LF]*/
2296  U_WB_MIDNUMLET =11, /*[MB]*/
2298  U_WB_NEWLINE =12, /*[NL]*/
2300  U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2302  U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2304  U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2306  U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2308  U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2310  U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2312  U_WB_E_MODIFIER = 19, /*[EM]*/
2314  U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2316  U_WB_ZWJ = 21, /*[ZWJ]*/
2318  U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/
2319 
2320 #ifndef U_HIDE_DEPRECATED_API
2321 
2328 #endif // U_HIDE_DEPRECATED_API
2330 
2337 typedef enum USentenceBreak {
2338  /*
2339  * Note: USentenceBreak constants are parsed by preparseucd.py.
2340  * It matches lines like
2341  * U_SB_<Unicode Sentence_Break value name>
2342  */
2343 
2344  U_SB_OTHER = 0, /*[XX]*/
2345  U_SB_ATERM = 1, /*[AT]*/
2346  U_SB_CLOSE = 2, /*[CL]*/
2347  U_SB_FORMAT = 3, /*[FO]*/
2348  U_SB_LOWER = 4, /*[LO]*/
2349  U_SB_NUMERIC = 5, /*[NU]*/
2350  U_SB_OLETTER = 6, /*[LE]*/
2351  U_SB_SEP = 7, /*[SE]*/
2352  U_SB_SP = 8, /*[SP]*/
2353  U_SB_STERM = 9, /*[ST]*/
2354  U_SB_UPPER = 10, /*[UP]*/
2355  U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2356  U_SB_EXTEND = 12, /*[EX]*/
2357  U_SB_LF = 13, /*[LF]*/
2358  U_SB_SCONTINUE = 14, /*[SC]*/
2359 #ifndef U_HIDE_DEPRECATED_API
2360 
2367 #endif // U_HIDE_DEPRECATED_API
2368 } USentenceBreak;
2369 
2376 typedef enum ULineBreak {
2377  /*
2378  * Note: ULineBreak constants are parsed by preparseucd.py.
2379  * It matches lines like
2380  * U_LB_<Unicode Line_Break value name>
2381  */
2382 
2383  U_LB_UNKNOWN = 0, /*[XX]*/
2384  U_LB_AMBIGUOUS = 1, /*[AI]*/
2385  U_LB_ALPHABETIC = 2, /*[AL]*/
2386  U_LB_BREAK_BOTH = 3, /*[B2]*/
2387  U_LB_BREAK_AFTER = 4, /*[BA]*/
2388  U_LB_BREAK_BEFORE = 5, /*[BB]*/
2389  U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2390  U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2391  U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2392  U_LB_COMBINING_MARK = 9, /*[CM]*/
2393  U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2394  U_LB_EXCLAMATION = 11, /*[EX]*/
2395  U_LB_GLUE = 12, /*[GL]*/
2396  U_LB_HYPHEN = 13, /*[HY]*/
2397  U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2399  U_LB_INSEPARABLE = 15, /*[IN]*/
2400  U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2401  U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2402  U_LB_LINE_FEED = 17, /*[LF]*/
2403  U_LB_NONSTARTER = 18, /*[NS]*/
2404  U_LB_NUMERIC = 19, /*[NU]*/
2405  U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2406  U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2407  U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2408  U_LB_QUOTATION = 23, /*[QU]*/
2409  U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2410  U_LB_SURROGATE = 25, /*[SG]*/
2411  U_LB_SPACE = 26, /*[SP]*/
2412  U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2413  U_LB_ZWSPACE = 28, /*[ZW]*/
2415  U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2417  U_LB_WORD_JOINER = 30, /*[WJ]*/
2419  U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2421  U_LB_H3 = 32, /*[H3]*/
2423  U_LB_JL = 33, /*[JL]*/
2425  U_LB_JT = 34, /*[JT]*/
2427  U_LB_JV = 35, /*[JV]*/
2429  U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2431  U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2433  U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2435  U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2437  U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2439  U_LB_E_MODIFIER = 41, /*[EM]*/
2441  U_LB_ZWJ = 42, /*[ZWJ]*/
2442 #ifndef U_HIDE_DEPRECATED_API
2443 
2450 #endif // U_HIDE_DEPRECATED_API
2451 } ULineBreak;
2452 
2459 typedef enum UNumericType {
2460  /*
2461  * Note: UNumericType constants are parsed by preparseucd.py.
2462  * It matches lines like
2463  * U_NT_<Unicode Numeric_Type value name>
2464  */
2465 
2466  U_NT_NONE, /*[None]*/
2467  U_NT_DECIMAL, /*[de]*/
2468  U_NT_DIGIT, /*[di]*/
2469  U_NT_NUMERIC, /*[nu]*/
2470 #ifndef U_HIDE_DEPRECATED_API
2471 
2478 #endif // U_HIDE_DEPRECATED_API
2479 } UNumericType;
2480 
2487 typedef enum UHangulSyllableType {
2488  /*
2489  * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2490  * It matches lines like
2491  * U_HST_<Unicode Hangul_Syllable_Type value name>
2492  */
2493 
2494  U_HST_NOT_APPLICABLE, /*[NA]*/
2495  U_HST_LEADING_JAMO, /*[L]*/
2496  U_HST_VOWEL_JAMO, /*[V]*/
2497  U_HST_TRAILING_JAMO, /*[T]*/
2498  U_HST_LV_SYLLABLE, /*[LV]*/
2499  U_HST_LVT_SYLLABLE, /*[LVT]*/
2500 #ifndef U_HIDE_DEPRECATED_API
2501 
2508 #endif // U_HIDE_DEPRECATED_API
2510 
2518  /*
2519  * Note: UIndicPositionalCategory constants are parsed by preparseucd.py.
2520  * It matches lines like
2521  * U_INPC_<Unicode Indic_Positional_Category value name>
2522  */
2523 
2557 
2565  /*
2566  * Note: UIndicSyllabicCategory constants are parsed by preparseucd.py.
2567  * It matches lines like
2568  * U_INSC_<Unicode Indic_Syllabic_Category value name>
2569  */
2570 
2644 
2651 typedef enum UVerticalOrientation {
2652  /*
2653  * Note: UVerticalOrientation constants are parsed by preparseucd.py.
2654  * It matches lines like
2655  * U_VO_<Unicode Vertical_Orientation value name>
2656  */
2657 
2667 
2695 U_CAPI UBool U_EXPORT2
2697 
2722 U_CAPI UBool U_EXPORT2
2723 u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which);
2724 
2740 U_CAPI const USet * U_EXPORT2
2741 u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode);
2742 
2755 U_CAPI UBool U_EXPORT2
2757 
2770 U_CAPI UBool U_EXPORT2
2772 
2785 U_CAPI UBool U_EXPORT2
2787 
2806 U_CAPI UBool U_EXPORT2
2808 
2847 U_CAPI int32_t U_EXPORT2
2849 
2868 U_CAPI int32_t U_EXPORT2
2870 
2897 U_CAPI int32_t U_EXPORT2
2899 
2915 U_CAPI const UCPMap * U_EXPORT2
2916 u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode);
2917 
2940 U_CAPI double U_EXPORT2
2942 
2950 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
2951 
2975 U_CAPI UBool U_EXPORT2
2976 u_islower(UChar32 c);
2977 
3002 U_CAPI UBool U_EXPORT2
3003 u_isupper(UChar32 c);
3004 
3019 U_CAPI UBool U_EXPORT2
3020 u_istitle(UChar32 c);
3021 
3040 U_CAPI UBool U_EXPORT2
3041 u_isdigit(UChar32 c);
3042 
3061 U_CAPI UBool U_EXPORT2
3062 u_isalpha(UChar32 c);
3063 
3082 U_CAPI UBool U_EXPORT2
3083 u_isalnum(UChar32 c);
3084 
3105 U_CAPI UBool U_EXPORT2
3106 u_isxdigit(UChar32 c);
3107 
3121 U_CAPI UBool U_EXPORT2
3122 u_ispunct(UChar32 c);
3123 
3140 U_CAPI UBool U_EXPORT2
3141 u_isgraph(UChar32 c);
3142 
3168 U_CAPI UBool U_EXPORT2
3169 u_isblank(UChar32 c);
3170 
3193 U_CAPI UBool U_EXPORT2
3194 u_isdefined(UChar32 c);
3195 
3214 U_CAPI UBool U_EXPORT2
3215 u_isspace(UChar32 c);
3216 
3235 U_CAPI UBool U_EXPORT2
3237 
3275 U_CAPI UBool U_EXPORT2
3277 
3299 U_CAPI UBool U_EXPORT2
3300 u_iscntrl(UChar32 c);
3301 
3314 U_CAPI UBool U_EXPORT2
3316 
3332 U_CAPI UBool U_EXPORT2
3333 u_isprint(UChar32 c);
3334 
3352 U_CAPI UBool U_EXPORT2
3353 u_isbase(UChar32 c);
3354 
3371 U_CAPI UCharDirection U_EXPORT2
3373 
3389 U_CAPI UBool U_EXPORT2
3391 
3411 U_CAPI UChar32 U_EXPORT2
3413 
3430 U_CAPI UChar32 U_EXPORT2
3432 
3444 U_CAPI int8_t U_EXPORT2
3445 u_charType(UChar32 c);
3446 
3460 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3461 
3479 typedef UBool U_CALLCONV
3480 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3481 
3501 U_CAPI void U_EXPORT2
3502 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3503 
3504 #if !UCONFIG_NO_NORMALIZATION
3505 
3513 U_CAPI uint8_t U_EXPORT2
3515 
3516 #endif
3517 
3541 U_CAPI int32_t U_EXPORT2
3543 
3553 U_CAPI UBlockCode U_EXPORT2
3555 
3588 U_CAPI int32_t U_EXPORT2
3589 u_charName(UChar32 code, UCharNameChoice nameChoice,
3590  char *buffer, int32_t bufferLength,
3591  UErrorCode *pErrorCode);
3592 
3593 #ifndef U_HIDE_DEPRECATED_API
3594 
3612 U_DEPRECATED int32_t U_EXPORT2
3614  char *dest, int32_t destCapacity,
3615  UErrorCode *pErrorCode);
3616 #endif /* U_HIDE_DEPRECATED_API */
3617 
3638 U_CAPI UChar32 U_EXPORT2
3639 u_charFromName(UCharNameChoice nameChoice,
3640  const char *name,
3641  UErrorCode *pErrorCode);
3642 
3660 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3661  UChar32 code,
3662  UCharNameChoice nameChoice,
3663  const char *name,
3664  int32_t length);
3665 
3687 U_CAPI void U_EXPORT2
3688 u_enumCharNames(UChar32 start, UChar32 limit,
3689  UEnumCharNamesFn *fn,
3690  void *context,
3691  UCharNameChoice nameChoice,
3692  UErrorCode *pErrorCode);
3693 
3725 U_CAPI const char* U_EXPORT2
3726 u_getPropertyName(UProperty property,
3727  UPropertyNameChoice nameChoice);
3728 
3748 U_CAPI UProperty U_EXPORT2
3749 u_getPropertyEnum(const char* alias);
3750 
3798 U_CAPI const char* U_EXPORT2
3800  int32_t value,
3801  UPropertyNameChoice nameChoice);
3802 
3834 U_CAPI int32_t U_EXPORT2
3836  const char* alias);
3837 
3852 U_CAPI UBool U_EXPORT2
3853 u_isIDStart(UChar32 c);
3854 
3869 U_CAPI UBool U_EXPORT2
3870 u_isIDPart(UChar32 c);
3871 
3892 U_CAPI UBool U_EXPORT2
3894 
3911 U_CAPI UBool U_EXPORT2
3913 
3932 U_CAPI UBool U_EXPORT2
3934 
3957 U_CAPI UChar32 U_EXPORT2
3958 u_tolower(UChar32 c);
3959 
3982 U_CAPI UChar32 U_EXPORT2
3983 u_toupper(UChar32 c);
3984 
4007 U_CAPI UChar32 U_EXPORT2
4008 u_totitle(UChar32 c);
4009 
4032 U_CAPI UChar32 U_EXPORT2
4033 u_foldCase(UChar32 c, uint32_t options);
4034 
4073 U_CAPI int32_t U_EXPORT2
4074 u_digit(UChar32 ch, int8_t radix);
4075 
4104 U_CAPI UChar32 U_EXPORT2
4105 u_forDigit(int32_t digit, int8_t radix);
4106 
4121 U_CAPI void U_EXPORT2
4122 u_charAge(UChar32 c, UVersionInfo versionArray);
4123 
4135 U_CAPI void U_EXPORT2
4136 u_getUnicodeVersion(UVersionInfo versionArray);
4137 
4138 #if !UCONFIG_NO_NORMALIZATION
4139 
4160 U_CAPI int32_t U_EXPORT2
4161 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
4162 
4163 #endif
4164 
4165 
4167 
4168 #endif /*_UCHAR*/
4169 /*eof*/
uint8_t UVersionInfo[U_MAX_VERSION_LENGTH]
The binary form of a version on ICU APIs is an array of 4 uint8_t.
Definition: uversion.h:59
Binary property Ideographic.
Definition: uchar.h:271
Unicode 4.0.1 renames the "Cyrillic Supplementary" block to "Cyrillic Supplement".
Definition: uchar.h:1393
Enumerated property Indic_Positional_Category.
Definition: uchar.h:635
Binary property Changes_When_Lowercased.
Definition: uchar.h:423
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition: uchar.h:275
U_CAPI UBool u_isalnum(UChar32 c)
Determines whether the specified code point is an alphanumeric character (letter or digit) according ...
Binary property Case_Ignorable.
Definition: uchar.h:421
Enumerated property NFC_Quick_Check.
Definition: uchar.h:590
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:2212
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:753
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1335
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:547
Binary property XID_Start.
Definition: uchar.h:328
Binary property Join_Control.
Definition: uchar.h:282
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition: uchar.h:286
struct UCPMap UCPMap
Abstract map from Unicode code points (U+0000..U+10FFFF) to integer values.
Definition: ucpmap.h:30
Binary property White_Space.
Definition: uchar.h:321
String property Titlecase_Mapping.
Definition: uchar.h:726
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 "blank" or "horizontal space", a character that visi...
One more than the highest normal UPropertyNameChoice value.
Definition: uchar.h:2007
U_CAPI UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:655
Enumerated property Numeric_Type.
Definition: uchar.h:575
Binary property xdigit (a C/POSIX character class).
Definition: uchar.h:417
Binary property Alphabetic.
Definition: uchar.h:209
First constant for double Unicode properties.
Definition: uchar.h:681
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 "defined", 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:3660
Binary property of strings RGI_Emoji.
Definition: uchar.h:534
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:2366
Binary property Emoji_Component.
Definition: uchar.h:468
Binary property of strings Emoji_Keycap_Sequence.
Definition: uchar.h:499
Binary property Emoji.
Definition: uchar.h:440
Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNE...
Definition: uchar.h:785
U_CAPI UProperty u_getPropertyEnum(const char *alias)
Return the UProperty enum for a given property name, as specified in the Unicode database file Proper...
String property Simple_Case_Folding.
Definition: uchar.h:714
U_CAPI int32_t u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Get the FC_NFKC_Closure property string for a character.
Binary property NFC_Inert.
Definition: uchar.h:364
Binary property graph (a C/POSIX character class).
Definition: uchar.h:407
String property Bidi_Mirroring_Glyph.
Definition: uchar.h:697
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:674
U_CAPI UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
Enumerated property Block.
Definition: uchar.h:550
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:765
Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2399
Binary property Radical (new in Unicode 3.2).
Definition: uchar.h:301
UCharDirection
This specifies the language directional property of a character set.
Definition: uchar.h:970
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition: uchar.h:279
Binary property Grapheme_Link (new in Unicode 3.2).
Definition: uchar.h:253
Enumerated property Decomposition_Type.
Definition: uchar.h:556
Binary property Regional_Indicator.
Definition: uchar.h:473
U_CAPI UChar32 u_foldCase(UChar32 c, uint32_t options)
The given character is mapped to its case folding equivalent according to UnicodeData.txt and CaseFolding.txt; if the character has no case folding equivalent, the character itself is returned.
#define U_CALLCONV
Similar to U_CDECL_BEGIN/U_CDECL_END, this qualifier is necessary in callback function typedefs to ma...
Definition: platform.h:866
String property Case_Folding.
Definition: uchar.h:700
String property Name.
Definition: uchar.h:711
C API: This file defines an abstract map from Unicode code points to integer values.
String property Simple_Uppercase_Mapping.
Definition: uchar.h:723
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 "graphic" character (printable, excluding spaces)...
Enumerated property Bidi_Class.
Definition: uchar.h:545
Enumerated property General_Category.
Definition: uchar.h:563
Binary property Prepended_Concatenation_Mark.
Definition: uchar.h:478
U_CAPI UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
Sm.
Definition: uchar.h:833
U_CAPI void u_charAge(UChar32 c, UVersionInfo versionArray)
Get the "age" of the code point.
String property Unicode_1_Name.
Definition: uchar.h:732
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:2459
U_CAPI UChar32 u_charMirror(UChar32 c)
Maps the specified character to a "mirror-image" character.
One more than the highest normal UBidiPairedBracketType value.
Definition: uchar.h:1061
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:392
Enumerated property Vertical_Orientation.
Definition: uchar.h:649
Close paired bracket.
Definition: uchar.h:1053
UBool UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type)
Callback from u_enumCharTypes(), is called for each contiguous range of code points c (where start<=c...
Definition: uchar.h:3480
Binary property Changes_When_Casefolded.
Definition: uchar.h:429
U_CAPI void u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context)
Enumerate efficiently all code points with their Unicode general categories.
Binary property NFD_Inert.
Definition: uchar.h:350
Binary property Diacritic.
Definition: uchar.h:234
Binary property Terminal_Punctuation.
Definition: uchar.h:310
U_CAPI UBool u_isMirrored(UChar32 c)
Determines whether the code point has the Bidi_Mirrored property.
U_CAPI const UCPMap * u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode)
Returns an immutable UCPMap for an enumerated/catalog/int-valued property.
One more than the highest normal UNumericType value.
Definition: uchar.h:2477
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:584
U_CAPI int32_t u_digit(UChar32 ch, int8_t radix)
Returns the decimal digit value of the code point in the specified radix.
U_CAPI uint8_t u_getCombiningClass(UChar32 c)
Returns the combining class of the code point as specified in UnicodeData.txt.
#define U_CDECL_BEGIN
This is used to begin a declaration of a library private ICU C API.
Definition: umachine.h:85
Binary property STerm (new in Unicode 4.0.1).
Definition: uchar.h:337
Enumerated property Joining_Group.
Definition: uchar.h:566
Binary property ID_Continue.
Definition: uchar.h:264
Binary property Emoji_Presentation.
Definition: uchar.h:447
U_CAPI int32_t u_getPropertyValueEnum(UProperty property, const char *alias)
Return the property value integer for a given value name, as specified in the Unicode database file P...
Open paired bracket.
Definition: uchar.h:1051
Binary property blank (a C/POSIX character class).
Definition: uchar.h:402
Binary property Quotation_Mark.
Definition: uchar.h:297
#define U_DEPRECATED
This is used to declare a function as a deprecated public ICU C API.
Definition: umachine.h:116
Binary property Changes_When_NFKC_Casefolded.
Definition: uchar.h:433
U_CAPI UBlockCode ublock_getCode(UChar32 c)
Returns the Unicode allocation block that contains the character.
First constant for binary Unicode properties.
Definition: uchar.h:211
U_CAPI UBool u_isIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in an identifier accordin...
Binary property Noncharacter_Code_Point.
Definition: uchar.h:295
Binary property Hyphen.
Definition: uchar.h:259
U_CAPI UBool u_ispunct(UChar32 c)
Determines whether the specified code point is a punctuation character.
U_CAPI int32_t u_charName(UChar32 code, UCharNameChoice nameChoice, char *buffer, int32_t bufferLength, UErrorCode *pErrorCode)
Retrieve the name of a Unicode character.
Enumerated property East_Asian_Width.
Definition: uchar.h:560
ULineBreak
Line Break constants.
Definition: uchar.h:2376
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:1031
U_CAPI UChar32 u_tolower(UChar32 c)
The given character is mapped to its lowercase equivalent according to UnicodeData.txt; if the character has no lowercase equivalent, the character itself is returned.
Binary property Full_Composition_Exclusion.
Definition: uchar.h:242
Bitmask property General_Category_Mask.
Definition: uchar.h:666
String property Simple_Titlecase_Mapping.
Definition: uchar.h:720
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2263
Unicode 3.2 renames this block to "Combining Diacritical Marks for Symbols".
Definition: uchar.h:1212
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:2017
One more than the highest normal ULineBreak value.
Definition: uchar.h:2449
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2487
String property Simple_Lowercase_Mapping.
Definition: uchar.h:717
UIndicPositionalCategory
Indic Positional Category constants.
Definition: uchar.h:2517
Binary property print (a C/POSIX character class).
Definition: uchar.h:412
Binary property Case_Sensitive.
Definition: uchar.h:332
Standard or synthetic character name.
Definition: uchar.h:1974
Binary property Bidi_Mirrored.
Definition: uchar.h:222
UVerticalOrientation
Vertical Orientation constants.
Definition: uchar.h:2651
Binary property NFKC_Inert.
Definition: uchar.h:371
U_CAPI void u_enumCharNames(UChar32 start, UChar32 limit, UEnumCharNamesFn *fn, void *context, UCharNameChoice nameChoice, UErrorCode *pErrorCode)
Enumerate all assigned Unicode characters between the start and limit code points (start inclusive...
Binary property Changes_When_Casemapped.
Definition: uchar.h:431
First constant for string Unicode properties.
Definition: uchar.h:694
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition: uchar.h:250
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:435
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:2222
New No_Block value in Unicode 4.
Definition: uchar.h:1077
Binary property Extender.
Definition: uchar.h:238
Double property Numeric_Value.
Definition: uchar.h:679
Binary property Math.
Definition: uchar.h:291
Unicode character name (Name property).
Definition: uchar.h:1964
U_CAPI const char * u_getPropertyName(UProperty property, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property, as given in the Unicode database file PropertyAliases...
Binary property Grapheme_Base (new in Unicode 3.2).
Definition: uchar.h:246
Binary property NFKD_Inert.
Definition: uchar.h:357
U_CAPI UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier...
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:773
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:617
Binary property Lowercase.
Definition: uchar.h:289
First constant for bit-mask Unicode properties.
Definition: uchar.h:668
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:627
Binary property of strings Basic_Emoji.
Definition: uchar.h:492
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:2337
Binary property Extended_Pictographic.
Definition: uchar.h:485
Enumerated property Indic_Syllabic_Category.
Definition: uchar.h:642
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:314
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:553
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:1962
One more than the last constant for binary Unicode properties.
Definition: uchar.h:540
Enumerated property Script.
Definition: uchar.h:578
Unicode 3.2 renames this block to "Greek and Coptic".
Definition: uchar.h:1104
Binary property Hex_Digit.
Definition: uchar.h:256
String property Uppercase_Mapping.
Definition: uchar.h:736
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:1999
U_CAPI UBool u_isupper(UChar32 c)
Determines whether the specified code point has the general category "Lu" (uppercase letter)...
String property Lowercase_Mapping.
Definition: uchar.h:708
char16_t UChar
The base type for UTF-16 code units and pointers.
Definition: umachine.h:386
#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:593
Not a paired bracket.
Definition: uchar.h:1049
U_CAPI UBool u_stringHasBinaryProperty(const UChar *s, int32_t length, UProperty which)
Returns true if the property is true for the string.
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:195
U_CAPI const USet * u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode)
Returns a frozen USet for a binary property.
Enumerated property Hangul_Syllable_Type, new in Unicode 4.
Definition: uchar.h:581
Binary property Dash.
Definition: uchar.h:224
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:397
Cf.
Definition: uchar.h:817
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition: uchar.h:343
Binary property of strings RGI_Emoji_ZWJ_Sequence.
Definition: uchar.h:527
One more than the highest normal UHangulSyllableType value.
Definition: uchar.h:2507
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:1069
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:622
Binary property Deprecated (new in Unicode 3.2).
Definition: uchar.h:231
Binary property Bidi_Control.
Definition: uchar.h:217
Binary property XID_Continue.
Definition: uchar.h:325
Same as UBLOCK_PRIVATE_USE.
Definition: uchar.h:1325
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415
Binary property Uppercase.
Definition: uchar.h:317
Binary property Changes_When_Uppercased.
Definition: uchar.h:425
UJoiningGroup
Joining Group constants.
Definition: uchar.h:2089
Binary property Cased.
Definition: uchar.h:419
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition: uset.h:50
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:1947
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:2049
Cs.
Definition: uchar.h:821
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:1927
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:607
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:783
First constant for Unicode properties with unusual value types.
Definition: uchar.h:755
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:2274
Binary property of strings RGI_Emoji_Modifier_Sequence.
Definition: uchar.h:506
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:569
One more than the last constant for double Unicode properties.
Definition: uchar.h:687
Binary property Emoji_Modifier.
Definition: uchar.h:454
Basic definitions for ICU, for both C and C++ APIs.
U_CAPI UBool u_isWhitespace(UChar32 c)
Determines if the specified code point is a whitespace character according to Java/ICU.
U_CAPI UBool u_isJavaSpaceChar(UChar32 c)
Determine if the specified code point is a space character according to Java.
String property Bidi_Paired_Bracket (new in Unicode 6.3).
Definition: uchar.h:739
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:600
Binary property ASCII_Hex_Digit.
Definition: uchar.h:213
UJoiningType
Joining Type constants.
Definition: uchar.h:2059
Binary property Soft_Dotted (new in Unicode 3.2).
Definition: uchar.h:306
One more than the last constant for string Unicode properties.
Definition: uchar.h:745
Binary property Emoji_Modifier_Base.
Definition: uchar.h:461
Binary Property Segment_Starter.
Definition: uchar.h:382
U_CAPI UChar32 u_forDigit(int32_t digit, int8_t radix)
Determines the character representation for a specific digit in the specified radix.
Binary property ID_Start.
Definition: uchar.h:268
Corrected name from NameAliases.txt.
Definition: uchar.h:1976
Binary property Changes_When_Titlecased.
Definition: uchar.h:427
Enumerated property Line_Break.
Definition: uchar.h:572
One more than the highest normal UJoiningType value.
Definition: uchar.h:2079
U_CAPI UBool u_isalpha(UChar32 c)
Determines whether the specified code point is a letter character.
U_CAPI const char * u_getPropertyValueName(UProperty property, int32_t value, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property value, as given in the Unicode database file PropertyVal...
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:612
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:704
Binary property of strings RGI_Emoji_Flag_Sequence.
Definition: uchar.h:513
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 "Ll" (lowercase letter)...
Binary property Default_Ignorable_Code_Point (new in Unicode 3.2).
Definition: uchar.h:228
One more than the highest normal UBlockCode value.
Definition: uchar.h:1910
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:1971
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:851
Binary property of strings RGI_Emoji_Tag_Sequence.
Definition: uchar.h:520
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2327
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition: uchar.h:387
One more than the last constant for Unicode properties with unusual value types.
Definition: uchar.h:761
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:1041
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:1982
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition: uchar.h:2564
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:692
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:587
U_CAPI UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.