ICU 60.1  60.1
uchar.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 **********************************************************************
5 * Copyright (C) 1997-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
8 *
9 * File UCHAR.H
10 *
11 * Modification History:
12 *
13 * Date Name Description
14 * 04/02/97 aliu Creation.
15 * 03/29/99 helena Updated for C APIs.
16 * 4/15/99 Madhu Updated for C Implementation and Javadoc
17 * 5/20/99 Madhu Added the function u_getVersion()
18 * 8/19/1999 srl Upgraded scripts to Unicode 3.0
19 * 8/27/1999 schererm UCharDirection constants: U_...
20 * 11/11/1999 weiv added u_isalnum(), cleaned comments
21 * 01/11/2000 helena Renamed u_getVersion to u_getUnicodeVersion().
22 ******************************************************************************
23 */
24 
25 #ifndef UCHAR_H
26 #define UCHAR_H
27 
28 #include "unicode/utypes.h"
29 #include "unicode/stringoptions.h"
30 
32 
33 /*==========================================================================*/
34 /* Unicode version number */
35 /*==========================================================================*/
45 #define U_UNICODE_VERSION "10.0"
46 
127 #define UCHAR_MIN_VALUE 0
128 
137 #define UCHAR_MAX_VALUE 0x10ffff
138 
143 #define U_MASK(x) ((uint32_t)1<<(x))
144 
165 typedef enum UProperty {
166  /*
167  * Note: UProperty constants are parsed by preparseucd.py.
168  * It matches lines like
169  * UCHAR_<Unicode property name>=<integer>,
170  */
171 
172  /* Note: Place UCHAR_ALPHABETIC before UCHAR_BINARY_START so that
173  debuggers display UCHAR_ALPHABETIC as the symbolic name for 0,
174  rather than UCHAR_BINARY_START. Likewise for other *_START
175  identifiers. */
176 
449 #ifndef U_HIDE_DEPRECATED_API
450 
455 #endif // U_HIDE_DEPRECATED_API
456 
464  UCHAR_BLOCK=0x1001,
492  UCHAR_SCRIPT=0x100A,
542 #ifndef U_HIDE_DEPRECATED_API
543 
548 #endif // U_HIDE_DEPRECATED_API
549 
561 #ifndef U_HIDE_DEPRECATED_API
562 
567 #endif // U_HIDE_DEPRECATED_API
568 
574 #ifndef U_HIDE_DEPRECATED_API
575 
580 #endif // U_HIDE_DEPRECATED_API
581 
584  UCHAR_AGE=0x4000,
593 #ifndef U_HIDE_DEPRECATED_API
594 
597 #endif /* U_HIDE_DEPRECATED_API */
598 
603  UCHAR_NAME=0x4005,
619 #ifndef U_HIDE_DEPRECATED_API
620 
625 #endif /* U_HIDE_DEPRECATED_API */
626 
632 #ifndef U_HIDE_DEPRECATED_API
633 
638 #endif // U_HIDE_DEPRECATED_API
639 
648 #ifndef U_HIDE_DEPRECATED_API
649 
654 #endif // U_HIDE_DEPRECATED_API
655 
658 } UProperty;
659 
665 typedef enum UCharCategory
666 {
667  /*
668  * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
669  * It matches pairs of lines like
670  * / ** <Unicode 2-letter General_Category value> comment... * /
671  * U_<[A-Z_]+> = <integer>,
672  */
673 
744 } UCharCategory;
745 
760 #define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES)
761 
763 #define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER)
764 
765 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
766 
767 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
768 
769 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
770 
771 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
772 
774 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
775 
776 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
777 
778 #define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK)
779 
781 #define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER)
782 
783 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
784 
785 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
786 
788 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
789 
790 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
791 
792 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
793 
795 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
796 
797 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
798 
799 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
800 
801 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
802 
804 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
805 
806 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
807 
808 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
809 
810 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
811 
812 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
813 
815 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
816 
817 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
818 
819 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
820 
821 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
822 
824 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
825 
826 #define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION)
827 
828 
830 #define U_GC_L_MASK \
831  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
832 
834 #define U_GC_LC_MASK \
835  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
836 
838 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
839 
841 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
842 
844 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
845 
847 #define U_GC_C_MASK \
848  (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
849 
851 #define U_GC_P_MASK \
852  (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
853  U_GC_PI_MASK|U_GC_PF_MASK)
854 
856 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
857 
862 typedef enum UCharDirection {
863  /*
864  * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
865  * It matches pairs of lines like
866  * / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
867  * U_<[A-Z_]+> = <integer>,
868  */
869 
916 #ifndef U_HIDE_DEPRECATED_API
917 
924 #endif // U_HIDE_DEPRECATED_API
926 
934  /*
935  * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
936  * It matches lines like
937  * U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
938  */
939 
946 #ifndef U_HIDE_DEPRECATED_API
947 
953  U_BPT_COUNT /* 3 */
954 #endif // U_HIDE_DEPRECATED_API
956 
962  /*
963  * Note: UBlockCode constants are parsed by preparseucd.py.
964  * It matches lines like
965  * UBLOCK_<Unicode Block value name> = <integer>,
966  */
967 
969  UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
970 
972  UBLOCK_BASIC_LATIN = 1, /*[0000]*/
973 
976 
978  UBLOCK_LATIN_EXTENDED_A =3, /*[0100]*/
979 
981  UBLOCK_LATIN_EXTENDED_B =4, /*[0180]*/
982 
984  UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
985 
988 
991 
996  UBLOCK_GREEK =8, /*[0370]*/
997 
999  UBLOCK_CYRILLIC =9, /*[0400]*/
1000 
1002  UBLOCK_ARMENIAN =10, /*[0530]*/
1003 
1005  UBLOCK_HEBREW =11, /*[0590]*/
1006 
1008  UBLOCK_ARABIC =12, /*[0600]*/
1009 
1011  UBLOCK_SYRIAC =13, /*[0700]*/
1012 
1014  UBLOCK_THAANA =14, /*[0780]*/
1015 
1017  UBLOCK_DEVANAGARI =15, /*[0900]*/
1018 
1020  UBLOCK_BENGALI =16, /*[0980]*/
1021 
1023  UBLOCK_GURMUKHI =17, /*[0A00]*/
1024 
1026  UBLOCK_GUJARATI =18, /*[0A80]*/
1027 
1029  UBLOCK_ORIYA =19, /*[0B00]*/
1030 
1032  UBLOCK_TAMIL =20, /*[0B80]*/
1033 
1035  UBLOCK_TELUGU =21, /*[0C00]*/
1036 
1038  UBLOCK_KANNADA =22, /*[0C80]*/
1039 
1041  UBLOCK_MALAYALAM =23, /*[0D00]*/
1042 
1044  UBLOCK_SINHALA =24, /*[0D80]*/
1045 
1047  UBLOCK_THAI =25, /*[0E00]*/
1048 
1050  UBLOCK_LAO =26, /*[0E80]*/
1051 
1053  UBLOCK_TIBETAN =27, /*[0F00]*/
1054 
1056  UBLOCK_MYANMAR =28, /*[1000]*/
1057 
1059  UBLOCK_GEORGIAN =29, /*[10A0]*/
1060 
1062  UBLOCK_HANGUL_JAMO =30, /*[1100]*/
1063 
1065  UBLOCK_ETHIOPIC =31, /*[1200]*/
1066 
1068  UBLOCK_CHEROKEE =32, /*[13A0]*/
1069 
1072 
1074  UBLOCK_OGHAM =34, /*[1680]*/
1075 
1077  UBLOCK_RUNIC =35, /*[16A0]*/
1078 
1080  UBLOCK_KHMER =36, /*[1780]*/
1081 
1083  UBLOCK_MONGOLIAN =37, /*[1800]*/
1084 
1087 
1089  UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
1090 
1093 
1096 
1098  UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
1099 
1105 
1108 
1110  UBLOCK_NUMBER_FORMS =45, /*[2150]*/
1111 
1113  UBLOCK_ARROWS =46, /*[2190]*/
1114 
1117 
1120 
1122  UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
1123 
1126 
1129 
1131  UBLOCK_BOX_DRAWING =52, /*[2500]*/
1132 
1134  UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
1135 
1137  UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
1138 
1141 
1143  UBLOCK_DINGBATS =56, /*[2700]*/
1144 
1146  UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
1147 
1150 
1152  UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
1153 
1156 
1159 
1161  UBLOCK_HIRAGANA =62, /*[3040]*/
1162 
1164  UBLOCK_KATAKANA =63, /*[30A0]*/
1165 
1167  UBLOCK_BOPOMOFO =64, /*[3100]*/
1168 
1171 
1173  UBLOCK_KANBUN =66, /*[3190]*/
1174 
1177 
1180 
1183 
1186 
1189 
1191  UBLOCK_YI_SYLLABLES =72, /*[A000]*/
1192 
1194  UBLOCK_YI_RADICALS =73, /*[A490]*/
1195 
1197  UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
1198 
1200  UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
1201 
1204 
1206  UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
1207 
1217  UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
1228 
1231 
1234 
1237 
1240 
1243 
1246 
1249 
1251  UBLOCK_SPECIALS =86, /*[FFF0]*/
1252 
1255 
1256  /* New blocks in Unicode 3.1 */
1257 
1259  UBLOCK_OLD_ITALIC = 88, /*[10300]*/
1261  UBLOCK_GOTHIC = 89, /*[10330]*/
1263  UBLOCK_DESERET = 90, /*[10400]*/
1267  UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
1275  UBLOCK_TAGS = 96, /*[E0000]*/
1276 
1277  /* New blocks in Unicode 3.2 */
1278 
1287  UBLOCK_TAGALOG = 98, /*[1700]*/
1289  UBLOCK_HANUNOO = 99, /*[1720]*/
1291  UBLOCK_BUHID = 100, /*[1740]*/
1293  UBLOCK_TAGBANWA = 101, /*[1760]*/
1307  UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
1312 
1313  /* New blocks in Unicode 4 */
1314 
1316  UBLOCK_LIMBU = 111, /*[1900]*/
1318  UBLOCK_TAI_LE = 112, /*[1950]*/
1320  UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
1322  UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
1328  UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
1330  UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
1332  UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
1334  UBLOCK_UGARITIC = 120, /*[10380]*/
1336  UBLOCK_SHAVIAN = 121, /*[10450]*/
1338  UBLOCK_OSMANYA = 122, /*[10480]*/
1340  UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
1342  UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
1345 
1346  /* New blocks in Unicode 4.1 */
1347 
1351  UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
1353  UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
1355  UBLOCK_BUGINESE = 129, /*[1A00]*/
1357  UBLOCK_CJK_STROKES = 130, /*[31C0]*/
1361  UBLOCK_COPTIC = 132, /*[2C80]*/
1363  UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
1365  UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
1367  UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
1369  UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
1371  UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
1375  UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
1377  UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
1383  UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
1385  UBLOCK_TIFINAGH = 144, /*[2D30]*/
1387  UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
1388 
1389  /* New blocks in Unicode 5.0 */
1390 
1392  UBLOCK_NKO = 146, /*[07C0]*/
1394  UBLOCK_BALINESE = 147, /*[1B00]*/
1396  UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
1398  UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
1400  UBLOCK_PHAGS_PA = 150, /*[A840]*/
1402  UBLOCK_PHOENICIAN = 151, /*[10900]*/
1404  UBLOCK_CUNEIFORM = 152, /*[12000]*/
1408  UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
1409 
1410  /* New blocks in Unicode 5.1 */
1411 
1413  UBLOCK_SUNDANESE = 155, /*[1B80]*/
1415  UBLOCK_LEPCHA = 156, /*[1C00]*/
1417  UBLOCK_OL_CHIKI = 157, /*[1C50]*/
1419  UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
1421  UBLOCK_VAI = 159, /*[A500]*/
1423  UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
1425  UBLOCK_SAURASHTRA = 161, /*[A880]*/
1427  UBLOCK_KAYAH_LI = 162, /*[A900]*/
1429  UBLOCK_REJANG = 163, /*[A930]*/
1431  UBLOCK_CHAM = 164, /*[AA00]*/
1433  UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
1435  UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
1437  UBLOCK_LYCIAN = 167, /*[10280]*/
1439  UBLOCK_CARIAN = 168, /*[102A0]*/
1441  UBLOCK_LYDIAN = 169, /*[10920]*/
1443  UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
1445  UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
1446 
1447  /* New blocks in Unicode 5.2 */
1448 
1450  UBLOCK_SAMARITAN = 172, /*[0800]*/
1454  UBLOCK_TAI_THAM = 174, /*[1A20]*/
1456  UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
1458  UBLOCK_LISU = 176, /*[A4D0]*/
1460  UBLOCK_BAMUM = 177, /*[A6A0]*/
1464  UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
1468  UBLOCK_JAVANESE = 181, /*[A980]*/
1470  UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
1472  UBLOCK_TAI_VIET = 183, /*[AA80]*/
1474  UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
1478  UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
1480  UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
1482  UBLOCK_AVESTAN = 188, /*[10B00]*/
1486  UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
1488  UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
1490  UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
1492  UBLOCK_KAITHI = 193, /*[11080]*/
1494  UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
1501 
1502  /* New blocks in Unicode 6.0 */
1503 
1505  UBLOCK_MANDAIC = 198, /*[0840]*/
1507  UBLOCK_BATAK = 199, /*[1BC0]*/
1509  UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
1511  UBLOCK_BRAHMI = 201, /*[11000]*/
1513  UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
1515  UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
1517  UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
1521  UBLOCK_EMOTICONS = 206, /*[1F600]*/
1525  UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
1528 
1529  /* New blocks in Unicode 6.1 */
1530 
1532  UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
1536  UBLOCK_CHAKMA = 212, /*[11100]*/
1540  UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
1542  UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
1544  UBLOCK_MIAO = 216, /*[16F00]*/
1546  UBLOCK_SHARADA = 217, /*[11180]*/
1548  UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
1552  UBLOCK_TAKRI = 220, /*[11680]*/
1553 
1554  /* New blocks in Unicode 7.0 */
1555 
1557  UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
1559  UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
1561  UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
1565  UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
1567  UBLOCK_ELBASAN = 226, /*[10500]*/
1571  UBLOCK_GRANTHA = 228, /*[11300]*/
1573  UBLOCK_KHOJKI = 229, /*[11200]*/
1575  UBLOCK_KHUDAWADI = 230, /*[112B0]*/
1577  UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
1579  UBLOCK_LINEAR_A = 232, /*[10600]*/
1581  UBLOCK_MAHAJANI = 233, /*[11150]*/
1583  UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
1585  UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
1587  UBLOCK_MODI = 236, /*[11600]*/
1589  UBLOCK_MRO = 237, /*[16A40]*/
1591  UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
1593  UBLOCK_NABATAEAN = 239, /*[10880]*/
1595  UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
1597  UBLOCK_OLD_PERMIC = 241, /*[10350]*/
1599  UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
1601  UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
1603  UBLOCK_PALMYRENE = 244, /*[10860]*/
1605  UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
1607  UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
1611  UBLOCK_SIDDHAM = 248, /*[11580]*/
1615  UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
1617  UBLOCK_TIRHUTA = 251, /*[11480]*/
1619  UBLOCK_WARANG_CITI = 252, /*[118A0]*/
1620 
1621  /* New blocks in Unicode 8.0 */
1622 
1624  UBLOCK_AHOM = 253, /*[11700]*/
1626  UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
1628  UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
1634  UBLOCK_HATRAN = 258, /*[108E0]*/
1636  UBLOCK_MULTANI = 259, /*[11280]*/
1638  UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
1642  UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/
1643 
1644  /* New blocks in Unicode 9.0 */
1645 
1647  UBLOCK_ADLAM = 263, /*[1E900]*/
1649  UBLOCK_BHAIKSUKI = 264, /*[11C00]*/
1651  UBLOCK_CYRILLIC_EXTENDED_C = 265, /*[1C80]*/
1653  UBLOCK_GLAGOLITIC_SUPPLEMENT = 266, /*[1E000]*/
1657  UBLOCK_MARCHEN = 268, /*[11C70]*/
1659  UBLOCK_MONGOLIAN_SUPPLEMENT = 269, /*[11660]*/
1661  UBLOCK_NEWA = 270, /*[11400]*/
1663  UBLOCK_OSAGE = 271, /*[104B0]*/
1665  UBLOCK_TANGUT = 272, /*[17000]*/
1667  UBLOCK_TANGUT_COMPONENTS = 273, /*[18800]*/
1668 
1669  // New blocks in Unicode 10.0
1670 
1674  UBLOCK_KANA_EXTENDED_A = 275, /*[1B100]*/
1676  UBLOCK_MASARAM_GONDI = 276, /*[11D00]*/
1678  UBLOCK_NUSHU = 277, /*[1B170]*/
1680  UBLOCK_SOYOMBO = 278, /*[11A50]*/
1682  UBLOCK_SYRIAC_SUPPLEMENT = 279, /*[0860]*/
1684  UBLOCK_ZANABAZAR_SQUARE = 280, /*[11A00]*/
1685 
1686 #ifndef U_HIDE_DEPRECATED_API
1687 
1694 #endif // U_HIDE_DEPRECATED_API
1695 
1698 };
1699 
1701 typedef enum UBlockCode UBlockCode;
1702 
1710 typedef enum UEastAsianWidth {
1711  /*
1712  * Note: UEastAsianWidth constants are parsed by preparseucd.py.
1713  * It matches lines like
1714  * U_EA_<Unicode East_Asian_Width value name>
1715  */
1716 
1717  U_EA_NEUTRAL, /*[N]*/
1718  U_EA_AMBIGUOUS, /*[A]*/
1719  U_EA_HALFWIDTH, /*[H]*/
1720  U_EA_FULLWIDTH, /*[F]*/
1721  U_EA_NARROW, /*[Na]*/
1722  U_EA_WIDE, /*[W]*/
1723 #ifndef U_HIDE_DEPRECATED_API
1724 
1731 #endif // U_HIDE_DEPRECATED_API
1732 } UEastAsianWidth;
1733 
1745 typedef enum UCharNameChoice {
1748 #ifndef U_HIDE_DEPRECATED_API
1749 
1755 #endif /* U_HIDE_DEPRECATED_API */
1756 
1760 #ifndef U_HIDE_DEPRECATED_API
1761 
1766 #endif // U_HIDE_DEPRECATED_API
1767 } UCharNameChoice;
1768 
1782 typedef enum UPropertyNameChoice {
1783  U_SHORT_PROPERTY_NAME,
1784  U_LONG_PROPERTY_NAME,
1785 #ifndef U_HIDE_DEPRECATED_API
1786 
1791 #endif // U_HIDE_DEPRECATED_API
1793 
1800 typedef enum UDecompositionType {
1801  /*
1802  * Note: UDecompositionType constants are parsed by preparseucd.py.
1803  * It matches lines like
1804  * U_DT_<Unicode Decomposition_Type value name>
1805  */
1806 
1807  U_DT_NONE, /*[none]*/
1808  U_DT_CANONICAL, /*[can]*/
1809  U_DT_COMPAT, /*[com]*/
1810  U_DT_CIRCLE, /*[enc]*/
1811  U_DT_FINAL, /*[fin]*/
1812  U_DT_FONT, /*[font]*/
1813  U_DT_FRACTION, /*[fra]*/
1814  U_DT_INITIAL, /*[init]*/
1815  U_DT_ISOLATED, /*[iso]*/
1816  U_DT_MEDIAL, /*[med]*/
1817  U_DT_NARROW, /*[nar]*/
1818  U_DT_NOBREAK, /*[nb]*/
1819  U_DT_SMALL, /*[sml]*/
1820  U_DT_SQUARE, /*[sqr]*/
1821  U_DT_SUB, /*[sub]*/
1822  U_DT_SUPER, /*[sup]*/
1823  U_DT_VERTICAL, /*[vert]*/
1824  U_DT_WIDE, /*[wide]*/
1825 #ifndef U_HIDE_DEPRECATED_API
1826 
1832  U_DT_COUNT /* 18 */
1833 #endif // U_HIDE_DEPRECATED_API
1835 
1842 typedef enum UJoiningType {
1843  /*
1844  * Note: UJoiningType constants are parsed by preparseucd.py.
1845  * It matches lines like
1846  * U_JT_<Unicode Joining_Type value name>
1847  */
1848 
1849  U_JT_NON_JOINING, /*[U]*/
1850  U_JT_JOIN_CAUSING, /*[C]*/
1851  U_JT_DUAL_JOINING, /*[D]*/
1852  U_JT_LEFT_JOINING, /*[L]*/
1853  U_JT_RIGHT_JOINING, /*[R]*/
1854  U_JT_TRANSPARENT, /*[T]*/
1855 #ifndef U_HIDE_DEPRECATED_API
1856 
1862  U_JT_COUNT /* 6 */
1863 #endif // U_HIDE_DEPRECATED_API
1864 } UJoiningType;
1865 
1872 typedef enum UJoiningGroup {
1873  /*
1874  * Note: UJoiningGroup constants are parsed by preparseucd.py.
1875  * It matches lines like
1876  * U_JG_<Unicode Joining_Group value name>
1877  */
1878 
1879  U_JG_NO_JOINING_GROUP,
1880  U_JG_AIN,
1881  U_JG_ALAPH,
1882  U_JG_ALEF,
1883  U_JG_BEH,
1884  U_JG_BETH,
1885  U_JG_DAL,
1886  U_JG_DALATH_RISH,
1887  U_JG_E,
1888  U_JG_FEH,
1889  U_JG_FINAL_SEMKATH,
1890  U_JG_GAF,
1891  U_JG_GAMAL,
1892  U_JG_HAH,
1894  U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
1895  U_JG_HE,
1896  U_JG_HEH,
1897  U_JG_HEH_GOAL,
1898  U_JG_HETH,
1899  U_JG_KAF,
1900  U_JG_KAPH,
1901  U_JG_KNOTTED_HEH,
1902  U_JG_LAM,
1903  U_JG_LAMADH,
1904  U_JG_MEEM,
1905  U_JG_MIM,
1906  U_JG_NOON,
1907  U_JG_NUN,
1908  U_JG_PE,
1909  U_JG_QAF,
1910  U_JG_QAPH,
1911  U_JG_REH,
1912  U_JG_REVERSED_PE,
1913  U_JG_SAD,
1914  U_JG_SADHE,
1915  U_JG_SEEN,
1916  U_JG_SEMKATH,
1917  U_JG_SHIN,
1918  U_JG_SWASH_KAF,
1919  U_JG_SYRIAC_WAW,
1920  U_JG_TAH,
1921  U_JG_TAW,
1922  U_JG_TEH_MARBUTA,
1923  U_JG_TETH,
1924  U_JG_WAW,
1925  U_JG_YEH,
1926  U_JG_YEH_BARREE,
1927  U_JG_YEH_WITH_TAIL,
1928  U_JG_YUDH,
1929  U_JG_YUDH_HE,
1930  U_JG_ZAIN,
1982 #ifndef U_HIDE_DEPRECATED_API
1983 
1990 #endif // U_HIDE_DEPRECATED_API
1991 } UJoiningGroup;
1992 
2000  /*
2001  * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2002  * It matches lines like
2003  * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2004  */
2005 
2006  U_GCB_OTHER = 0, /*[XX]*/
2007  U_GCB_CONTROL = 1, /*[CN]*/
2008  U_GCB_CR = 2, /*[CR]*/
2009  U_GCB_EXTEND = 3, /*[EX]*/
2010  U_GCB_L = 4, /*[L]*/
2011  U_GCB_LF = 5, /*[LF]*/
2012  U_GCB_LV = 6, /*[LV]*/
2013  U_GCB_LVT = 7, /*[LVT]*/
2014  U_GCB_T = 8, /*[T]*/
2015  U_GCB_V = 9, /*[V]*/
2017  U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2019  U_GCB_PREPEND = 11, /*[PP]*/
2021  U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2023  U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2025  U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2027  U_GCB_E_MODIFIER = 15, /*[EM]*/
2029  U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2031  U_GCB_ZWJ = 17, /*[ZWJ]*/
2032 #ifndef U_HIDE_DEPRECATED_API
2033 
2040 #endif // U_HIDE_DEPRECATED_API
2042 
2050 typedef enum UWordBreakValues {
2051  /*
2052  * Note: UWordBreakValues constants are parsed by preparseucd.py.
2053  * It matches lines like
2054  * U_WB_<Unicode Word_Break value name>
2055  */
2056 
2057  U_WB_OTHER = 0, /*[XX]*/
2058  U_WB_ALETTER = 1, /*[LE]*/
2059  U_WB_FORMAT = 2, /*[FO]*/
2060  U_WB_KATAKANA = 3, /*[KA]*/
2061  U_WB_MIDLETTER = 4, /*[ML]*/
2062  U_WB_MIDNUM = 5, /*[MN]*/
2063  U_WB_NUMERIC = 6, /*[NU]*/
2064  U_WB_EXTENDNUMLET = 7, /*[EX]*/
2066  U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2068  U_WB_EXTEND = 9, /*[Extend]*/
2070  U_WB_LF = 10, /*[LF]*/
2072  U_WB_MIDNUMLET =11, /*[MB]*/
2074  U_WB_NEWLINE =12, /*[NL]*/
2076  U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2078  U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2080  U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2082  U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2084  U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2086  U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2088  U_WB_E_MODIFIER = 19, /*[EM]*/
2090  U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2092  U_WB_ZWJ = 21, /*[ZWJ]*/
2093 #ifndef U_HIDE_DEPRECATED_API
2094 
2101 #endif // U_HIDE_DEPRECATED_API
2103 
2110 typedef enum USentenceBreak {
2111  /*
2112  * Note: USentenceBreak constants are parsed by preparseucd.py.
2113  * It matches lines like
2114  * U_SB_<Unicode Sentence_Break value name>
2115  */
2116 
2117  U_SB_OTHER = 0, /*[XX]*/
2118  U_SB_ATERM = 1, /*[AT]*/
2119  U_SB_CLOSE = 2, /*[CL]*/
2120  U_SB_FORMAT = 3, /*[FO]*/
2121  U_SB_LOWER = 4, /*[LO]*/
2122  U_SB_NUMERIC = 5, /*[NU]*/
2123  U_SB_OLETTER = 6, /*[LE]*/
2124  U_SB_SEP = 7, /*[SE]*/
2125  U_SB_SP = 8, /*[SP]*/
2126  U_SB_STERM = 9, /*[ST]*/
2127  U_SB_UPPER = 10, /*[UP]*/
2128  U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2129  U_SB_EXTEND = 12, /*[EX]*/
2130  U_SB_LF = 13, /*[LF]*/
2131  U_SB_SCONTINUE = 14, /*[SC]*/
2132 #ifndef U_HIDE_DEPRECATED_API
2133 
2140 #endif // U_HIDE_DEPRECATED_API
2141 } USentenceBreak;
2142 
2149 typedef enum ULineBreak {
2150  /*
2151  * Note: ULineBreak constants are parsed by preparseucd.py.
2152  * It matches lines like
2153  * U_LB_<Unicode Line_Break value name>
2154  */
2155 
2156  U_LB_UNKNOWN = 0, /*[XX]*/
2157  U_LB_AMBIGUOUS = 1, /*[AI]*/
2158  U_LB_ALPHABETIC = 2, /*[AL]*/
2159  U_LB_BREAK_BOTH = 3, /*[B2]*/
2160  U_LB_BREAK_AFTER = 4, /*[BA]*/
2161  U_LB_BREAK_BEFORE = 5, /*[BB]*/
2162  U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2163  U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2164  U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2165  U_LB_COMBINING_MARK = 9, /*[CM]*/
2166  U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2167  U_LB_EXCLAMATION = 11, /*[EX]*/
2168  U_LB_GLUE = 12, /*[GL]*/
2169  U_LB_HYPHEN = 13, /*[HY]*/
2170  U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2172  U_LB_INSEPARABLE = 15, /*[IN]*/
2173  U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2174  U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2175  U_LB_LINE_FEED = 17, /*[LF]*/
2176  U_LB_NONSTARTER = 18, /*[NS]*/
2177  U_LB_NUMERIC = 19, /*[NU]*/
2178  U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2179  U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2180  U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2181  U_LB_QUOTATION = 23, /*[QU]*/
2182  U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2183  U_LB_SURROGATE = 25, /*[SG]*/
2184  U_LB_SPACE = 26, /*[SP]*/
2185  U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2186  U_LB_ZWSPACE = 28, /*[ZW]*/
2188  U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2190  U_LB_WORD_JOINER = 30, /*[WJ]*/
2192  U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2194  U_LB_H3 = 32, /*[H3]*/
2196  U_LB_JL = 33, /*[JL]*/
2198  U_LB_JT = 34, /*[JT]*/
2200  U_LB_JV = 35, /*[JV]*/
2202  U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2204  U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2206  U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2208  U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2210  U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2212  U_LB_E_MODIFIER = 41, /*[EM]*/
2214  U_LB_ZWJ = 42, /*[ZWJ]*/
2215 #ifndef U_HIDE_DEPRECATED_API
2216 
2223 #endif // U_HIDE_DEPRECATED_API
2224 } ULineBreak;
2225 
2232 typedef enum UNumericType {
2233  /*
2234  * Note: UNumericType constants are parsed by preparseucd.py.
2235  * It matches lines like
2236  * U_NT_<Unicode Numeric_Type value name>
2237  */
2238 
2239  U_NT_NONE, /*[None]*/
2240  U_NT_DECIMAL, /*[de]*/
2241  U_NT_DIGIT, /*[di]*/
2242  U_NT_NUMERIC, /*[nu]*/
2243 #ifndef U_HIDE_DEPRECATED_API
2244 
2251 #endif // U_HIDE_DEPRECATED_API
2252 } UNumericType;
2253 
2260 typedef enum UHangulSyllableType {
2261  /*
2262  * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2263  * It matches lines like
2264  * U_HST_<Unicode Hangul_Syllable_Type value name>
2265  */
2266 
2267  U_HST_NOT_APPLICABLE, /*[NA]*/
2268  U_HST_LEADING_JAMO, /*[L]*/
2269  U_HST_VOWEL_JAMO, /*[V]*/
2270  U_HST_TRAILING_JAMO, /*[T]*/
2271  U_HST_LV_SYLLABLE, /*[LV]*/
2272  U_HST_LVT_SYLLABLE, /*[LVT]*/
2273 #ifndef U_HIDE_DEPRECATED_API
2274 
2281 #endif // U_HIDE_DEPRECATED_API
2283 
2310 U_STABLE UBool U_EXPORT2
2312 
2325 U_STABLE UBool U_EXPORT2
2327 
2340 U_STABLE UBool U_EXPORT2
2342 
2355 U_STABLE UBool U_EXPORT2
2357 
2376 U_STABLE UBool U_EXPORT2
2378 
2416 U_STABLE int32_t U_EXPORT2
2418 
2437 U_STABLE int32_t U_EXPORT2
2439 
2466 U_STABLE int32_t U_EXPORT2
2468 
2491 U_STABLE double U_EXPORT2
2493 
2501 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
2502 
2526 U_STABLE UBool U_EXPORT2
2527 u_islower(UChar32 c);
2528 
2553 U_STABLE UBool U_EXPORT2
2554 u_isupper(UChar32 c);
2555 
2570 U_STABLE UBool U_EXPORT2
2571 u_istitle(UChar32 c);
2572 
2591 U_STABLE UBool U_EXPORT2
2592 u_isdigit(UChar32 c);
2593 
2612 U_STABLE UBool U_EXPORT2
2613 u_isalpha(UChar32 c);
2614 
2633 U_STABLE UBool U_EXPORT2
2634 u_isalnum(UChar32 c);
2635 
2656 U_STABLE UBool U_EXPORT2
2657 u_isxdigit(UChar32 c);
2658 
2672 U_STABLE UBool U_EXPORT2
2673 u_ispunct(UChar32 c);
2674 
2691 U_STABLE UBool U_EXPORT2
2692 u_isgraph(UChar32 c);
2693 
2720 U_STABLE UBool U_EXPORT2
2721 u_isblank(UChar32 c);
2722 
2745 U_STABLE UBool U_EXPORT2
2746 u_isdefined(UChar32 c);
2747 
2766 U_STABLE UBool U_EXPORT2
2767 u_isspace(UChar32 c);
2768 
2787 U_STABLE UBool U_EXPORT2
2789 
2827 U_STABLE UBool U_EXPORT2
2829 
2851 U_STABLE UBool U_EXPORT2
2852 u_iscntrl(UChar32 c);
2853 
2866 U_STABLE UBool U_EXPORT2
2868 
2884 U_STABLE UBool U_EXPORT2
2885 u_isprint(UChar32 c);
2886 
2905 U_STABLE UBool U_EXPORT2
2906 u_isbase(UChar32 c);
2907 
2924 U_STABLE UCharDirection U_EXPORT2
2926 
2942 U_STABLE UBool U_EXPORT2
2944 
2964 U_STABLE UChar32 U_EXPORT2
2966 
2983 U_STABLE UChar32 U_EXPORT2
2985 
2997 U_STABLE int8_t U_EXPORT2
2998 u_charType(UChar32 c);
2999 
3013 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3014 
3032 typedef UBool U_CALLCONV
3033 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3034 
3054 U_STABLE void U_EXPORT2
3055 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3056 
3057 #if !UCONFIG_NO_NORMALIZATION
3058 
3066 U_STABLE uint8_t U_EXPORT2
3068 
3069 #endif
3070 
3094 U_STABLE int32_t U_EXPORT2
3096 
3106 U_STABLE UBlockCode U_EXPORT2
3108 
3141 U_STABLE int32_t U_EXPORT2
3142 u_charName(UChar32 code, UCharNameChoice nameChoice,
3143  char *buffer, int32_t bufferLength,
3144  UErrorCode *pErrorCode);
3145 
3146 #ifndef U_HIDE_DEPRECATED_API
3147 
3165 U_DEPRECATED int32_t U_EXPORT2
3167  char *dest, int32_t destCapacity,
3168  UErrorCode *pErrorCode);
3169 #endif /* U_HIDE_DEPRECATED_API */
3170 
3191 U_STABLE UChar32 U_EXPORT2
3192 u_charFromName(UCharNameChoice nameChoice,
3193  const char *name,
3194  UErrorCode *pErrorCode);
3195 
3213 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3214  UChar32 code,
3215  UCharNameChoice nameChoice,
3216  const char *name,
3217  int32_t length);
3218 
3240 U_STABLE void U_EXPORT2
3241 u_enumCharNames(UChar32 start, UChar32 limit,
3242  UEnumCharNamesFn *fn,
3243  void *context,
3244  UCharNameChoice nameChoice,
3245  UErrorCode *pErrorCode);
3246 
3278 U_STABLE const char* U_EXPORT2
3279 u_getPropertyName(UProperty property,
3280  UPropertyNameChoice nameChoice);
3281 
3301 U_STABLE UProperty U_EXPORT2
3302 u_getPropertyEnum(const char* alias);
3303 
3351 U_STABLE const char* U_EXPORT2
3353  int32_t value,
3354  UPropertyNameChoice nameChoice);
3355 
3387 U_STABLE int32_t U_EXPORT2
3389  const char* alias);
3390 
3408 U_STABLE UBool U_EXPORT2
3409 u_isIDStart(UChar32 c);
3410 
3432 U_STABLE UBool U_EXPORT2
3433 u_isIDPart(UChar32 c);
3434 
3455 U_STABLE UBool U_EXPORT2
3457 
3474 U_STABLE UBool U_EXPORT2
3476 
3495 U_STABLE UBool U_EXPORT2
3497 
3520 U_STABLE UChar32 U_EXPORT2
3521 u_tolower(UChar32 c);
3522 
3545 U_STABLE UChar32 U_EXPORT2
3546 u_toupper(UChar32 c);
3547 
3570 U_STABLE UChar32 U_EXPORT2
3571 u_totitle(UChar32 c);
3572 
3595 U_STABLE UChar32 U_EXPORT2
3596 u_foldCase(UChar32 c, uint32_t options);
3597 
3636 U_STABLE int32_t U_EXPORT2
3637 u_digit(UChar32 ch, int8_t radix);
3638 
3667 U_STABLE UChar32 U_EXPORT2
3668 u_forDigit(int32_t digit, int8_t radix);
3669 
3684 U_STABLE void U_EXPORT2
3685 u_charAge(UChar32 c, UVersionInfo versionArray);
3686 
3698 U_STABLE void U_EXPORT2
3699 u_getUnicodeVersion(UVersionInfo versionArray);
3700 
3701 #if !UCONFIG_NO_NORMALIZATION
3702 
3723 U_STABLE int32_t U_EXPORT2
3724 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
3725 
3726 #endif
3727 
3728 
3730 
3731 #endif /*_UCHAR*/
3732 /*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:241
Unicode 4.0.1 renames the "Cyrillic Supplementary" block to "Cyrillic Supplement".
Definition: uchar.h:1285
Binary property Changes_When_Lowercased.
Definition: uchar.h:393
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition: uchar.h:245
Binary property Case_Ignorable.
Definition: uchar.h:391
UBool u_isUWhiteSpace(UChar32 c)
Check if a code point has the White_Space Unicode property.
UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.
Enumerated property NFC_Quick_Check.
Definition: uchar.h:504
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.
One more than the highest normal UJoiningGroup value.
Definition: uchar.h:1989
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:645
const char * u_getPropertyName(UProperty property, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property, as given in the Unicode database file PropertyAliases...
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1227
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.
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:461
UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
Binary property XID_Start.
Definition: uchar.h:298
Binary property Join_Control.
Definition: uchar.h:252
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition: uchar.h:256
Binary property White_Space.
Definition: uchar.h:291
String property Titlecase_Mapping.
Definition: uchar.h:618
One more than the highest normal UPropertyNameChoice value.
Definition: uchar.h:1790
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:547
Enumerated property Numeric_Type.
Definition: uchar.h:489
Binary property xdigit (a C/POSIX character class).
Definition: uchar.h:387
UBlockCode ublock_getCode(UChar32 c)
Returns the Unicode allocation block that contains the character.
Binary property Alphabetic.
Definition: uchar.h:179
First constant for double Unicode properties.
Definition: uchar.h:573
C API: Bit set option bit constants for various string and character processing functions.
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:3213
One more than the highest normal USentenceBreak value.
Definition: uchar.h:2139
UBool u_isgraph(UChar32 c)
Determines whether the specified code point is a "graphic" character (printable, excluding spaces)...
Binary property Emoji_Component.
Definition: uchar.h:438
Binary property Emoji.
Definition: uchar.h:410
Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNE...
Definition: uchar.h:677
String property Simple_Case_Folding.
Definition: uchar.h:606
Binary property NFC_Inert.
Definition: uchar.h:334
Binary property graph (a C/POSIX character class).
Definition: uchar.h:377
String property Bidi_Mirroring_Glyph.
Definition: uchar.h:589
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:566
UBool u_isdefined(UChar32 c)
Determines whether the specified code point is "defined", which usually means that it is assigned a c...
Enumerated property Block.
Definition: uchar.h:464
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:657
Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2172
Binary property Radical (new in Unicode 3.2).
Definition: uchar.h:271
UCharDirection
This specifies the language directional property of a character set.
Definition: uchar.h:862
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition: uchar.h:249
Binary property Grapheme_Link (new in Unicode 3.2).
Definition: uchar.h:223
Enumerated property Decomposition_Type.
Definition: uchar.h:470
Binary property Regional_Indicator.
Definition: uchar.h:443
#define U_CALLCONV
Similar to U_CDECL_BEGIN/U_CDECL_END, this qualifier is necessary in callback function typedefs to ma...
Definition: platform.h:848
String property Case_Folding.
Definition: uchar.h:592
String property Name.
Definition: uchar.h:603
String property Simple_Uppercase_Mapping.
Definition: uchar.h:615
UChar32 u_forDigit(int32_t digit, int8_t radix)
Determines the character representation for a specific digit in the specified radix.
Enumerated property Bidi_Class.
Definition: uchar.h:459
int32_t u_charDigitValue(UChar32 c)
Returns the decimal digit value of a decimal digit character.
Enumerated property General_Category.
Definition: uchar.h:477
Binary property Prepended_Concatenation_Mark.
Definition: uchar.h:448
Sm.
Definition: uchar.h:725
String property Unicode_1_Name.
Definition: uchar.h:624
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.
UNumericType
Numeric Type constants.
Definition: uchar.h:2232
One more than the highest normal UBidiPairedBracketType value.
Definition: uchar.h:953
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:362
Close paired bracket.
Definition: uchar.h:945
UBool u_iscntrl(UChar32 c)
Determines whether the specified code point is a control character (as defined by this function)...
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:3033
Binary property Changes_When_Casefolded.
Definition: uchar.h:399
Binary property NFD_Inert.
Definition: uchar.h:320
Binary property Diacritic.
Definition: uchar.h:204
Binary property Terminal_Punctuation.
Definition: uchar.h:280
UChar32 u_charFromName(UCharNameChoice nameChoice, const char *name, UErrorCode *pErrorCode)
Find a Unicode character by its name and return its code point value.
One more than the highest normal UNumericType value.
Definition: uchar.h:2250
UBool u_isUAlphabetic(UChar32 c)
Check if a code point has the Alphabetic Unicode property.
Enumerated property NFD_Quick_Check.
Definition: uchar.h:498
void u_charAge(UChar32 c, UVersionInfo versionArray)
Get the "age" of the code point.
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...
#define U_CDECL_BEGIN
This is used to begin a declaration of a library private ICU C API.
Definition: umachine.h:84
Binary property STerm (new in Unicode 4.0.1).
Definition: uchar.h:307
Enumerated property Joining_Group.
Definition: uchar.h:480
Binary property ID_Continue.
Definition: uchar.h:234
Binary property Emoji_Presentation.
Definition: uchar.h:417
Open paired bracket.
Definition: uchar.h:943
Binary property blank (a C/POSIX character class).
Definition: uchar.h:372
Binary property Quotation_Mark.
Definition: uchar.h:267
#define U_DEPRECATED
This is used to declare a function as a deprecated public ICU C API.
Definition: umachine.h:115
Binary property Changes_When_NFKC_Casefolded.
Definition: uchar.h:403
First constant for binary Unicode properties.
Definition: uchar.h:181
Binary property Noncharacter_Code_Point.
Definition: uchar.h:265
Binary property Hyphen.
Definition: uchar.h:229
Enumerated property East_Asian_Width.
Definition: uchar.h:474
ULineBreak
Line Break constants.
Definition: uchar.h:2149
One more than the highest UCharDirection value.
Definition: uchar.h:923
Binary property Full_Composition_Exclusion.
Definition: uchar.h:212
Bitmask property General_Category_Mask.
Definition: uchar.h:558
String property Simple_Titlecase_Mapping.
Definition: uchar.h:612
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2039
Unicode 3.2 renames this block to "Combining Diacritical Marks for Symbols".
Definition: uchar.h:1104
int32_t u_digit(UChar32 ch, int8_t radix)
Returns the decimal digit value of the code point in the specified radix.
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:1800
UBool u_isprint(UChar32 c)
Determines whether the specified code point is a printable character.
UBool u_isxdigit(UChar32 c)
Determines whether the specified code point is a hexadecimal digit.
One more than the highest normal ULineBreak value.
Definition: uchar.h:2222
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2260
String property Simple_Lowercase_Mapping.
Definition: uchar.h:609
Binary property print (a C/POSIX character class).
Definition: uchar.h:382
Binary property Case_Sensitive.
Definition: uchar.h:302
Standard or synthetic character name.
Definition: uchar.h:1757
Binary property Bidi_Mirrored.
Definition: uchar.h:192
Binary property NFKC_Inert.
Definition: uchar.h:341
int32_t u_getIntPropertyValue(UChar32 c, UProperty which)
Get the property value for an enumerated or integer Unicode property for a code point.
Binary property Changes_When_Casemapped.
Definition: uchar.h:401
First constant for string Unicode properties.
Definition: uchar.h:586
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition: uchar.h:220
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:396
UGraphemeClusterBreak
Grapheme Cluster Break constants.
Definition: uchar.h:1999
New No_Block value in Unicode 4.
Definition: uchar.h:969
Binary property Extender.
Definition: uchar.h:208
Double property Numeric_Value.
Definition: uchar.h:571
Binary property Math.
Definition: uchar.h:261
Unicode character name (Name property).
Definition: uchar.h:1747
Binary property Grapheme_Base (new in Unicode 3.2).
Definition: uchar.h:216
Binary property NFKD_Inert.
Definition: uchar.h:327
uint8_t u_getCombiningClass(UChar32 c)
Returns the combining class of the code point as specified in UnicodeData.txt.
UBool u_isbase(UChar32 c)
Determines whether the specified code point is a base character.
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:665
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 Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:531
double u_getNumericValue(UChar32 c)
Get the numeric value for a Unicode code point as defined in the Unicode Character Database...
Binary property Lowercase.
Definition: uchar.h:259
UBool u_isJavaIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in a Java identifier...
First constant for bit-mask Unicode properties.
Definition: uchar.h:560
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:541
UBool u_isspace(UChar32 c)
Determines if the specified character is a space character or not.
USentenceBreak
Sentence Break constants.
Definition: uchar.h:2110
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:284
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:467
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:1745
One more than the last constant for binary Unicode properties.
Definition: uchar.h:454
UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
Enumerated property Script.
Definition: uchar.h:492
Unicode 3.2 renames this block to "Greek and Coptic".
Definition: uchar.h:996
Binary property Hex_Digit.
Definition: uchar.h:226
String property Uppercase_Mapping.
Definition: uchar.h:628
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:1782
String property Lowercase_Mapping.
Definition: uchar.h:600
UCharDirection u_charDirection(UChar32 c)
Returns the bidirectional category value for the code point, which is used in the Unicode bidirection...
UBool u_islower(UChar32 c)
Determines whether the specified code point has the general category "Ll" (lowercase letter)...
uint16_t UChar
The base type for UTF-16 code units and pointers.
Definition: umachine.h:349
#define U_CDECL_END
This is used to end a declaration of a library private ICU C API.
Definition: umachine.h:85
Enumerated property NFKC_Quick_Check.
Definition: uchar.h:507
Not a paired bracket.
Definition: uchar.h:941
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:165
void u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context)
Enumerate efficiently all code points with their Unicode general categories.
Enumerated property Hangul_Syllable_Type, new in Unicode 4.
Definition: uchar.h:495
Binary property Dash.
Definition: uchar.h:194
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:367
Cf.
Definition: uchar.h:709
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition: uchar.h:313
One more than the highest normal UHangulSyllableType value.
Definition: uchar.h:2280
UBool u_isUUppercase(UChar32 c)
Check if a code point has the Uppercase Unicode property.
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:961
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:536
Binary property Deprecated (new in Unicode 3.2).
Definition: uchar.h:201
Binary property Bidi_Control.
Definition: uchar.h:187
Binary property XID_Continue.
Definition: uchar.h:295
Same as UBLOCK_PRIVATE_USE.
Definition: uchar.h:1217
UBool u_hasBinaryProperty(UChar32 c, UProperty which)
Check a binary Unicode property for a code point.
UErrorCode
Error code to replace exception handling, so that the code is compatible with all C++ compilers...
Definition: utypes.h:396
Binary property Uppercase.
Definition: uchar.h:287
void u_getUnicodeVersion(UVersionInfo versionArray)
Gets the Unicode version information.
Binary property Changes_When_Uppercased.
Definition: uchar.h:395
UJoiningGroup
Joining Group constants.
Definition: uchar.h:1872
Binary property Cased.
Definition: uchar.h:389
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:1730
One more than the highest normal UDecompositionType value.
Definition: uchar.h:1832
Cs.
Definition: uchar.h:713
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:1710
UBool u_isupper(UChar32 c)
Determines whether the specified code point has the general category "Lu" (uppercase letter)...
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:521
UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
Non-category for unassigned and non-character code points.
Definition: uchar.h:675
UBool u_ispunct(UChar32 c)
Determines whether the specified code point is a punctuation character.
First constant for Unicode properties with unusual value types.
Definition: uchar.h:647
UWordBreakValues
Word Break constants.
Definition: uchar.h:2050
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...
int32_t u_getIntPropertyMaxValue(UProperty which)
Get the maximum value for an enumerated/integer/binary Unicode property.
Enumerated property Joining_Type.
Definition: uchar.h:483
One more than the last constant for double Unicode properties.
Definition: uchar.h:579
Binary property Emoji_Modifier.
Definition: uchar.h:424
Basic definitions for ICU, for both C and C++ APIs.
UBool u_isIDPart(UChar32 c)
Determines if the specified character is permissible in an identifier according to Java...
String property Bidi_Paired_Bracket (new in Unicode 6.3).
Definition: uchar.h:631
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:514
Binary property ASCII_Hex_Digit.
Definition: uchar.h:183
UJoiningType
Joining Type constants.
Definition: uchar.h:1842
UBool u_isIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in an identifier accordin...
Binary property Soft_Dotted (new in Unicode 3.2).
Definition: uchar.h:276
One more than the last constant for string Unicode properties.
Definition: uchar.h:637
Binary property Emoji_Modifier_Base.
Definition: uchar.h:431
Binary Property Segment_Starter.
Definition: uchar.h:352
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.
Binary property ID_Start.
Definition: uchar.h:238
int32_t u_getIntPropertyMinValue(UProperty which)
Get the minimum value for an enumerated/integer/binary Unicode property.
UBool u_isJavaSpaceChar(UChar32 c)
Determine if the specified code point is a space character according to Java.
UBool u_isMirrored(UChar32 c)
Determines whether the code point has the Bidi_Mirrored property.
Corrected name from NameAliases.txt.
Definition: uchar.h:1759
Binary property Changes_When_Titlecased.
Definition: uchar.h:397
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.
Enumerated property Line_Break.
Definition: uchar.h:486
One more than the highest normal UJoiningType value.
Definition: uchar.h:1862
UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier...
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:526
UBool u_isdigit(UChar32 c)
Determines whether the specified code point is a digit character according to Java.
UProperty u_getPropertyEnum(const char *alias)
Return the UProperty enum for a given property name, as specified in the Unicode database file Proper...
UBool u_isWhitespace(UChar32 c)
Determines if the specified code point is a whitespace character according to Java/ICU.
Deprecated string property ISO_Comment.
Definition: uchar.h:596
UBool u_isblank(UChar32 c)
Determines whether the specified code point is a "blank" or "horizontal space", a character that visi...
UBool u_isalpha(UChar32 c)
Determines whether the specified code point is a letter character.
Binary property Default_Ignorable_Code_Point (new in Unicode 3.2).
Definition: uchar.h:198
One more than the highest normal UBlockCode value.
Definition: uchar.h:1693
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:1754
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:743
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2100
UBool u_isalnum(UChar32 c)
Determines whether the specified code point is an alphanumeric character (letter or digit) according ...
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition: uchar.h:357
One more than the last constant for Unicode properties with unusual value types.
Definition: uchar.h:653
UBidiPairedBracketType
Bidi Paired Bracket Type constants.
Definition: uchar.h:933
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:1765
UChar32 u_charMirror(UChar32 c)
Maps the specified character to a "mirror-image" character.
UBool u_isISOControl(UChar32 c)
Determines whether the specified code point is an ISO control code.
int8_t u_charType(UChar32 c)
Returns the general category value for the code point.
int32_t u_charName(UChar32 code, UCharNameChoice nameChoice, char *buffer, int32_t bufferLength, UErrorCode *pErrorCode)
Retrieve the name of a Unicode character.
String property Age.
Definition: uchar.h:584
#define U_STABLE
This is used to declare a function as a stable public ICU C API.
Definition: umachine.h:111
int8_t UBool
The ICU boolean type.
Definition: umachine.h:236
Enumerated property NFKD_Quick_Check.
Definition: uchar.h:501