ICU 57.1: uchar.h Source File

00001 /*
00002 **********************************************************************
00003 *   Copyright (C) 1997-2016, International Business Machines
00004 *   Corporation and others.  All Rights Reserved.
00005 **********************************************************************
00006 *
00007 * File UCHAR.H
00008 *
00009 * Modification History:
00010 *
00011 *   Date        Name        Description
00012 *   04/02/97    aliu        Creation.
00013 *   03/29/99    helena      Updated for C APIs.
00014 *   4/15/99     Madhu       Updated for C Implementation and Javadoc
00015 *   5/20/99     Madhu       Added the function u_getVersion()
00016 *   8/19/1999   srl         Upgraded scripts to Unicode 3.0
00017 *   8/27/1999   schererm    UCharDirection constants: U_...
00018 *   11/11/1999  weiv        added u_isalnum(), cleaned comments
00019 *   01/11/2000  helena      Renamed u_getVersion to u_getUnicodeVersion().
00020 ******************************************************************************
00021 */
00022 
00023 #ifndef UCHAR_H
00024 #define UCHAR_H
00025 
00026 #include "unicode/utypes.h"
00027 
00028 U_CDECL_BEGIN
00029 
00030 /*==========================================================================*/
00031 /* Unicode version number                                                   */
00032 /*==========================================================================*/
00042 #define U_UNICODE_VERSION "8.0"
00043 
00124 #define UCHAR_MIN_VALUE 0
00125 
00134 #define UCHAR_MAX_VALUE 0x10ffff
00135 
00140 #define U_MASK(x) ((uint32_t)1<<(x))
00141 
00161 typedef enum UProperty {
00162     /*
00163      * Note: UProperty constants are parsed by preparseucd.py.
00164      * It matches lines like
00165      *     UCHAR_<Unicode property name>=<integer>,
00166      */
00167 
00168     /*  Note: Place UCHAR_ALPHABETIC before UCHAR_BINARY_START so that
00169     debuggers display UCHAR_ALPHABETIC as the symbolic name for 0,
00170     rather than UCHAR_BINARY_START.  Likewise for other *_START
00171     identifiers. */
00172 
00175     UCHAR_ALPHABETIC=0,
00177     UCHAR_BINARY_START=UCHAR_ALPHABETIC,
00179     UCHAR_ASCII_HEX_DIGIT=1,
00183     UCHAR_BIDI_CONTROL=2,
00188     UCHAR_BIDI_MIRRORED=3,
00190     UCHAR_DASH=4,
00194     UCHAR_DEFAULT_IGNORABLE_CODE_POINT=5,
00197     UCHAR_DEPRECATED=6,
00200     UCHAR_DIACRITIC=7,
00204     UCHAR_EXTENDER=8,
00208     UCHAR_FULL_COMPOSITION_EXCLUSION=9,
00212     UCHAR_GRAPHEME_BASE=10,
00216     UCHAR_GRAPHEME_EXTEND=11,
00219     UCHAR_GRAPHEME_LINK=12,
00222     UCHAR_HEX_DIGIT=13,
00225     UCHAR_HYPHEN=14,
00230     UCHAR_ID_CONTINUE=15,
00234     UCHAR_ID_START=16,
00237     UCHAR_IDEOGRAPHIC=17,
00241     UCHAR_IDS_BINARY_OPERATOR=18,
00245     UCHAR_IDS_TRINARY_OPERATOR=19,
00248     UCHAR_JOIN_CONTROL=20,
00252     UCHAR_LOGICAL_ORDER_EXCEPTION=21,
00255     UCHAR_LOWERCASE=22,
00257     UCHAR_MATH=23,
00261     UCHAR_NONCHARACTER_CODE_POINT=24,
00263     UCHAR_QUOTATION_MARK=25,
00267     UCHAR_RADICAL=26,
00272     UCHAR_SOFT_DOTTED=27,
00276     UCHAR_TERMINAL_PUNCTUATION=28,
00280     UCHAR_UNIFIED_IDEOGRAPH=29,
00283     UCHAR_UPPERCASE=30,
00287     UCHAR_WHITE_SPACE=31,
00291     UCHAR_XID_CONTINUE=32,
00294     UCHAR_XID_START=33,
00298    UCHAR_CASE_SENSITIVE=34,
00303     UCHAR_S_TERM=35,
00309     UCHAR_VARIATION_SELECTOR=36,
00316     UCHAR_NFD_INERT=37,
00323     UCHAR_NFKD_INERT=38,
00330     UCHAR_NFC_INERT=39,
00337     UCHAR_NFKC_INERT=40,
00348     UCHAR_SEGMENT_STARTER=41,
00353     UCHAR_PATTERN_SYNTAX=42,
00358     UCHAR_PATTERN_WHITE_SPACE=43,
00363     UCHAR_POSIX_ALNUM=44,
00368     UCHAR_POSIX_BLANK=45,
00373     UCHAR_POSIX_GRAPH=46,
00378     UCHAR_POSIX_PRINT=47,
00383     UCHAR_POSIX_XDIGIT=48,
00385     UCHAR_CASED=49,
00387     UCHAR_CASE_IGNORABLE=50,
00389     UCHAR_CHANGES_WHEN_LOWERCASED=51,
00391     UCHAR_CHANGES_WHEN_UPPERCASED=52,
00393     UCHAR_CHANGES_WHEN_TITLECASED=53,
00395     UCHAR_CHANGES_WHEN_CASEFOLDED=54,
00397     UCHAR_CHANGES_WHEN_CASEMAPPED=55,
00399     UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED=56,
00400 #ifndef U_HIDE_DRAFT_API
00401 
00407     UCHAR_EMOJI=57,
00414     UCHAR_EMOJI_PRESENTATION=58,
00421     UCHAR_EMOJI_MODIFIER=59,
00428     UCHAR_EMOJI_MODIFIER_BASE=60,
00429 #endif /* U_HIDE_DRAFT_API */
00430 
00431     UCHAR_BINARY_LIMIT=61,
00432 
00435     UCHAR_BIDI_CLASS=0x1000,
00437     UCHAR_INT_START=UCHAR_BIDI_CLASS,
00440     UCHAR_BLOCK=0x1001,
00443     UCHAR_CANONICAL_COMBINING_CLASS=0x1002,
00446     UCHAR_DECOMPOSITION_TYPE=0x1003,
00450     UCHAR_EAST_ASIAN_WIDTH=0x1004,
00453     UCHAR_GENERAL_CATEGORY=0x1005,
00456     UCHAR_JOINING_GROUP=0x1006,
00459     UCHAR_JOINING_TYPE=0x1007,
00462     UCHAR_LINE_BREAK=0x1008,
00465     UCHAR_NUMERIC_TYPE=0x1009,
00468     UCHAR_SCRIPT=0x100A,
00471     UCHAR_HANGUL_SYLLABLE_TYPE=0x100B,
00474     UCHAR_NFD_QUICK_CHECK=0x100C,
00477     UCHAR_NFKD_QUICK_CHECK=0x100D,
00480     UCHAR_NFC_QUICK_CHECK=0x100E,
00483     UCHAR_NFKC_QUICK_CHECK=0x100F,
00490     UCHAR_LEAD_CANONICAL_COMBINING_CLASS=0x1010,
00497     UCHAR_TRAIL_CANONICAL_COMBINING_CLASS=0x1011,
00502     UCHAR_GRAPHEME_CLUSTER_BREAK=0x1012,
00507     UCHAR_SENTENCE_BREAK=0x1013,
00512     UCHAR_WORD_BREAK=0x1014,
00517     UCHAR_BIDI_PAIRED_BRACKET_TYPE=0x1015,
00519     UCHAR_INT_LIMIT=0x1016,
00520 
00529     UCHAR_GENERAL_CATEGORY_MASK=0x2000,
00531     UCHAR_MASK_START=UCHAR_GENERAL_CATEGORY_MASK,
00533     UCHAR_MASK_LIMIT=0x2001,
00534 
00537     UCHAR_NUMERIC_VALUE=0x3000,
00539     UCHAR_DOUBLE_START=UCHAR_NUMERIC_VALUE,
00541     UCHAR_DOUBLE_LIMIT=0x3001,
00542 
00545     UCHAR_AGE=0x4000,
00547     UCHAR_STRING_START=UCHAR_AGE,
00550     UCHAR_BIDI_MIRRORING_GLYPH=0x4001,
00553     UCHAR_CASE_FOLDING=0x4002,
00554 #ifndef U_HIDE_DEPRECATED_API
00555 
00557     UCHAR_ISO_COMMENT=0x4003,
00558 #endif  /* U_HIDE_DEPRECATED_API */
00559 
00561     UCHAR_LOWERCASE_MAPPING=0x4004,
00564     UCHAR_NAME=0x4005,
00567     UCHAR_SIMPLE_CASE_FOLDING=0x4006,
00570     UCHAR_SIMPLE_LOWERCASE_MAPPING=0x4007,
00573     UCHAR_SIMPLE_TITLECASE_MAPPING=0x4008,
00576     UCHAR_SIMPLE_UPPERCASE_MAPPING=0x4009,
00579     UCHAR_TITLECASE_MAPPING=0x400A,
00580 #ifndef U_HIDE_DEPRECATED_API
00581 
00585     UCHAR_UNICODE_1_NAME=0x400B,
00586 #endif  /* U_HIDE_DEPRECATED_API */
00587 
00589     UCHAR_UPPERCASE_MAPPING=0x400C,
00592     UCHAR_BIDI_PAIRED_BRACKET=0x400D,
00594     UCHAR_STRING_LIMIT=0x400E,
00595 
00601     UCHAR_SCRIPT_EXTENSIONS=0x7000,
00603     UCHAR_OTHER_PROPERTY_START=UCHAR_SCRIPT_EXTENSIONS,
00606     UCHAR_OTHER_PROPERTY_LIMIT=0x7001,
00608     UCHAR_INVALID_CODE = -1
00609 } UProperty;
00610 
00616 typedef enum UCharCategory
00617 {
00618     /*
00619      * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
00620      * It matches pairs of lines like
00621      *     / ** <Unicode 2-letter General_Category value> comment... * /
00622      *     U_<[A-Z_]+> = <integer>,
00623      */
00624 
00626     U_UNASSIGNED              = 0,
00628     U_GENERAL_OTHER_TYPES     = 0,
00630     U_UPPERCASE_LETTER        = 1,
00632     U_LOWERCASE_LETTER        = 2,
00634     U_TITLECASE_LETTER        = 3,
00636     U_MODIFIER_LETTER         = 4,
00638     U_OTHER_LETTER            = 5,
00640     U_NON_SPACING_MARK        = 6,
00642     U_ENCLOSING_MARK          = 7,
00644     U_COMBINING_SPACING_MARK  = 8,
00646     U_DECIMAL_DIGIT_NUMBER    = 9,
00648     U_LETTER_NUMBER           = 10,
00650     U_OTHER_NUMBER            = 11,
00652     U_SPACE_SEPARATOR         = 12,
00654     U_LINE_SEPARATOR          = 13,
00656     U_PARAGRAPH_SEPARATOR     = 14,
00658     U_CONTROL_CHAR            = 15,
00660     U_FORMAT_CHAR             = 16,
00662     U_PRIVATE_USE_CHAR        = 17,
00664     U_SURROGATE               = 18,
00666     U_DASH_PUNCTUATION        = 19,
00668     U_START_PUNCTUATION       = 20,
00670     U_END_PUNCTUATION         = 21,
00672     U_CONNECTOR_PUNCTUATION   = 22,
00674     U_OTHER_PUNCTUATION       = 23,
00676     U_MATH_SYMBOL             = 24,
00678     U_CURRENCY_SYMBOL         = 25,
00680     U_MODIFIER_SYMBOL         = 26,
00682     U_OTHER_SYMBOL            = 27,
00684     U_INITIAL_PUNCTUATION     = 28,
00686     U_FINAL_PUNCTUATION       = 29,
00688     U_CHAR_CATEGORY_COUNT
00689 } UCharCategory;
00690 
00705 #define U_GC_CN_MASK    U_MASK(U_GENERAL_OTHER_TYPES)
00706 
00708 #define U_GC_LU_MASK    U_MASK(U_UPPERCASE_LETTER)
00709 
00710 #define U_GC_LL_MASK    U_MASK(U_LOWERCASE_LETTER)
00711 
00712 #define U_GC_LT_MASK    U_MASK(U_TITLECASE_LETTER)
00713 
00714 #define U_GC_LM_MASK    U_MASK(U_MODIFIER_LETTER)
00715 
00716 #define U_GC_LO_MASK    U_MASK(U_OTHER_LETTER)
00717 
00719 #define U_GC_MN_MASK    U_MASK(U_NON_SPACING_MARK)
00720 
00721 #define U_GC_ME_MASK    U_MASK(U_ENCLOSING_MARK)
00722 
00723 #define U_GC_MC_MASK    U_MASK(U_COMBINING_SPACING_MARK)
00724 
00726 #define U_GC_ND_MASK    U_MASK(U_DECIMAL_DIGIT_NUMBER)
00727 
00728 #define U_GC_NL_MASK    U_MASK(U_LETTER_NUMBER)
00729 
00730 #define U_GC_NO_MASK    U_MASK(U_OTHER_NUMBER)
00731 
00733 #define U_GC_ZS_MASK    U_MASK(U_SPACE_SEPARATOR)
00734 
00735 #define U_GC_ZL_MASK    U_MASK(U_LINE_SEPARATOR)
00736 
00737 #define U_GC_ZP_MASK    U_MASK(U_PARAGRAPH_SEPARATOR)
00738 
00740 #define U_GC_CC_MASK    U_MASK(U_CONTROL_CHAR)
00741 
00742 #define U_GC_CF_MASK    U_MASK(U_FORMAT_CHAR)
00743 
00744 #define U_GC_CO_MASK    U_MASK(U_PRIVATE_USE_CHAR)
00745 
00746 #define U_GC_CS_MASK    U_MASK(U_SURROGATE)
00747 
00749 #define U_GC_PD_MASK    U_MASK(U_DASH_PUNCTUATION)
00750 
00751 #define U_GC_PS_MASK    U_MASK(U_START_PUNCTUATION)
00752 
00753 #define U_GC_PE_MASK    U_MASK(U_END_PUNCTUATION)
00754 
00755 #define U_GC_PC_MASK    U_MASK(U_CONNECTOR_PUNCTUATION)
00756 
00757 #define U_GC_PO_MASK    U_MASK(U_OTHER_PUNCTUATION)
00758 
00760 #define U_GC_SM_MASK    U_MASK(U_MATH_SYMBOL)
00761 
00762 #define U_GC_SC_MASK    U_MASK(U_CURRENCY_SYMBOL)
00763 
00764 #define U_GC_SK_MASK    U_MASK(U_MODIFIER_SYMBOL)
00765 
00766 #define U_GC_SO_MASK    U_MASK(U_OTHER_SYMBOL)
00767 
00769 #define U_GC_PI_MASK    U_MASK(U_INITIAL_PUNCTUATION)
00770 
00771 #define U_GC_PF_MASK    U_MASK(U_FINAL_PUNCTUATION)
00772 
00773 
00775 #define U_GC_L_MASK \
00776             (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
00777 
00779 #define U_GC_LC_MASK \
00780             (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
00781 
00783 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
00784 
00786 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
00787 
00789 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
00790 
00792 #define U_GC_C_MASK \
00793             (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
00794 
00796 #define U_GC_P_MASK \
00797             (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
00798              U_GC_PI_MASK|U_GC_PF_MASK)
00799 
00801 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
00802 
00807 typedef enum UCharDirection {
00808     /*
00809      * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
00810      * It matches pairs of lines like
00811      *     / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
00812      *     U_<[A-Z_]+> = <integer>,
00813      */
00814 
00816     U_LEFT_TO_RIGHT               = 0,
00818     U_RIGHT_TO_LEFT               = 1,
00820     U_EUROPEAN_NUMBER             = 2,
00822     U_EUROPEAN_NUMBER_SEPARATOR   = 3,
00824     U_EUROPEAN_NUMBER_TERMINATOR  = 4,
00826     U_ARABIC_NUMBER               = 5,
00828     U_COMMON_NUMBER_SEPARATOR     = 6,
00830     U_BLOCK_SEPARATOR             = 7,
00832     U_SEGMENT_SEPARATOR           = 8,
00834     U_WHITE_SPACE_NEUTRAL         = 9,
00836     U_OTHER_NEUTRAL               = 10,
00838     U_LEFT_TO_RIGHT_EMBEDDING     = 11,
00840     U_LEFT_TO_RIGHT_OVERRIDE      = 12,
00842     U_RIGHT_TO_LEFT_ARABIC        = 13,
00844     U_RIGHT_TO_LEFT_EMBEDDING     = 14,
00846     U_RIGHT_TO_LEFT_OVERRIDE      = 15,
00848     U_POP_DIRECTIONAL_FORMAT      = 16,
00850     U_DIR_NON_SPACING_MARK        = 17,
00852     U_BOUNDARY_NEUTRAL            = 18,
00854     U_FIRST_STRONG_ISOLATE        = 19,
00856     U_LEFT_TO_RIGHT_ISOLATE       = 20,
00858     U_RIGHT_TO_LEFT_ISOLATE       = 21,
00860     U_POP_DIRECTIONAL_ISOLATE     = 22,
00862     U_CHAR_DIRECTION_COUNT
00863 } UCharDirection;
00864 
00871 typedef enum UBidiPairedBracketType {
00872     /*
00873      * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
00874      * It matches lines like
00875      *     U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
00876      */
00877 
00879     U_BPT_NONE,
00881     U_BPT_OPEN,
00883     U_BPT_CLOSE,
00885     U_BPT_COUNT /* 3 */
00886 } UBidiPairedBracketType;
00887 
00892 enum UBlockCode {
00893     /*
00894      * Note: UBlockCode constants are parsed by preparseucd.py.
00895      * It matches lines like
00896      *     UBLOCK_<Unicode Block value name> = <integer>,
00897      */
00898 
00900     UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
00901 
00903     UBLOCK_BASIC_LATIN = 1, /*[0000]*/
00904 
00906     UBLOCK_LATIN_1_SUPPLEMENT=2, /*[0080]*/
00907 
00909     UBLOCK_LATIN_EXTENDED_A =3, /*[0100]*/
00910 
00912     UBLOCK_LATIN_EXTENDED_B =4, /*[0180]*/
00913 
00915     UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
00916 
00918     UBLOCK_SPACING_MODIFIER_LETTERS =6, /*[02B0]*/
00919 
00921     UBLOCK_COMBINING_DIACRITICAL_MARKS =7, /*[0300]*/
00922 
00927     UBLOCK_GREEK =8, /*[0370]*/
00928 
00930     UBLOCK_CYRILLIC =9, /*[0400]*/
00931 
00933     UBLOCK_ARMENIAN =10, /*[0530]*/
00934 
00936     UBLOCK_HEBREW =11, /*[0590]*/
00937 
00939     UBLOCK_ARABIC =12, /*[0600]*/
00940 
00942     UBLOCK_SYRIAC =13, /*[0700]*/
00943 
00945     UBLOCK_THAANA =14, /*[0780]*/
00946 
00948     UBLOCK_DEVANAGARI =15, /*[0900]*/
00949 
00951     UBLOCK_BENGALI =16, /*[0980]*/
00952 
00954     UBLOCK_GURMUKHI =17, /*[0A00]*/
00955 
00957     UBLOCK_GUJARATI =18, /*[0A80]*/
00958 
00960     UBLOCK_ORIYA =19, /*[0B00]*/
00961 
00963     UBLOCK_TAMIL =20, /*[0B80]*/
00964 
00966     UBLOCK_TELUGU =21, /*[0C00]*/
00967 
00969     UBLOCK_KANNADA =22, /*[0C80]*/
00970 
00972     UBLOCK_MALAYALAM =23, /*[0D00]*/
00973 
00975     UBLOCK_SINHALA =24, /*[0D80]*/
00976 
00978     UBLOCK_THAI =25, /*[0E00]*/
00979 
00981     UBLOCK_LAO =26, /*[0E80]*/
00982 
00984     UBLOCK_TIBETAN =27, /*[0F00]*/
00985 
00987     UBLOCK_MYANMAR =28, /*[1000]*/
00988 
00990     UBLOCK_GEORGIAN =29, /*[10A0]*/
00991 
00993     UBLOCK_HANGUL_JAMO =30, /*[1100]*/
00994 
00996     UBLOCK_ETHIOPIC =31, /*[1200]*/
00997 
00999     UBLOCK_CHEROKEE =32, /*[13A0]*/
01000 
01002     UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS =33, /*[1400]*/
01003 
01005     UBLOCK_OGHAM =34, /*[1680]*/
01006 
01008     UBLOCK_RUNIC =35, /*[16A0]*/
01009 
01011     UBLOCK_KHMER =36, /*[1780]*/
01012 
01014     UBLOCK_MONGOLIAN =37, /*[1800]*/
01015 
01017     UBLOCK_LATIN_EXTENDED_ADDITIONAL =38, /*[1E00]*/
01018 
01020     UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
01021 
01023     UBLOCK_GENERAL_PUNCTUATION =40, /*[2000]*/
01024 
01026     UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS =41, /*[2070]*/
01027 
01029     UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
01030 
01035     UBLOCK_COMBINING_MARKS_FOR_SYMBOLS =43, /*[20D0]*/
01036 
01038     UBLOCK_LETTERLIKE_SYMBOLS =44, /*[2100]*/
01039 
01041     UBLOCK_NUMBER_FORMS =45, /*[2150]*/
01042 
01044     UBLOCK_ARROWS =46, /*[2190]*/
01045 
01047     UBLOCK_MATHEMATICAL_OPERATORS =47, /*[2200]*/
01048 
01050     UBLOCK_MISCELLANEOUS_TECHNICAL =48, /*[2300]*/
01051 
01053     UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
01054 
01056     UBLOCK_OPTICAL_CHARACTER_RECOGNITION =50, /*[2440]*/
01057 
01059     UBLOCK_ENCLOSED_ALPHANUMERICS =51, /*[2460]*/
01060 
01062     UBLOCK_BOX_DRAWING =52, /*[2500]*/
01063 
01065     UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
01066 
01068     UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
01069 
01071     UBLOCK_MISCELLANEOUS_SYMBOLS =55, /*[2600]*/
01072 
01074     UBLOCK_DINGBATS =56, /*[2700]*/
01075 
01077     UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
01078 
01080     UBLOCK_CJK_RADICALS_SUPPLEMENT =58, /*[2E80]*/
01081 
01083     UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
01084 
01086     UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS =60, /*[2FF0]*/
01087 
01089     UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION =61, /*[3000]*/
01090 
01092     UBLOCK_HIRAGANA =62, /*[3040]*/
01093 
01095     UBLOCK_KATAKANA =63, /*[30A0]*/
01096 
01098     UBLOCK_BOPOMOFO =64, /*[3100]*/
01099 
01101     UBLOCK_HANGUL_COMPATIBILITY_JAMO =65, /*[3130]*/
01102 
01104     UBLOCK_KANBUN =66, /*[3190]*/
01105 
01107     UBLOCK_BOPOMOFO_EXTENDED =67, /*[31A0]*/
01108 
01110     UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS =68, /*[3200]*/
01111 
01113     UBLOCK_CJK_COMPATIBILITY =69, /*[3300]*/
01114 
01116     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A =70, /*[3400]*/
01117 
01119     UBLOCK_CJK_UNIFIED_IDEOGRAPHS =71, /*[4E00]*/
01120 
01122     UBLOCK_YI_SYLLABLES =72, /*[A000]*/
01123 
01125     UBLOCK_YI_RADICALS =73, /*[A490]*/
01126 
01128     UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
01129 
01131     UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
01132 
01134     UBLOCK_HIGH_PRIVATE_USE_SURROGATES =76, /*[DB80]*/
01135 
01137     UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
01138 
01148     UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
01158     UBLOCK_PRIVATE_USE = UBLOCK_PRIVATE_USE_AREA,
01159 
01161     UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS =79, /*[F900]*/
01162 
01164     UBLOCK_ALPHABETIC_PRESENTATION_FORMS =80, /*[FB00]*/
01165 
01167     UBLOCK_ARABIC_PRESENTATION_FORMS_A =81, /*[FB50]*/
01168 
01170     UBLOCK_COMBINING_HALF_MARKS =82, /*[FE20]*/
01171 
01173     UBLOCK_CJK_COMPATIBILITY_FORMS =83, /*[FE30]*/
01174 
01176     UBLOCK_SMALL_FORM_VARIANTS =84, /*[FE50]*/
01177 
01179     UBLOCK_ARABIC_PRESENTATION_FORMS_B =85, /*[FE70]*/
01180 
01182     UBLOCK_SPECIALS =86, /*[FFF0]*/
01183 
01185     UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS =87, /*[FF00]*/
01186 
01187     /* New blocks in Unicode 3.1 */
01188 
01190     UBLOCK_OLD_ITALIC = 88, /*[10300]*/
01192     UBLOCK_GOTHIC = 89, /*[10330]*/
01194     UBLOCK_DESERET = 90, /*[10400]*/
01196     UBLOCK_BYZANTINE_MUSICAL_SYMBOLS = 91, /*[1D000]*/
01198     UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
01200     UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS = 93, /*[1D400]*/
01202     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B  = 94, /*[20000]*/
01204     UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT = 95, /*[2F800]*/
01206     UBLOCK_TAGS = 96, /*[E0000]*/
01207 
01208     /* New blocks in Unicode 3.2 */
01209 
01211     UBLOCK_CYRILLIC_SUPPLEMENT = 97, /*[0500]*/
01216     UBLOCK_CYRILLIC_SUPPLEMENTARY = UBLOCK_CYRILLIC_SUPPLEMENT,
01218     UBLOCK_TAGALOG = 98, /*[1700]*/
01220     UBLOCK_HANUNOO = 99, /*[1720]*/
01222     UBLOCK_BUHID = 100, /*[1740]*/
01224     UBLOCK_TAGBANWA = 101, /*[1760]*/
01226     UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A = 102, /*[27C0]*/
01228     UBLOCK_SUPPLEMENTAL_ARROWS_A = 103, /*[27F0]*/
01230     UBLOCK_SUPPLEMENTAL_ARROWS_B = 104, /*[2900]*/
01232     UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B = 105, /*[2980]*/
01234     UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS = 106, /*[2A00]*/
01236     UBLOCK_KATAKANA_PHONETIC_EXTENSIONS = 107, /*[31F0]*/
01238     UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
01240     UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A = 109, /*[F0000]*/
01242     UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B = 110, /*[100000]*/
01243 
01244     /* New blocks in Unicode 4 */
01245 
01247     UBLOCK_LIMBU = 111, /*[1900]*/
01249     UBLOCK_TAI_LE = 112, /*[1950]*/
01251     UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
01253     UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
01255     UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS = 115, /*[2B00]*/
01257     UBLOCK_YIJING_HEXAGRAM_SYMBOLS = 116, /*[4DC0]*/
01259     UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
01261     UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
01263     UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
01265     UBLOCK_UGARITIC = 120, /*[10380]*/
01267     UBLOCK_SHAVIAN = 121, /*[10450]*/
01269     UBLOCK_OSMANYA = 122, /*[10480]*/
01271     UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
01273     UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
01275     UBLOCK_VARIATION_SELECTORS_SUPPLEMENT = 125, /*[E0100]*/
01276 
01277     /* New blocks in Unicode 4.1 */
01278 
01280     UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION = 126, /*[1D200]*/
01282     UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
01284     UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
01286     UBLOCK_BUGINESE = 129, /*[1A00]*/
01288     UBLOCK_CJK_STROKES = 130, /*[31C0]*/
01290     UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT = 131, /*[1DC0]*/
01292     UBLOCK_COPTIC = 132, /*[2C80]*/
01294     UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
01296     UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
01298     UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
01300     UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
01302     UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
01304     UBLOCK_MODIFIER_TONE_LETTERS = 138, /*[A700]*/
01306     UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
01308     UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
01310     UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT = 141, /*[1D80]*/
01312     UBLOCK_SUPPLEMENTAL_PUNCTUATION = 142, /*[2E00]*/
01314     UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
01316     UBLOCK_TIFINAGH = 144, /*[2D30]*/
01318     UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
01319 
01320     /* New blocks in Unicode 5.0 */
01321 
01323     UBLOCK_NKO = 146, /*[07C0]*/
01325     UBLOCK_BALINESE = 147, /*[1B00]*/
01327     UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
01329     UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
01331     UBLOCK_PHAGS_PA = 150, /*[A840]*/
01333     UBLOCK_PHOENICIAN = 151, /*[10900]*/
01335     UBLOCK_CUNEIFORM = 152, /*[12000]*/
01337     UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION = 153, /*[12400]*/
01339     UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
01340 
01341     /* New blocks in Unicode 5.1 */
01342 
01344     UBLOCK_SUNDANESE = 155, /*[1B80]*/
01346     UBLOCK_LEPCHA = 156, /*[1C00]*/
01348     UBLOCK_OL_CHIKI = 157, /*[1C50]*/
01350     UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
01352     UBLOCK_VAI = 159, /*[A500]*/
01354     UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
01356     UBLOCK_SAURASHTRA = 161, /*[A880]*/
01358     UBLOCK_KAYAH_LI = 162, /*[A900]*/
01360     UBLOCK_REJANG = 163, /*[A930]*/
01362     UBLOCK_CHAM = 164, /*[AA00]*/
01364     UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
01366     UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
01368     UBLOCK_LYCIAN = 167, /*[10280]*/
01370     UBLOCK_CARIAN = 168, /*[102A0]*/
01372     UBLOCK_LYDIAN = 169, /*[10920]*/
01374     UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
01376     UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
01377 
01378     /* New blocks in Unicode 5.2 */
01379 
01381     UBLOCK_SAMARITAN = 172, /*[0800]*/
01383     UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED = 173, /*[18B0]*/
01385     UBLOCK_TAI_THAM = 174, /*[1A20]*/
01387     UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
01389     UBLOCK_LISU = 176, /*[A4D0]*/
01391     UBLOCK_BAMUM = 177, /*[A6A0]*/
01393     UBLOCK_COMMON_INDIC_NUMBER_FORMS = 178, /*[A830]*/
01395     UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
01397     UBLOCK_HANGUL_JAMO_EXTENDED_A = 180, /*[A960]*/
01399     UBLOCK_JAVANESE = 181, /*[A980]*/
01401     UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
01403     UBLOCK_TAI_VIET = 183, /*[AA80]*/
01405     UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
01407     UBLOCK_HANGUL_JAMO_EXTENDED_B = 185, /*[D7B0]*/
01409     UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
01411     UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
01413     UBLOCK_AVESTAN = 188, /*[10B00]*/
01415     UBLOCK_INSCRIPTIONAL_PARTHIAN = 189, /*[10B40]*/
01417     UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
01419     UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
01421     UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
01423     UBLOCK_KAITHI = 193, /*[11080]*/
01425     UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
01427     UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT = 195, /*[1F100]*/
01429     UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT = 196, /*[1F200]*/
01431     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C = 197, /*[2A700]*/
01432 
01433     /* New blocks in Unicode 6.0 */
01434 
01436     UBLOCK_MANDAIC = 198, /*[0840]*/
01438     UBLOCK_BATAK = 199, /*[1BC0]*/
01440     UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
01442     UBLOCK_BRAHMI = 201, /*[11000]*/
01444     UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
01446     UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
01448     UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
01450     UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS = 205, /*[1F300]*/
01452     UBLOCK_EMOTICONS = 206, /*[1F600]*/
01454     UBLOCK_TRANSPORT_AND_MAP_SYMBOLS = 207, /*[1F680]*/
01456     UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
01458     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D = 209, /*[2B740]*/
01459 
01460     /* New blocks in Unicode 6.1 */
01461 
01463     UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
01465     UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS = 211, /*[1EE00]*/
01467     UBLOCK_CHAKMA = 212, /*[11100]*/
01469     UBLOCK_MEETEI_MAYEK_EXTENSIONS = 213, /*[AAE0]*/
01471     UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
01473     UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
01475     UBLOCK_MIAO = 216, /*[16F00]*/
01477     UBLOCK_SHARADA = 217, /*[11180]*/
01479     UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
01481     UBLOCK_SUNDANESE_SUPPLEMENT = 219, /*[1CC0]*/
01483     UBLOCK_TAKRI = 220, /*[11680]*/
01484 
01485     /* New blocks in Unicode 7.0 */
01486 
01488     UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
01490     UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
01492     UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
01494     UBLOCK_COMBINING_DIACRITICAL_MARKS_EXTENDED = 224, /*[1AB0]*/
01496     UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
01498     UBLOCK_ELBASAN = 226, /*[10500]*/
01500     UBLOCK_GEOMETRIC_SHAPES_EXTENDED = 227, /*[1F780]*/
01502     UBLOCK_GRANTHA = 228, /*[11300]*/
01504     UBLOCK_KHOJKI = 229, /*[11200]*/
01506     UBLOCK_KHUDAWADI = 230, /*[112B0]*/
01508     UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
01510     UBLOCK_LINEAR_A = 232, /*[10600]*/
01512     UBLOCK_MAHAJANI = 233, /*[11150]*/
01514     UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
01516     UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
01518     UBLOCK_MODI = 236, /*[11600]*/
01520     UBLOCK_MRO = 237, /*[16A40]*/
01522     UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
01524     UBLOCK_NABATAEAN = 239, /*[10880]*/
01526     UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
01528     UBLOCK_OLD_PERMIC = 241, /*[10350]*/
01530     UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
01532     UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
01534     UBLOCK_PALMYRENE = 244, /*[10860]*/
01536     UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
01538     UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
01540     UBLOCK_SHORTHAND_FORMAT_CONTROLS = 247, /*[1BCA0]*/
01542     UBLOCK_SIDDHAM = 248, /*[11580]*/
01544     UBLOCK_SINHALA_ARCHAIC_NUMBERS = 249, /*[111E0]*/
01546     UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
01548     UBLOCK_TIRHUTA = 251, /*[11480]*/
01550     UBLOCK_WARANG_CITI = 252, /*[118A0]*/
01551 
01552     /* New blocks in Unicode 8.0 */
01553 
01555     UBLOCK_AHOM = 253, /*[11700]*/
01557     UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
01559     UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
01561     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E = 256, /*[2B820]*/
01563     UBLOCK_EARLY_DYNASTIC_CUNEIFORM = 257, /*[12480]*/
01565     UBLOCK_HATRAN = 258, /*[108E0]*/
01567     UBLOCK_MULTANI = 259, /*[11280]*/
01569     UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
01571     UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS = 261, /*[1F900]*/
01573     UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/
01574 
01576     UBLOCK_COUNT = 263,
01577 
01579     UBLOCK_INVALID_CODE=-1
01580 };
01581 
01583 typedef enum UBlockCode UBlockCode;
01584 
01592 typedef enum UEastAsianWidth {
01593     /*
01594      * Note: UEastAsianWidth constants are parsed by preparseucd.py.
01595      * It matches lines like
01596      *     U_EA_<Unicode East_Asian_Width value name>
01597      */
01598 
01599     U_EA_NEUTRAL,   /*[N]*/
01600     U_EA_AMBIGUOUS, /*[A]*/
01601     U_EA_HALFWIDTH, /*[H]*/
01602     U_EA_FULLWIDTH, /*[F]*/
01603     U_EA_NARROW,    /*[Na]*/
01604     U_EA_WIDE,      /*[W]*/
01605     U_EA_COUNT
01606 } UEastAsianWidth;
01607 
01619 typedef enum UCharNameChoice {
01621     U_UNICODE_CHAR_NAME,
01622 #ifndef U_HIDE_DEPRECATED_API
01623 
01628     U_UNICODE_10_CHAR_NAME,
01629 #endif  /* U_HIDE_DEPRECATED_API */
01630 
01631     U_EXTENDED_CHAR_NAME = U_UNICODE_CHAR_NAME+2,
01633     U_CHAR_NAME_ALIAS,
01635     U_CHAR_NAME_CHOICE_COUNT
01636 } UCharNameChoice;
01637 
01651 typedef enum UPropertyNameChoice {
01652     U_SHORT_PROPERTY_NAME,
01653     U_LONG_PROPERTY_NAME,
01654     U_PROPERTY_NAME_CHOICE_COUNT
01655 } UPropertyNameChoice;
01656 
01663 typedef enum UDecompositionType {
01664     /*
01665      * Note: UDecompositionType constants are parsed by preparseucd.py.
01666      * It matches lines like
01667      *     U_DT_<Unicode Decomposition_Type value name>
01668      */
01669 
01670     U_DT_NONE,              /*[none]*/
01671     U_DT_CANONICAL,         /*[can]*/
01672     U_DT_COMPAT,            /*[com]*/
01673     U_DT_CIRCLE,            /*[enc]*/
01674     U_DT_FINAL,             /*[fin]*/
01675     U_DT_FONT,              /*[font]*/
01676     U_DT_FRACTION,          /*[fra]*/
01677     U_DT_INITIAL,           /*[init]*/
01678     U_DT_ISOLATED,          /*[iso]*/
01679     U_DT_MEDIAL,            /*[med]*/
01680     U_DT_NARROW,            /*[nar]*/
01681     U_DT_NOBREAK,           /*[nb]*/
01682     U_DT_SMALL,             /*[sml]*/
01683     U_DT_SQUARE,            /*[sqr]*/
01684     U_DT_SUB,               /*[sub]*/
01685     U_DT_SUPER,             /*[sup]*/
01686     U_DT_VERTICAL,          /*[vert]*/
01687     U_DT_WIDE,              /*[wide]*/
01688     U_DT_COUNT /* 18 */
01689 } UDecompositionType;
01690 
01697 typedef enum UJoiningType {
01698     /*
01699      * Note: UJoiningType constants are parsed by preparseucd.py.
01700      * It matches lines like
01701      *     U_JT_<Unicode Joining_Type value name>
01702      */
01703 
01704     U_JT_NON_JOINING,       /*[U]*/
01705     U_JT_JOIN_CAUSING,      /*[C]*/
01706     U_JT_DUAL_JOINING,      /*[D]*/
01707     U_JT_LEFT_JOINING,      /*[L]*/
01708     U_JT_RIGHT_JOINING,     /*[R]*/
01709     U_JT_TRANSPARENT,       /*[T]*/
01710     U_JT_COUNT /* 6 */
01711 } UJoiningType;
01712 
01719 typedef enum UJoiningGroup {
01720     /*
01721      * Note: UJoiningGroup constants are parsed by preparseucd.py.
01722      * It matches lines like
01723      *     U_JG_<Unicode Joining_Group value name>
01724      */
01725 
01726     U_JG_NO_JOINING_GROUP,
01727     U_JG_AIN,
01728     U_JG_ALAPH,
01729     U_JG_ALEF,
01730     U_JG_BEH,
01731     U_JG_BETH,
01732     U_JG_DAL,
01733     U_JG_DALATH_RISH,
01734     U_JG_E,
01735     U_JG_FEH,
01736     U_JG_FINAL_SEMKATH,
01737     U_JG_GAF,
01738     U_JG_GAMAL,
01739     U_JG_HAH,
01740     U_JG_TEH_MARBUTA_GOAL,  
01741     U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
01742     U_JG_HE,
01743     U_JG_HEH,
01744     U_JG_HEH_GOAL,
01745     U_JG_HETH,
01746     U_JG_KAF,
01747     U_JG_KAPH,
01748     U_JG_KNOTTED_HEH,
01749     U_JG_LAM,
01750     U_JG_LAMADH,
01751     U_JG_MEEM,
01752     U_JG_MIM,
01753     U_JG_NOON,
01754     U_JG_NUN,
01755     U_JG_PE,
01756     U_JG_QAF,
01757     U_JG_QAPH,
01758     U_JG_REH,
01759     U_JG_REVERSED_PE,
01760     U_JG_SAD,
01761     U_JG_SADHE,
01762     U_JG_SEEN,
01763     U_JG_SEMKATH,
01764     U_JG_SHIN,
01765     U_JG_SWASH_KAF,
01766     U_JG_SYRIAC_WAW,
01767     U_JG_TAH,
01768     U_JG_TAW,
01769     U_JG_TEH_MARBUTA,
01770     U_JG_TETH,
01771     U_JG_WAW,
01772     U_JG_YEH,
01773     U_JG_YEH_BARREE,
01774     U_JG_YEH_WITH_TAIL,
01775     U_JG_YUDH,
01776     U_JG_YUDH_HE,
01777     U_JG_ZAIN,
01778     U_JG_FE,        
01779     U_JG_KHAPH,     
01780     U_JG_ZHAIN,     
01781     U_JG_BURUSHASKI_YEH_BARREE, 
01782     U_JG_FARSI_YEH, 
01783     U_JG_NYA,       
01784     U_JG_ROHINGYA_YEH,  
01785     U_JG_MANICHAEAN_ALEPH,  
01786     U_JG_MANICHAEAN_AYIN,  
01787     U_JG_MANICHAEAN_BETH,  
01788     U_JG_MANICHAEAN_DALETH,  
01789     U_JG_MANICHAEAN_DHAMEDH,  
01790     U_JG_MANICHAEAN_FIVE,  
01791     U_JG_MANICHAEAN_GIMEL,  
01792     U_JG_MANICHAEAN_HETH,  
01793     U_JG_MANICHAEAN_HUNDRED,  
01794     U_JG_MANICHAEAN_KAPH,  
01795     U_JG_MANICHAEAN_LAMEDH,  
01796     U_JG_MANICHAEAN_MEM,  
01797     U_JG_MANICHAEAN_NUN,  
01798     U_JG_MANICHAEAN_ONE,  
01799     U_JG_MANICHAEAN_PE,  
01800     U_JG_MANICHAEAN_QOPH,  
01801     U_JG_MANICHAEAN_RESH,  
01802     U_JG_MANICHAEAN_SADHE,  
01803     U_JG_MANICHAEAN_SAMEKH,  
01804     U_JG_MANICHAEAN_TAW,  
01805     U_JG_MANICHAEAN_TEN,  
01806     U_JG_MANICHAEAN_TETH,  
01807     U_JG_MANICHAEAN_THAMEDH,  
01808     U_JG_MANICHAEAN_TWENTY,  
01809     U_JG_MANICHAEAN_WAW,  
01810     U_JG_MANICHAEAN_YODH,  
01811     U_JG_MANICHAEAN_ZAYIN,  
01812     U_JG_STRAIGHT_WAW,  
01813     U_JG_COUNT
01814 } UJoiningGroup;
01815 
01822 typedef enum UGraphemeClusterBreak {
01823     /*
01824      * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
01825      * It matches lines like
01826      *     U_GCB_<Unicode Grapheme_Cluster_Break value name>
01827      */
01828 
01829     U_GCB_OTHER = 0,            /*[XX]*/
01830     U_GCB_CONTROL = 1,          /*[CN]*/
01831     U_GCB_CR = 2,               /*[CR]*/
01832     U_GCB_EXTEND = 3,           /*[EX]*/
01833     U_GCB_L = 4,                /*[L]*/
01834     U_GCB_LF = 5,               /*[LF]*/
01835     U_GCB_LV = 6,               /*[LV]*/
01836     U_GCB_LVT = 7,              /*[LVT]*/
01837     U_GCB_T = 8,                /*[T]*/
01838     U_GCB_V = 9,                /*[V]*/
01839     U_GCB_SPACING_MARK = 10,    /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
01840     U_GCB_PREPEND = 11,         /*[PP]*/
01841     U_GCB_REGIONAL_INDICATOR = 12,  /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
01842     U_GCB_COUNT = 13
01843 } UGraphemeClusterBreak;
01844 
01852 typedef enum UWordBreakValues {
01853     /*
01854      * Note: UWordBreakValues constants are parsed by preparseucd.py.
01855      * It matches lines like
01856      *     U_WB_<Unicode Word_Break value name>
01857      */
01858 
01859     U_WB_OTHER = 0,             /*[XX]*/
01860     U_WB_ALETTER = 1,           /*[LE]*/
01861     U_WB_FORMAT = 2,            /*[FO]*/
01862     U_WB_KATAKANA = 3,          /*[KA]*/
01863     U_WB_MIDLETTER = 4,         /*[ML]*/
01864     U_WB_MIDNUM = 5,            /*[MN]*/
01865     U_WB_NUMERIC = 6,           /*[NU]*/
01866     U_WB_EXTENDNUMLET = 7,      /*[EX]*/
01867     U_WB_CR = 8,                /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
01868     U_WB_EXTEND = 9,            /*[Extend]*/
01869     U_WB_LF = 10,               /*[LF]*/
01870     U_WB_MIDNUMLET =11,         /*[MB]*/
01871     U_WB_NEWLINE =12,           /*[NL]*/
01872     U_WB_REGIONAL_INDICATOR = 13,   /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
01873     U_WB_HEBREW_LETTER = 14,    /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
01874     U_WB_SINGLE_QUOTE = 15,     /*[SQ]*/
01875     U_WB_DOUBLE_QUOTE = 16,     /*[DQ]*/
01876     U_WB_COUNT = 17
01877 } UWordBreakValues;
01878 
01885 typedef enum USentenceBreak {
01886     /*
01887      * Note: USentenceBreak constants are parsed by preparseucd.py.
01888      * It matches lines like
01889      *     U_SB_<Unicode Sentence_Break value name>
01890      */
01891 
01892     U_SB_OTHER = 0,             /*[XX]*/
01893     U_SB_ATERM = 1,             /*[AT]*/
01894     U_SB_CLOSE = 2,             /*[CL]*/
01895     U_SB_FORMAT = 3,            /*[FO]*/
01896     U_SB_LOWER = 4,             /*[LO]*/
01897     U_SB_NUMERIC = 5,           /*[NU]*/
01898     U_SB_OLETTER = 6,           /*[LE]*/
01899     U_SB_SEP = 7,               /*[SE]*/
01900     U_SB_SP = 8,                /*[SP]*/
01901     U_SB_STERM = 9,             /*[ST]*/
01902     U_SB_UPPER = 10,            /*[UP]*/
01903     U_SB_CR = 11,               /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
01904     U_SB_EXTEND = 12,           /*[EX]*/
01905     U_SB_LF = 13,               /*[LF]*/
01906     U_SB_SCONTINUE = 14,        /*[SC]*/
01907     U_SB_COUNT = 15
01908 } USentenceBreak;
01909 
01916 typedef enum ULineBreak {
01917     /*
01918      * Note: ULineBreak constants are parsed by preparseucd.py.
01919      * It matches lines like
01920      *     U_LB_<Unicode Line_Break value name>
01921      */
01922 
01923     U_LB_UNKNOWN = 0,           /*[XX]*/
01924     U_LB_AMBIGUOUS = 1,         /*[AI]*/
01925     U_LB_ALPHABETIC = 2,        /*[AL]*/
01926     U_LB_BREAK_BOTH = 3,        /*[B2]*/
01927     U_LB_BREAK_AFTER = 4,       /*[BA]*/
01928     U_LB_BREAK_BEFORE = 5,      /*[BB]*/
01929     U_LB_MANDATORY_BREAK = 6,   /*[BK]*/
01930     U_LB_CONTINGENT_BREAK = 7,  /*[CB]*/
01931     U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
01932     U_LB_COMBINING_MARK = 9,    /*[CM]*/
01933     U_LB_CARRIAGE_RETURN = 10,   /*[CR]*/
01934     U_LB_EXCLAMATION = 11,       /*[EX]*/
01935     U_LB_GLUE = 12,              /*[GL]*/
01936     U_LB_HYPHEN = 13,            /*[HY]*/
01937     U_LB_IDEOGRAPHIC = 14,       /*[ID]*/
01939     U_LB_INSEPARABLE = 15,       /*[IN]*/
01940     U_LB_INSEPERABLE = U_LB_INSEPARABLE,
01941     U_LB_INFIX_NUMERIC = 16,     /*[IS]*/
01942     U_LB_LINE_FEED = 17,         /*[LF]*/
01943     U_LB_NONSTARTER = 18,        /*[NS]*/
01944     U_LB_NUMERIC = 19,           /*[NU]*/
01945     U_LB_OPEN_PUNCTUATION = 20,  /*[OP]*/
01946     U_LB_POSTFIX_NUMERIC = 21,   /*[PO]*/
01947     U_LB_PREFIX_NUMERIC = 22,    /*[PR]*/
01948     U_LB_QUOTATION = 23,         /*[QU]*/
01949     U_LB_COMPLEX_CONTEXT = 24,   /*[SA]*/
01950     U_LB_SURROGATE = 25,         /*[SG]*/
01951     U_LB_SPACE = 26,             /*[SP]*/
01952     U_LB_BREAK_SYMBOLS = 27,     /*[SY]*/
01953     U_LB_ZWSPACE = 28,           /*[ZW]*/
01954     U_LB_NEXT_LINE = 29,         /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
01955     U_LB_WORD_JOINER = 30,       /*[WJ]*/
01956     U_LB_H2 = 31,                /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
01957     U_LB_H3 = 32,                /*[H3]*/
01958     U_LB_JL = 33,                /*[JL]*/
01959     U_LB_JT = 34,                /*[JT]*/
01960     U_LB_JV = 35,                /*[JV]*/
01961     U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
01962     U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
01963     U_LB_HEBREW_LETTER = 38,     /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
01964     U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
01965     U_LB_COUNT = 40
01966 } ULineBreak;
01967 
01974 typedef enum UNumericType {
01975     /*
01976      * Note: UNumericType constants are parsed by preparseucd.py.
01977      * It matches lines like
01978      *     U_NT_<Unicode Numeric_Type value name>
01979      */
01980 
01981     U_NT_NONE,              /*[None]*/
01982     U_NT_DECIMAL,           /*[de]*/
01983     U_NT_DIGIT,             /*[di]*/
01984     U_NT_NUMERIC,           /*[nu]*/
01985     U_NT_COUNT
01986 } UNumericType;
01987 
01994 typedef enum UHangulSyllableType {
01995     /*
01996      * Note: UHangulSyllableType constants are parsed by preparseucd.py.
01997      * It matches lines like
01998      *     U_HST_<Unicode Hangul_Syllable_Type value name>
01999      */
02000 
02001     U_HST_NOT_APPLICABLE,   /*[NA]*/
02002     U_HST_LEADING_JAMO,     /*[L]*/
02003     U_HST_VOWEL_JAMO,       /*[V]*/
02004     U_HST_TRAILING_JAMO,    /*[T]*/
02005     U_HST_LV_SYLLABLE,      /*[LV]*/
02006     U_HST_LVT_SYLLABLE,     /*[LVT]*/
02007     U_HST_COUNT
02008 } UHangulSyllableType;
02009 
02036 U_STABLE UBool U_EXPORT2
02037 u_hasBinaryProperty(UChar32 c, UProperty which);
02038 
02051 U_STABLE UBool U_EXPORT2
02052 u_isUAlphabetic(UChar32 c);
02053 
02066 U_STABLE UBool U_EXPORT2
02067 u_isULowercase(UChar32 c);
02068 
02081 U_STABLE UBool U_EXPORT2
02082 u_isUUppercase(UChar32 c);
02083 
02102 U_STABLE UBool U_EXPORT2
02103 u_isUWhiteSpace(UChar32 c);
02104 
02142 U_STABLE int32_t U_EXPORT2
02143 u_getIntPropertyValue(UChar32 c, UProperty which);
02144 
02163 U_STABLE int32_t U_EXPORT2
02164 u_getIntPropertyMinValue(UProperty which);
02165 
02192 U_STABLE int32_t U_EXPORT2
02193 u_getIntPropertyMaxValue(UProperty which);
02194 
02217 U_STABLE double U_EXPORT2
02218 u_getNumericValue(UChar32 c);
02219 
02227 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
02228 
02252 U_STABLE UBool U_EXPORT2
02253 u_islower(UChar32 c);
02254 
02279 U_STABLE UBool U_EXPORT2
02280 u_isupper(UChar32 c);
02281 
02296 U_STABLE UBool U_EXPORT2
02297 u_istitle(UChar32 c);
02298 
02317 U_STABLE UBool U_EXPORT2
02318 u_isdigit(UChar32 c);
02319 
02338 U_STABLE UBool U_EXPORT2
02339 u_isalpha(UChar32 c);
02340 
02359 U_STABLE UBool U_EXPORT2
02360 u_isalnum(UChar32 c);
02361 
02382 U_STABLE UBool U_EXPORT2
02383 u_isxdigit(UChar32 c);
02384 
02398 U_STABLE UBool U_EXPORT2
02399 u_ispunct(UChar32 c);
02400 
02417 U_STABLE UBool U_EXPORT2
02418 u_isgraph(UChar32 c);
02419 
02446 U_STABLE UBool U_EXPORT2
02447 u_isblank(UChar32 c);
02448 
02471 U_STABLE UBool U_EXPORT2
02472 u_isdefined(UChar32 c);
02473 
02492 U_STABLE UBool U_EXPORT2
02493 u_isspace(UChar32 c);
02494 
02513 U_STABLE UBool U_EXPORT2
02514 u_isJavaSpaceChar(UChar32 c);
02515 
02553 U_STABLE UBool U_EXPORT2
02554 u_isWhitespace(UChar32 c);
02555 
02577 U_STABLE UBool U_EXPORT2
02578 u_iscntrl(UChar32 c);
02579 
02592 U_STABLE UBool U_EXPORT2
02593 u_isISOControl(UChar32 c);
02594 
02610 U_STABLE UBool U_EXPORT2
02611 u_isprint(UChar32 c);
02612 
02631 U_STABLE UBool U_EXPORT2
02632 u_isbase(UChar32 c);
02633 
02650 U_STABLE UCharDirection U_EXPORT2
02651 u_charDirection(UChar32 c);
02652 
02668 U_STABLE UBool U_EXPORT2
02669 u_isMirrored(UChar32 c);
02670 
02690 U_STABLE UChar32 U_EXPORT2
02691 u_charMirror(UChar32 c);
02692 
02709 U_STABLE UChar32 U_EXPORT2
02710 u_getBidiPairedBracket(UChar32 c);
02711 
02723 U_STABLE int8_t U_EXPORT2
02724 u_charType(UChar32 c);
02725 
02739 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
02740 
02758 typedef UBool U_CALLCONV
02759 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
02760 
02780 U_STABLE void U_EXPORT2
02781 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
02782 
02783 #if !UCONFIG_NO_NORMALIZATION
02784 
02792 U_STABLE uint8_t U_EXPORT2
02793 u_getCombiningClass(UChar32 c);
02794 
02795 #endif
02796 
02820 U_STABLE int32_t U_EXPORT2
02821 u_charDigitValue(UChar32 c);
02822 
02832 U_STABLE UBlockCode U_EXPORT2
02833 ublock_getCode(UChar32 c);
02834 
02867 U_STABLE int32_t U_EXPORT2
02868 u_charName(UChar32 code, UCharNameChoice nameChoice,
02869            char *buffer, int32_t bufferLength,
02870            UErrorCode *pErrorCode);
02871 
02872 #ifndef U_HIDE_DEPRECATED_API
02873 
02891 U_DEPRECATED int32_t U_EXPORT2
02892 u_getISOComment(UChar32 c,
02893                 char *dest, int32_t destCapacity,
02894                 UErrorCode *pErrorCode);
02895 #endif  /* U_HIDE_DEPRECATED_API */
02896 
02917 U_STABLE UChar32 U_EXPORT2
02918 u_charFromName(UCharNameChoice nameChoice,
02919                const char *name,
02920                UErrorCode *pErrorCode);
02921 
02939 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
02940                                UChar32 code,
02941                                UCharNameChoice nameChoice,
02942                                const char *name,
02943                                int32_t length);
02944 
02966 U_STABLE void U_EXPORT2
02967 u_enumCharNames(UChar32 start, UChar32 limit,
02968                 UEnumCharNamesFn *fn,
02969                 void *context,
02970                 UCharNameChoice nameChoice,
02971                 UErrorCode *pErrorCode);
02972 
03004 U_STABLE const char* U_EXPORT2
03005 u_getPropertyName(UProperty property,
03006                   UPropertyNameChoice nameChoice);
03007 
03027 U_STABLE UProperty U_EXPORT2
03028 u_getPropertyEnum(const char* alias);
03029 
03077 U_STABLE const char* U_EXPORT2
03078 u_getPropertyValueName(UProperty property,
03079                        int32_t value,
03080                        UPropertyNameChoice nameChoice);
03081 
03113 U_STABLE int32_t U_EXPORT2
03114 u_getPropertyValueEnum(UProperty property,
03115                        const char* alias);
03116 
03134 U_STABLE UBool U_EXPORT2
03135 u_isIDStart(UChar32 c);
03136 
03158 U_STABLE UBool U_EXPORT2
03159 u_isIDPart(UChar32 c);
03160 
03181 U_STABLE UBool U_EXPORT2
03182 u_isIDIgnorable(UChar32 c);
03183 
03200 U_STABLE UBool U_EXPORT2
03201 u_isJavaIDStart(UChar32 c);
03202 
03221 U_STABLE UBool U_EXPORT2
03222 u_isJavaIDPart(UChar32 c);
03223 
03246 U_STABLE UChar32 U_EXPORT2
03247 u_tolower(UChar32 c);
03248 
03271 U_STABLE UChar32 U_EXPORT2
03272 u_toupper(UChar32 c);
03273 
03296 U_STABLE UChar32 U_EXPORT2
03297 u_totitle(UChar32 c);
03298 
03300 #define U_FOLD_CASE_DEFAULT 0
03301 
03318 #define U_FOLD_CASE_EXCLUDE_SPECIAL_I 1
03319 
03342 U_STABLE UChar32 U_EXPORT2
03343 u_foldCase(UChar32 c, uint32_t options);
03344 
03383 U_STABLE int32_t U_EXPORT2
03384 u_digit(UChar32 ch, int8_t radix);
03385 
03414 U_STABLE UChar32 U_EXPORT2
03415 u_forDigit(int32_t digit, int8_t radix);
03416 
03431 U_STABLE void U_EXPORT2
03432 u_charAge(UChar32 c, UVersionInfo versionArray);
03433 
03445 U_STABLE void U_EXPORT2
03446 u_getUnicodeVersion(UVersionInfo versionArray);
03447 
03448 #if !UCONFIG_NO_NORMALIZATION
03449 
03470 U_STABLE int32_t U_EXPORT2
03471 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
03472 
03473 #endif
03474 
03475 
03476 U_CDECL_END
03477 
03478 #endif /*_UCHAR*/
03479 /*eof*/