uchar.h

Go to the documentation of this file.
00001 /*
00002 **********************************************************************
00003 *   Copyright (C) 1997-2012, 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 "6.2"
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,
00401     UCHAR_BINARY_LIMIT=57,
00402 
00405     UCHAR_BIDI_CLASS=0x1000,
00407     UCHAR_INT_START=UCHAR_BIDI_CLASS,
00410     UCHAR_BLOCK=0x1001,
00413     UCHAR_CANONICAL_COMBINING_CLASS=0x1002,
00416     UCHAR_DECOMPOSITION_TYPE=0x1003,
00420     UCHAR_EAST_ASIAN_WIDTH=0x1004,
00423     UCHAR_GENERAL_CATEGORY=0x1005,
00426     UCHAR_JOINING_GROUP=0x1006,
00429     UCHAR_JOINING_TYPE=0x1007,
00432     UCHAR_LINE_BREAK=0x1008,
00435     UCHAR_NUMERIC_TYPE=0x1009,
00438     UCHAR_SCRIPT=0x100A,
00441     UCHAR_HANGUL_SYLLABLE_TYPE=0x100B,
00444     UCHAR_NFD_QUICK_CHECK=0x100C,
00447     UCHAR_NFKD_QUICK_CHECK=0x100D,
00450     UCHAR_NFC_QUICK_CHECK=0x100E,
00453     UCHAR_NFKC_QUICK_CHECK=0x100F,
00460     UCHAR_LEAD_CANONICAL_COMBINING_CLASS=0x1010,
00467     UCHAR_TRAIL_CANONICAL_COMBINING_CLASS=0x1011,
00472     UCHAR_GRAPHEME_CLUSTER_BREAK=0x1012,
00477     UCHAR_SENTENCE_BREAK=0x1013,
00482     UCHAR_WORD_BREAK=0x1014,
00484     UCHAR_INT_LIMIT=0x1015,
00485 
00494     UCHAR_GENERAL_CATEGORY_MASK=0x2000,
00496     UCHAR_MASK_START=UCHAR_GENERAL_CATEGORY_MASK,
00498     UCHAR_MASK_LIMIT=0x2001,
00499 
00502     UCHAR_NUMERIC_VALUE=0x3000,
00504     UCHAR_DOUBLE_START=UCHAR_NUMERIC_VALUE,
00506     UCHAR_DOUBLE_LIMIT=0x3001,
00507 
00510     UCHAR_AGE=0x4000,
00512     UCHAR_STRING_START=UCHAR_AGE,
00515     UCHAR_BIDI_MIRRORING_GLYPH=0x4001,
00518     UCHAR_CASE_FOLDING=0x4002,
00521     UCHAR_ISO_COMMENT=0x4003,
00524     UCHAR_LOWERCASE_MAPPING=0x4004,
00527     UCHAR_NAME=0x4005,
00530     UCHAR_SIMPLE_CASE_FOLDING=0x4006,
00533     UCHAR_SIMPLE_LOWERCASE_MAPPING=0x4007,
00536     UCHAR_SIMPLE_TITLECASE_MAPPING=0x4008,
00539     UCHAR_SIMPLE_UPPERCASE_MAPPING=0x4009,
00542     UCHAR_TITLECASE_MAPPING=0x400A,
00547     UCHAR_UNICODE_1_NAME=0x400B,
00550     UCHAR_UPPERCASE_MAPPING=0x400C,
00552     UCHAR_STRING_LIMIT=0x400D,
00560     UCHAR_SCRIPT_EXTENSIONS=0x7000,
00562     UCHAR_OTHER_PROPERTY_START=UCHAR_SCRIPT_EXTENSIONS,
00565     UCHAR_OTHER_PROPERTY_LIMIT=0x7001,
00567     UCHAR_INVALID_CODE = -1
00568 } UProperty;
00569 
00575 typedef enum UCharCategory
00576 {
00577     /*
00578      * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
00579      * It matches pairs of lines like
00580      *     / ** <Unicode 2-letter General_Category value> comment... * /
00581      *     U_<[A-Z_]+> = <integer>,
00582      */
00583 
00585     U_UNASSIGNED              = 0,
00587     U_GENERAL_OTHER_TYPES     = 0,
00589     U_UPPERCASE_LETTER        = 1,
00591     U_LOWERCASE_LETTER        = 2,
00593     U_TITLECASE_LETTER        = 3,
00595     U_MODIFIER_LETTER         = 4,
00597     U_OTHER_LETTER            = 5,
00599     U_NON_SPACING_MARK        = 6,
00601     U_ENCLOSING_MARK          = 7,
00603     U_COMBINING_SPACING_MARK  = 8,
00605     U_DECIMAL_DIGIT_NUMBER    = 9,
00607     U_LETTER_NUMBER           = 10,
00609     U_OTHER_NUMBER            = 11,
00611     U_SPACE_SEPARATOR         = 12,
00613     U_LINE_SEPARATOR          = 13,
00615     U_PARAGRAPH_SEPARATOR     = 14,
00617     U_CONTROL_CHAR            = 15,
00619     U_FORMAT_CHAR             = 16,
00621     U_PRIVATE_USE_CHAR        = 17,
00623     U_SURROGATE               = 18,
00625     U_DASH_PUNCTUATION        = 19,
00627     U_START_PUNCTUATION       = 20,
00629     U_END_PUNCTUATION         = 21,
00631     U_CONNECTOR_PUNCTUATION   = 22,
00633     U_OTHER_PUNCTUATION       = 23,
00635     U_MATH_SYMBOL             = 24,
00637     U_CURRENCY_SYMBOL         = 25,
00639     U_MODIFIER_SYMBOL         = 26,
00641     U_OTHER_SYMBOL            = 27,
00643     U_INITIAL_PUNCTUATION     = 28,
00645     U_FINAL_PUNCTUATION       = 29,
00647     U_CHAR_CATEGORY_COUNT
00648 } UCharCategory;
00649 
00664 #define U_GC_CN_MASK    U_MASK(U_GENERAL_OTHER_TYPES)
00665 
00667 #define U_GC_LU_MASK    U_MASK(U_UPPERCASE_LETTER)
00668 
00669 #define U_GC_LL_MASK    U_MASK(U_LOWERCASE_LETTER)
00670 
00671 #define U_GC_LT_MASK    U_MASK(U_TITLECASE_LETTER)
00672 
00673 #define U_GC_LM_MASK    U_MASK(U_MODIFIER_LETTER)
00674 
00675 #define U_GC_LO_MASK    U_MASK(U_OTHER_LETTER)
00676 
00678 #define U_GC_MN_MASK    U_MASK(U_NON_SPACING_MARK)
00679 
00680 #define U_GC_ME_MASK    U_MASK(U_ENCLOSING_MARK)
00681 
00682 #define U_GC_MC_MASK    U_MASK(U_COMBINING_SPACING_MARK)
00683 
00685 #define U_GC_ND_MASK    U_MASK(U_DECIMAL_DIGIT_NUMBER)
00686 
00687 #define U_GC_NL_MASK    U_MASK(U_LETTER_NUMBER)
00688 
00689 #define U_GC_NO_MASK    U_MASK(U_OTHER_NUMBER)
00690 
00692 #define U_GC_ZS_MASK    U_MASK(U_SPACE_SEPARATOR)
00693 
00694 #define U_GC_ZL_MASK    U_MASK(U_LINE_SEPARATOR)
00695 
00696 #define U_GC_ZP_MASK    U_MASK(U_PARAGRAPH_SEPARATOR)
00697 
00699 #define U_GC_CC_MASK    U_MASK(U_CONTROL_CHAR)
00700 
00701 #define U_GC_CF_MASK    U_MASK(U_FORMAT_CHAR)
00702 
00703 #define U_GC_CO_MASK    U_MASK(U_PRIVATE_USE_CHAR)
00704 
00705 #define U_GC_CS_MASK    U_MASK(U_SURROGATE)
00706 
00708 #define U_GC_PD_MASK    U_MASK(U_DASH_PUNCTUATION)
00709 
00710 #define U_GC_PS_MASK    U_MASK(U_START_PUNCTUATION)
00711 
00712 #define U_GC_PE_MASK    U_MASK(U_END_PUNCTUATION)
00713 
00714 #define U_GC_PC_MASK    U_MASK(U_CONNECTOR_PUNCTUATION)
00715 
00716 #define U_GC_PO_MASK    U_MASK(U_OTHER_PUNCTUATION)
00717 
00719 #define U_GC_SM_MASK    U_MASK(U_MATH_SYMBOL)
00720 
00721 #define U_GC_SC_MASK    U_MASK(U_CURRENCY_SYMBOL)
00722 
00723 #define U_GC_SK_MASK    U_MASK(U_MODIFIER_SYMBOL)
00724 
00725 #define U_GC_SO_MASK    U_MASK(U_OTHER_SYMBOL)
00726 
00728 #define U_GC_PI_MASK    U_MASK(U_INITIAL_PUNCTUATION)
00729 
00730 #define U_GC_PF_MASK    U_MASK(U_FINAL_PUNCTUATION)
00731 
00732 
00734 #define U_GC_L_MASK \
00735             (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
00736 
00738 #define U_GC_LC_MASK \
00739             (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
00740 
00742 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
00743 
00745 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
00746 
00748 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
00749 
00751 #define U_GC_C_MASK \
00752             (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
00753 
00755 #define U_GC_P_MASK \
00756             (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
00757              U_GC_PI_MASK|U_GC_PF_MASK)
00758 
00760 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
00761 
00766 typedef enum UCharDirection {
00767     /*
00768      * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
00769      * It matches pairs of lines like
00770      *     / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
00771      *     U_<[A-Z_]+> = <integer>,
00772      */
00773 
00775     U_LEFT_TO_RIGHT               = 0,
00777     U_RIGHT_TO_LEFT               = 1,
00779     U_EUROPEAN_NUMBER             = 2,
00781     U_EUROPEAN_NUMBER_SEPARATOR   = 3,
00783     U_EUROPEAN_NUMBER_TERMINATOR  = 4,
00785     U_ARABIC_NUMBER               = 5,
00787     U_COMMON_NUMBER_SEPARATOR     = 6,
00789     U_BLOCK_SEPARATOR             = 7,
00791     U_SEGMENT_SEPARATOR           = 8,
00793     U_WHITE_SPACE_NEUTRAL         = 9,
00795     U_OTHER_NEUTRAL               = 10,
00797     U_LEFT_TO_RIGHT_EMBEDDING     = 11,
00799     U_LEFT_TO_RIGHT_OVERRIDE      = 12,
00801     U_RIGHT_TO_LEFT_ARABIC        = 13,
00803     U_RIGHT_TO_LEFT_EMBEDDING     = 14,
00805     U_RIGHT_TO_LEFT_OVERRIDE      = 15,
00807     U_POP_DIRECTIONAL_FORMAT      = 16,
00809     U_DIR_NON_SPACING_MARK        = 17,
00811     U_BOUNDARY_NEUTRAL            = 18,
00813     U_CHAR_DIRECTION_COUNT
00814 } UCharDirection;
00815 
00820 enum UBlockCode {
00821     /*
00822      * Note: UBlockCode constants are parsed by preparseucd.py.
00823      * It matches lines like
00824      *     UBLOCK_<Unicode Block value name> = <integer>,
00825      */
00826 
00828     UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
00829 
00831     UBLOCK_BASIC_LATIN = 1, /*[0000]*/
00832 
00834     UBLOCK_LATIN_1_SUPPLEMENT=2, /*[0080]*/
00835 
00837     UBLOCK_LATIN_EXTENDED_A =3, /*[0100]*/
00838 
00840     UBLOCK_LATIN_EXTENDED_B =4, /*[0180]*/
00841 
00843     UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
00844 
00846     UBLOCK_SPACING_MODIFIER_LETTERS =6, /*[02B0]*/
00847 
00849     UBLOCK_COMBINING_DIACRITICAL_MARKS =7, /*[0300]*/
00850 
00855     UBLOCK_GREEK =8, /*[0370]*/
00856 
00858     UBLOCK_CYRILLIC =9, /*[0400]*/
00859 
00861     UBLOCK_ARMENIAN =10, /*[0530]*/
00862 
00864     UBLOCK_HEBREW =11, /*[0590]*/
00865 
00867     UBLOCK_ARABIC =12, /*[0600]*/
00868 
00870     UBLOCK_SYRIAC =13, /*[0700]*/
00871 
00873     UBLOCK_THAANA =14, /*[0780]*/
00874 
00876     UBLOCK_DEVANAGARI =15, /*[0900]*/
00877 
00879     UBLOCK_BENGALI =16, /*[0980]*/
00880 
00882     UBLOCK_GURMUKHI =17, /*[0A00]*/
00883 
00885     UBLOCK_GUJARATI =18, /*[0A80]*/
00886 
00888     UBLOCK_ORIYA =19, /*[0B00]*/
00889 
00891     UBLOCK_TAMIL =20, /*[0B80]*/
00892 
00894     UBLOCK_TELUGU =21, /*[0C00]*/
00895 
00897     UBLOCK_KANNADA =22, /*[0C80]*/
00898 
00900     UBLOCK_MALAYALAM =23, /*[0D00]*/
00901 
00903     UBLOCK_SINHALA =24, /*[0D80]*/
00904 
00906     UBLOCK_THAI =25, /*[0E00]*/
00907 
00909     UBLOCK_LAO =26, /*[0E80]*/
00910 
00912     UBLOCK_TIBETAN =27, /*[0F00]*/
00913 
00915     UBLOCK_MYANMAR =28, /*[1000]*/
00916 
00918     UBLOCK_GEORGIAN =29, /*[10A0]*/
00919 
00921     UBLOCK_HANGUL_JAMO =30, /*[1100]*/
00922 
00924     UBLOCK_ETHIOPIC =31, /*[1200]*/
00925 
00927     UBLOCK_CHEROKEE =32, /*[13A0]*/
00928 
00930     UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS =33, /*[1400]*/
00931 
00933     UBLOCK_OGHAM =34, /*[1680]*/
00934 
00936     UBLOCK_RUNIC =35, /*[16A0]*/
00937 
00939     UBLOCK_KHMER =36, /*[1780]*/
00940 
00942     UBLOCK_MONGOLIAN =37, /*[1800]*/
00943 
00945     UBLOCK_LATIN_EXTENDED_ADDITIONAL =38, /*[1E00]*/
00946 
00948     UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
00949 
00951     UBLOCK_GENERAL_PUNCTUATION =40, /*[2000]*/
00952 
00954     UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS =41, /*[2070]*/
00955 
00957     UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
00958 
00963     UBLOCK_COMBINING_MARKS_FOR_SYMBOLS =43, /*[20D0]*/
00964 
00966     UBLOCK_LETTERLIKE_SYMBOLS =44, /*[2100]*/
00967 
00969     UBLOCK_NUMBER_FORMS =45, /*[2150]*/
00970 
00972     UBLOCK_ARROWS =46, /*[2190]*/
00973 
00975     UBLOCK_MATHEMATICAL_OPERATORS =47, /*[2200]*/
00976 
00978     UBLOCK_MISCELLANEOUS_TECHNICAL =48, /*[2300]*/
00979 
00981     UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
00982 
00984     UBLOCK_OPTICAL_CHARACTER_RECOGNITION =50, /*[2440]*/
00985 
00987     UBLOCK_ENCLOSED_ALPHANUMERICS =51, /*[2460]*/
00988 
00990     UBLOCK_BOX_DRAWING =52, /*[2500]*/
00991 
00993     UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
00994 
00996     UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
00997 
00999     UBLOCK_MISCELLANEOUS_SYMBOLS =55, /*[2600]*/
01000 
01002     UBLOCK_DINGBATS =56, /*[2700]*/
01003 
01005     UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
01006 
01008     UBLOCK_CJK_RADICALS_SUPPLEMENT =58, /*[2E80]*/
01009 
01011     UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
01012 
01014     UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS =60, /*[2FF0]*/
01015 
01017     UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION =61, /*[3000]*/
01018 
01020     UBLOCK_HIRAGANA =62, /*[3040]*/
01021 
01023     UBLOCK_KATAKANA =63, /*[30A0]*/
01024 
01026     UBLOCK_BOPOMOFO =64, /*[3100]*/
01027 
01029     UBLOCK_HANGUL_COMPATIBILITY_JAMO =65, /*[3130]*/
01030 
01032     UBLOCK_KANBUN =66, /*[3190]*/
01033 
01035     UBLOCK_BOPOMOFO_EXTENDED =67, /*[31A0]*/
01036 
01038     UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS =68, /*[3200]*/
01039 
01041     UBLOCK_CJK_COMPATIBILITY =69, /*[3300]*/
01042 
01044     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A =70, /*[3400]*/
01045 
01047     UBLOCK_CJK_UNIFIED_IDEOGRAPHS =71, /*[4E00]*/
01048 
01050     UBLOCK_YI_SYLLABLES =72, /*[A000]*/
01051 
01053     UBLOCK_YI_RADICALS =73, /*[A490]*/
01054 
01056     UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
01057 
01059     UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
01060 
01062     UBLOCK_HIGH_PRIVATE_USE_SURROGATES =76, /*[DB80]*/
01063 
01065     UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
01066 
01076     UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
01086     UBLOCK_PRIVATE_USE = UBLOCK_PRIVATE_USE_AREA,
01087 
01089     UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS =79, /*[F900]*/
01090 
01092     UBLOCK_ALPHABETIC_PRESENTATION_FORMS =80, /*[FB00]*/
01093 
01095     UBLOCK_ARABIC_PRESENTATION_FORMS_A =81, /*[FB50]*/
01096 
01098     UBLOCK_COMBINING_HALF_MARKS =82, /*[FE20]*/
01099 
01101     UBLOCK_CJK_COMPATIBILITY_FORMS =83, /*[FE30]*/
01102 
01104     UBLOCK_SMALL_FORM_VARIANTS =84, /*[FE50]*/
01105 
01107     UBLOCK_ARABIC_PRESENTATION_FORMS_B =85, /*[FE70]*/
01108 
01110     UBLOCK_SPECIALS =86, /*[FFF0]*/
01111 
01113     UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS =87, /*[FF00]*/
01114 
01115     /* New blocks in Unicode 3.1 */
01116 
01118     UBLOCK_OLD_ITALIC = 88, /*[10300]*/
01120     UBLOCK_GOTHIC = 89, /*[10330]*/
01122     UBLOCK_DESERET = 90, /*[10400]*/
01124     UBLOCK_BYZANTINE_MUSICAL_SYMBOLS = 91, /*[1D000]*/
01126     UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
01128     UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS = 93, /*[1D400]*/
01130     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B  = 94, /*[20000]*/
01132     UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT = 95, /*[2F800]*/
01134     UBLOCK_TAGS = 96, /*[E0000]*/
01135 
01136     /* New blocks in Unicode 3.2 */
01137 
01139     UBLOCK_CYRILLIC_SUPPLEMENT = 97, /*[0500]*/
01144     UBLOCK_CYRILLIC_SUPPLEMENTARY = UBLOCK_CYRILLIC_SUPPLEMENT, 
01146     UBLOCK_TAGALOG = 98, /*[1700]*/
01148     UBLOCK_HANUNOO = 99, /*[1720]*/
01150     UBLOCK_BUHID = 100, /*[1740]*/
01152     UBLOCK_TAGBANWA = 101, /*[1760]*/
01154     UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A = 102, /*[27C0]*/
01156     UBLOCK_SUPPLEMENTAL_ARROWS_A = 103, /*[27F0]*/
01158     UBLOCK_SUPPLEMENTAL_ARROWS_B = 104, /*[2900]*/
01160     UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B = 105, /*[2980]*/
01162     UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS = 106, /*[2A00]*/
01164     UBLOCK_KATAKANA_PHONETIC_EXTENSIONS = 107, /*[31F0]*/
01166     UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
01168     UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A = 109, /*[F0000]*/
01170     UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B = 110, /*[100000]*/
01171 
01172     /* New blocks in Unicode 4 */
01173 
01175     UBLOCK_LIMBU = 111, /*[1900]*/
01177     UBLOCK_TAI_LE = 112, /*[1950]*/
01179     UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
01181     UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
01183     UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS = 115, /*[2B00]*/
01185     UBLOCK_YIJING_HEXAGRAM_SYMBOLS = 116, /*[4DC0]*/
01187     UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
01189     UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
01191     UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
01193     UBLOCK_UGARITIC = 120, /*[10380]*/
01195     UBLOCK_SHAVIAN = 121, /*[10450]*/
01197     UBLOCK_OSMANYA = 122, /*[10480]*/
01199     UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
01201     UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
01203     UBLOCK_VARIATION_SELECTORS_SUPPLEMENT = 125, /*[E0100]*/
01204 
01205     /* New blocks in Unicode 4.1 */
01206 
01208     UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION = 126, /*[1D200]*/
01210     UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
01212     UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
01214     UBLOCK_BUGINESE = 129, /*[1A00]*/
01216     UBLOCK_CJK_STROKES = 130, /*[31C0]*/
01218     UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT = 131, /*[1DC0]*/
01220     UBLOCK_COPTIC = 132, /*[2C80]*/
01222     UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
01224     UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
01226     UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
01228     UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
01230     UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
01232     UBLOCK_MODIFIER_TONE_LETTERS = 138, /*[A700]*/
01234     UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
01236     UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
01238     UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT = 141, /*[1D80]*/
01240     UBLOCK_SUPPLEMENTAL_PUNCTUATION = 142, /*[2E00]*/
01242     UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
01244     UBLOCK_TIFINAGH = 144, /*[2D30]*/
01246     UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
01247 
01248     /* New blocks in Unicode 5.0 */
01249 
01251     UBLOCK_NKO = 146, /*[07C0]*/
01253     UBLOCK_BALINESE = 147, /*[1B00]*/
01255     UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
01257     UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
01259     UBLOCK_PHAGS_PA = 150, /*[A840]*/
01261     UBLOCK_PHOENICIAN = 151, /*[10900]*/
01263     UBLOCK_CUNEIFORM = 152, /*[12000]*/
01265     UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION = 153, /*[12400]*/
01267     UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
01268 
01269     /* New blocks in Unicode 5.1 */
01270 
01272     UBLOCK_SUNDANESE = 155, /*[1B80]*/
01274     UBLOCK_LEPCHA = 156, /*[1C00]*/
01276     UBLOCK_OL_CHIKI = 157, /*[1C50]*/
01278     UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
01280     UBLOCK_VAI = 159, /*[A500]*/
01282     UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
01284     UBLOCK_SAURASHTRA = 161, /*[A880]*/
01286     UBLOCK_KAYAH_LI = 162, /*[A900]*/
01288     UBLOCK_REJANG = 163, /*[A930]*/
01290     UBLOCK_CHAM = 164, /*[AA00]*/
01292     UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
01294     UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
01296     UBLOCK_LYCIAN = 167, /*[10280]*/
01298     UBLOCK_CARIAN = 168, /*[102A0]*/
01300     UBLOCK_LYDIAN = 169, /*[10920]*/
01302     UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
01304     UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
01305 
01306     /* New blocks in Unicode 5.2 */
01307 
01309     UBLOCK_SAMARITAN = 172, /*[0800]*/
01311     UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED = 173, /*[18B0]*/
01313     UBLOCK_TAI_THAM = 174, /*[1A20]*/
01315     UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
01317     UBLOCK_LISU = 176, /*[A4D0]*/
01319     UBLOCK_BAMUM = 177, /*[A6A0]*/
01321     UBLOCK_COMMON_INDIC_NUMBER_FORMS = 178, /*[A830]*/
01323     UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
01325     UBLOCK_HANGUL_JAMO_EXTENDED_A = 180, /*[A960]*/
01327     UBLOCK_JAVANESE = 181, /*[A980]*/
01329     UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
01331     UBLOCK_TAI_VIET = 183, /*[AA80]*/
01333     UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
01335     UBLOCK_HANGUL_JAMO_EXTENDED_B = 185, /*[D7B0]*/
01337     UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
01339     UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
01341     UBLOCK_AVESTAN = 188, /*[10B00]*/
01343     UBLOCK_INSCRIPTIONAL_PARTHIAN = 189, /*[10B40]*/
01345     UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
01347     UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
01349     UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
01351     UBLOCK_KAITHI = 193, /*[11080]*/
01353     UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
01355     UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT = 195, /*[1F100]*/
01357     UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT = 196, /*[1F200]*/
01359     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C = 197, /*[2A700]*/
01360 
01361     /* New blocks in Unicode 6.0 */
01362 
01364     UBLOCK_MANDAIC = 198, /*[0840]*/
01366     UBLOCK_BATAK = 199, /*[1BC0]*/
01368     UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
01370     UBLOCK_BRAHMI = 201, /*[11000]*/
01372     UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
01374     UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
01376     UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
01378     UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS = 205, /*[1F300]*/
01380     UBLOCK_EMOTICONS = 206, /*[1F600]*/
01382     UBLOCK_TRANSPORT_AND_MAP_SYMBOLS = 207, /*[1F680]*/
01384     UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
01386     UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D = 209, /*[2B740]*/
01387 
01388     /* New blocks in Unicode 6.1 */
01389 
01391     UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
01393     UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS = 211, /*[1EE00]*/
01395     UBLOCK_CHAKMA = 212, /*[11100]*/
01397     UBLOCK_MEETEI_MAYEK_EXTENSIONS = 213, /*[AAE0]*/
01399     UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
01401     UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
01403     UBLOCK_MIAO = 216, /*[16F00]*/
01405     UBLOCK_SHARADA = 217, /*[11180]*/
01407     UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
01409     UBLOCK_SUNDANESE_SUPPLEMENT = 219, /*[1CC0]*/
01411     UBLOCK_TAKRI = 220, /*[11680]*/
01412 
01414     UBLOCK_COUNT = 221,
01415 
01417     UBLOCK_INVALID_CODE=-1
01418 };
01419 
01421 typedef enum UBlockCode UBlockCode;
01422 
01430 typedef enum UEastAsianWidth {
01431     /*
01432      * Note: UEastAsianWidth constants are parsed by preparseucd.py.
01433      * It matches lines like
01434      *     U_EA_<Unicode East_Asian_Width value name>
01435      */
01436 
01437     U_EA_NEUTRAL,   /*[N]*/
01438     U_EA_AMBIGUOUS, /*[A]*/
01439     U_EA_HALFWIDTH, /*[H]*/
01440     U_EA_FULLWIDTH, /*[F]*/
01441     U_EA_NARROW,    /*[Na]*/
01442     U_EA_WIDE,      /*[W]*/
01443     U_EA_COUNT
01444 } UEastAsianWidth;
01445 
01457 typedef enum UCharNameChoice {
01459     U_UNICODE_CHAR_NAME,
01465     U_UNICODE_10_CHAR_NAME,
01467     U_EXTENDED_CHAR_NAME,
01469     U_CHAR_NAME_ALIAS,
01471     U_CHAR_NAME_CHOICE_COUNT
01472 } UCharNameChoice;
01473 
01487 typedef enum UPropertyNameChoice {
01488     U_SHORT_PROPERTY_NAME,
01489     U_LONG_PROPERTY_NAME,
01490     U_PROPERTY_NAME_CHOICE_COUNT
01491 } UPropertyNameChoice;
01492 
01499 typedef enum UDecompositionType {
01500     /*
01501      * Note: UDecompositionType constants are parsed by preparseucd.py.
01502      * It matches lines like
01503      *     U_DT_<Unicode Decomposition_Type value name>
01504      */
01505 
01506     U_DT_NONE,              /*[none]*/
01507     U_DT_CANONICAL,         /*[can]*/
01508     U_DT_COMPAT,            /*[com]*/
01509     U_DT_CIRCLE,            /*[enc]*/
01510     U_DT_FINAL,             /*[fin]*/
01511     U_DT_FONT,              /*[font]*/
01512     U_DT_FRACTION,          /*[fra]*/
01513     U_DT_INITIAL,           /*[init]*/
01514     U_DT_ISOLATED,          /*[iso]*/
01515     U_DT_MEDIAL,            /*[med]*/
01516     U_DT_NARROW,            /*[nar]*/
01517     U_DT_NOBREAK,           /*[nb]*/
01518     U_DT_SMALL,             /*[sml]*/
01519     U_DT_SQUARE,            /*[sqr]*/
01520     U_DT_SUB,               /*[sub]*/
01521     U_DT_SUPER,             /*[sup]*/
01522     U_DT_VERTICAL,          /*[vert]*/
01523     U_DT_WIDE,              /*[wide]*/
01524     U_DT_COUNT /* 18 */
01525 } UDecompositionType;
01526 
01533 typedef enum UJoiningType {
01534     /*
01535      * Note: UJoiningType constants are parsed by preparseucd.py.
01536      * It matches lines like
01537      *     U_JT_<Unicode Joining_Type value name>
01538      */
01539 
01540     U_JT_NON_JOINING,       /*[U]*/
01541     U_JT_JOIN_CAUSING,      /*[C]*/
01542     U_JT_DUAL_JOINING,      /*[D]*/
01543     U_JT_LEFT_JOINING,      /*[L]*/
01544     U_JT_RIGHT_JOINING,     /*[R]*/
01545     U_JT_TRANSPARENT,       /*[T]*/
01546     U_JT_COUNT /* 6 */
01547 } UJoiningType;
01548 
01555 typedef enum UJoiningGroup {
01556     /*
01557      * Note: UJoiningGroup constants are parsed by preparseucd.py.
01558      * It matches lines like
01559      *     U_JG_<Unicode Joining_Group value name>
01560      */
01561 
01562     U_JG_NO_JOINING_GROUP,
01563     U_JG_AIN,
01564     U_JG_ALAPH,
01565     U_JG_ALEF,
01566     U_JG_BEH,
01567     U_JG_BETH,
01568     U_JG_DAL,
01569     U_JG_DALATH_RISH,
01570     U_JG_E,
01571     U_JG_FEH,
01572     U_JG_FINAL_SEMKATH,
01573     U_JG_GAF,
01574     U_JG_GAMAL,
01575     U_JG_HAH,
01576     U_JG_TEH_MARBUTA_GOAL,  
01577     U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
01578     U_JG_HE,
01579     U_JG_HEH,
01580     U_JG_HEH_GOAL,
01581     U_JG_HETH,
01582     U_JG_KAF,
01583     U_JG_KAPH,
01584     U_JG_KNOTTED_HEH,
01585     U_JG_LAM,
01586     U_JG_LAMADH,
01587     U_JG_MEEM,
01588     U_JG_MIM,
01589     U_JG_NOON,
01590     U_JG_NUN,
01591     U_JG_PE,
01592     U_JG_QAF,
01593     U_JG_QAPH,
01594     U_JG_REH,
01595     U_JG_REVERSED_PE,
01596     U_JG_SAD,
01597     U_JG_SADHE,
01598     U_JG_SEEN,
01599     U_JG_SEMKATH,
01600     U_JG_SHIN,
01601     U_JG_SWASH_KAF,
01602     U_JG_SYRIAC_WAW,
01603     U_JG_TAH,
01604     U_JG_TAW,
01605     U_JG_TEH_MARBUTA,
01606     U_JG_TETH,
01607     U_JG_WAW,
01608     U_JG_YEH,
01609     U_JG_YEH_BARREE,
01610     U_JG_YEH_WITH_TAIL,
01611     U_JG_YUDH,
01612     U_JG_YUDH_HE,
01613     U_JG_ZAIN,
01614     U_JG_FE,        
01615     U_JG_KHAPH,     
01616     U_JG_ZHAIN,     
01617     U_JG_BURUSHASKI_YEH_BARREE, 
01618     U_JG_FARSI_YEH, 
01619     U_JG_NYA,       
01620     U_JG_ROHINGYA_YEH,  
01621     U_JG_COUNT
01622 } UJoiningGroup;
01623 
01630 typedef enum UGraphemeClusterBreak {
01631     /*
01632      * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
01633      * It matches lines like
01634      *     U_GCB_<Unicode Grapheme_Cluster_Break value name>
01635      */
01636 
01637     U_GCB_OTHER = 0,            /*[XX]*/
01638     U_GCB_CONTROL = 1,          /*[CN]*/
01639     U_GCB_CR = 2,               /*[CR]*/
01640     U_GCB_EXTEND = 3,           /*[EX]*/
01641     U_GCB_L = 4,                /*[L]*/
01642     U_GCB_LF = 5,               /*[LF]*/
01643     U_GCB_LV = 6,               /*[LV]*/
01644     U_GCB_LVT = 7,              /*[LVT]*/
01645     U_GCB_T = 8,                /*[T]*/
01646     U_GCB_V = 9,                /*[V]*/
01647     U_GCB_SPACING_MARK = 10,    /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
01648     U_GCB_PREPEND = 11,         /*[PP]*/
01649     U_GCB_REGIONAL_INDICATOR = 12,  /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
01650     U_GCB_COUNT = 13
01651 } UGraphemeClusterBreak;
01652 
01660 typedef enum UWordBreakValues {
01661     /*
01662      * Note: UWordBreakValues constants are parsed by preparseucd.py.
01663      * It matches lines like
01664      *     U_WB_<Unicode Word_Break value name>
01665      */
01666 
01667     U_WB_OTHER = 0,             /*[XX]*/
01668     U_WB_ALETTER = 1,           /*[LE]*/
01669     U_WB_FORMAT = 2,            /*[FO]*/
01670     U_WB_KATAKANA = 3,          /*[KA]*/
01671     U_WB_MIDLETTER = 4,         /*[ML]*/
01672     U_WB_MIDNUM = 5,            /*[MN]*/
01673     U_WB_NUMERIC = 6,           /*[NU]*/
01674     U_WB_EXTENDNUMLET = 7,      /*[EX]*/
01675     U_WB_CR = 8,                /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
01676     U_WB_EXTEND = 9,            /*[Extend]*/
01677     U_WB_LF = 10,               /*[LF]*/
01678     U_WB_MIDNUMLET =11,         /*[MB]*/
01679     U_WB_NEWLINE =12,           /*[NL]*/
01680     U_WB_REGIONAL_INDICATOR = 13,   /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
01681     U_WB_COUNT = 14
01682 } UWordBreakValues;
01683 
01690 typedef enum USentenceBreak {
01691     /*
01692      * Note: USentenceBreak constants are parsed by preparseucd.py.
01693      * It matches lines like
01694      *     U_SB_<Unicode Sentence_Break value name>
01695      */
01696 
01697     U_SB_OTHER = 0,             /*[XX]*/
01698     U_SB_ATERM = 1,             /*[AT]*/
01699     U_SB_CLOSE = 2,             /*[CL]*/
01700     U_SB_FORMAT = 3,            /*[FO]*/
01701     U_SB_LOWER = 4,             /*[LO]*/
01702     U_SB_NUMERIC = 5,           /*[NU]*/
01703     U_SB_OLETTER = 6,           /*[LE]*/
01704     U_SB_SEP = 7,               /*[SE]*/
01705     U_SB_SP = 8,                /*[SP]*/
01706     U_SB_STERM = 9,             /*[ST]*/
01707     U_SB_UPPER = 10,            /*[UP]*/
01708     U_SB_CR = 11,               /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
01709     U_SB_EXTEND = 12,           /*[EX]*/
01710     U_SB_LF = 13,               /*[LF]*/
01711     U_SB_SCONTINUE = 14,        /*[SC]*/
01712     U_SB_COUNT = 15
01713 } USentenceBreak;
01714 
01721 typedef enum ULineBreak {
01722     /*
01723      * Note: ULineBreak constants are parsed by preparseucd.py.
01724      * It matches lines like
01725      *     U_LB_<Unicode Line_Break value name>
01726      */
01727 
01728     U_LB_UNKNOWN = 0,           /*[XX]*/
01729     U_LB_AMBIGUOUS = 1,         /*[AI]*/
01730     U_LB_ALPHABETIC = 2,        /*[AL]*/
01731     U_LB_BREAK_BOTH = 3,        /*[B2]*/
01732     U_LB_BREAK_AFTER = 4,       /*[BA]*/
01733     U_LB_BREAK_BEFORE = 5,      /*[BB]*/
01734     U_LB_MANDATORY_BREAK = 6,   /*[BK]*/
01735     U_LB_CONTINGENT_BREAK = 7,  /*[CB]*/
01736     U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
01737     U_LB_COMBINING_MARK = 9,    /*[CM]*/
01738     U_LB_CARRIAGE_RETURN = 10,   /*[CR]*/
01739     U_LB_EXCLAMATION = 11,       /*[EX]*/
01740     U_LB_GLUE = 12,              /*[GL]*/
01741     U_LB_HYPHEN = 13,            /*[HY]*/
01742     U_LB_IDEOGRAPHIC = 14,       /*[ID]*/
01744     U_LB_INSEPARABLE = 15,       /*[IN]*/
01745     U_LB_INSEPERABLE = U_LB_INSEPARABLE,
01746     U_LB_INFIX_NUMERIC = 16,     /*[IS]*/
01747     U_LB_LINE_FEED = 17,         /*[LF]*/
01748     U_LB_NONSTARTER = 18,        /*[NS]*/
01749     U_LB_NUMERIC = 19,           /*[NU]*/
01750     U_LB_OPEN_PUNCTUATION = 20,  /*[OP]*/
01751     U_LB_POSTFIX_NUMERIC = 21,   /*[PO]*/
01752     U_LB_PREFIX_NUMERIC = 22,    /*[PR]*/
01753     U_LB_QUOTATION = 23,         /*[QU]*/
01754     U_LB_COMPLEX_CONTEXT = 24,   /*[SA]*/
01755     U_LB_SURROGATE = 25,         /*[SG]*/
01756     U_LB_SPACE = 26,             /*[SP]*/
01757     U_LB_BREAK_SYMBOLS = 27,     /*[SY]*/
01758     U_LB_ZWSPACE = 28,           /*[ZW]*/
01759     U_LB_NEXT_LINE = 29,         /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
01760     U_LB_WORD_JOINER = 30,       /*[WJ]*/
01761     U_LB_H2 = 31,                /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
01762     U_LB_H3 = 32,                /*[H3]*/
01763     U_LB_JL = 33,                /*[JL]*/
01764     U_LB_JT = 34,                /*[JT]*/
01765     U_LB_JV = 35,                /*[JV]*/
01766     U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
01767     U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
01768     U_LB_HEBREW_LETTER = 38,     /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
01769     U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
01770     U_LB_COUNT = 40
01771 } ULineBreak;
01772 
01779 typedef enum UNumericType {
01780     /*
01781      * Note: UNumericType constants are parsed by preparseucd.py.
01782      * It matches lines like
01783      *     U_NT_<Unicode Numeric_Type value name>
01784      */
01785 
01786     U_NT_NONE,              /*[None]*/
01787     U_NT_DECIMAL,           /*[de]*/
01788     U_NT_DIGIT,             /*[di]*/
01789     U_NT_NUMERIC,           /*[nu]*/
01790     U_NT_COUNT
01791 } UNumericType;
01792 
01799 typedef enum UHangulSyllableType {
01800     /*
01801      * Note: UHangulSyllableType constants are parsed by preparseucd.py.
01802      * It matches lines like
01803      *     U_HST_<Unicode Hangul_Syllable_Type value name>
01804      */
01805 
01806     U_HST_NOT_APPLICABLE,   /*[NA]*/
01807     U_HST_LEADING_JAMO,     /*[L]*/
01808     U_HST_VOWEL_JAMO,       /*[V]*/
01809     U_HST_TRAILING_JAMO,    /*[T]*/
01810     U_HST_LV_SYLLABLE,      /*[LV]*/
01811     U_HST_LVT_SYLLABLE,     /*[LVT]*/
01812     U_HST_COUNT
01813 } UHangulSyllableType;
01814 
01841 U_STABLE UBool U_EXPORT2
01842 u_hasBinaryProperty(UChar32 c, UProperty which);
01843 
01856 U_STABLE UBool U_EXPORT2
01857 u_isUAlphabetic(UChar32 c);
01858 
01871 U_STABLE UBool U_EXPORT2
01872 u_isULowercase(UChar32 c);
01873 
01886 U_STABLE UBool U_EXPORT2
01887 u_isUUppercase(UChar32 c);
01888 
01907 U_STABLE UBool U_EXPORT2
01908 u_isUWhiteSpace(UChar32 c);
01909 
01947 U_STABLE int32_t U_EXPORT2
01948 u_getIntPropertyValue(UChar32 c, UProperty which);
01949 
01968 U_STABLE int32_t U_EXPORT2
01969 u_getIntPropertyMinValue(UProperty which);
01970 
01997 U_STABLE int32_t U_EXPORT2
01998 u_getIntPropertyMaxValue(UProperty which);
01999 
02022 U_STABLE double U_EXPORT2
02023 u_getNumericValue(UChar32 c);
02024 
02032 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
02033 
02057 U_STABLE UBool U_EXPORT2
02058 u_islower(UChar32 c);
02059 
02084 U_STABLE UBool U_EXPORT2
02085 u_isupper(UChar32 c);
02086 
02101 U_STABLE UBool U_EXPORT2
02102 u_istitle(UChar32 c);
02103 
02122 U_STABLE UBool U_EXPORT2
02123 u_isdigit(UChar32 c);
02124 
02143 U_STABLE UBool U_EXPORT2
02144 u_isalpha(UChar32 c);
02145 
02164 U_STABLE UBool U_EXPORT2
02165 u_isalnum(UChar32 c);
02166 
02187 U_STABLE UBool U_EXPORT2
02188 u_isxdigit(UChar32 c);
02189 
02203 U_STABLE UBool U_EXPORT2
02204 u_ispunct(UChar32 c);
02205 
02222 U_STABLE UBool U_EXPORT2
02223 u_isgraph(UChar32 c);
02224 
02251 U_STABLE UBool U_EXPORT2
02252 u_isblank(UChar32 c);
02253 
02276 U_STABLE UBool U_EXPORT2
02277 u_isdefined(UChar32 c);
02278 
02297 U_STABLE UBool U_EXPORT2
02298 u_isspace(UChar32 c);
02299 
02318 U_STABLE UBool U_EXPORT2
02319 u_isJavaSpaceChar(UChar32 c);
02320 
02358 U_STABLE UBool U_EXPORT2
02359 u_isWhitespace(UChar32 c);
02360 
02382 U_STABLE UBool U_EXPORT2
02383 u_iscntrl(UChar32 c);
02384 
02397 U_STABLE UBool U_EXPORT2
02398 u_isISOControl(UChar32 c);
02399 
02415 U_STABLE UBool U_EXPORT2
02416 u_isprint(UChar32 c);
02417 
02436 U_STABLE UBool U_EXPORT2
02437 u_isbase(UChar32 c);
02438 
02455 U_STABLE UCharDirection U_EXPORT2
02456 u_charDirection(UChar32 c);
02457 
02473 U_STABLE UBool U_EXPORT2
02474 u_isMirrored(UChar32 c);
02475 
02495 U_STABLE UChar32 U_EXPORT2
02496 u_charMirror(UChar32 c);
02497 
02509 U_STABLE int8_t U_EXPORT2
02510 u_charType(UChar32 c);
02511 
02525 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
02526 
02544 typedef UBool U_CALLCONV
02545 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
02546 
02566 U_STABLE void U_EXPORT2
02567 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
02568 
02569 #if !UCONFIG_NO_NORMALIZATION
02570 
02578 U_STABLE uint8_t U_EXPORT2
02579 u_getCombiningClass(UChar32 c);
02580 
02581 #endif
02582 
02606 U_STABLE int32_t U_EXPORT2
02607 u_charDigitValue(UChar32 c);
02608 
02618 U_STABLE UBlockCode U_EXPORT2
02619 ublock_getCode(UChar32 c);
02620 
02653 U_STABLE int32_t U_EXPORT2
02654 u_charName(UChar32 code, UCharNameChoice nameChoice,
02655            char *buffer, int32_t bufferLength,
02656            UErrorCode *pErrorCode);
02657 
02676 U_STABLE int32_t U_EXPORT2
02677 u_getISOComment(UChar32 c,
02678                 char *dest, int32_t destCapacity,
02679                 UErrorCode *pErrorCode);
02680 
02701 U_STABLE UChar32 U_EXPORT2
02702 u_charFromName(UCharNameChoice nameChoice,
02703                const char *name,
02704                UErrorCode *pErrorCode);
02705 
02723 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
02724                                UChar32 code,
02725                                UCharNameChoice nameChoice,
02726                                const char *name,
02727                                int32_t length);
02728 
02750 U_STABLE void U_EXPORT2
02751 u_enumCharNames(UChar32 start, UChar32 limit,
02752                 UEnumCharNamesFn *fn,
02753                 void *context,
02754                 UCharNameChoice nameChoice,
02755                 UErrorCode *pErrorCode);
02756 
02788 U_STABLE const char* U_EXPORT2
02789 u_getPropertyName(UProperty property,
02790                   UPropertyNameChoice nameChoice);
02791 
02811 U_STABLE UProperty U_EXPORT2
02812 u_getPropertyEnum(const char* alias);
02813 
02861 U_STABLE const char* U_EXPORT2
02862 u_getPropertyValueName(UProperty property,
02863                        int32_t value,
02864                        UPropertyNameChoice nameChoice);
02865 
02897 U_STABLE int32_t U_EXPORT2
02898 u_getPropertyValueEnum(UProperty property,
02899                        const char* alias);
02900 
02918 U_STABLE UBool U_EXPORT2
02919 u_isIDStart(UChar32 c);
02920 
02942 U_STABLE UBool U_EXPORT2
02943 u_isIDPart(UChar32 c);
02944 
02965 U_STABLE UBool U_EXPORT2
02966 u_isIDIgnorable(UChar32 c);
02967 
02984 U_STABLE UBool U_EXPORT2
02985 u_isJavaIDStart(UChar32 c);
02986 
03005 U_STABLE UBool U_EXPORT2
03006 u_isJavaIDPart(UChar32 c);
03007 
03030 U_STABLE UChar32 U_EXPORT2
03031 u_tolower(UChar32 c);
03032 
03055 U_STABLE UChar32 U_EXPORT2
03056 u_toupper(UChar32 c);
03057 
03080 U_STABLE UChar32 U_EXPORT2
03081 u_totitle(UChar32 c);
03082 
03084 #define U_FOLD_CASE_DEFAULT 0
03085 
03102 #define U_FOLD_CASE_EXCLUDE_SPECIAL_I 1
03103 
03126 U_STABLE UChar32 U_EXPORT2
03127 u_foldCase(UChar32 c, uint32_t options);
03128 
03167 U_STABLE int32_t U_EXPORT2
03168 u_digit(UChar32 ch, int8_t radix);
03169 
03198 U_STABLE UChar32 U_EXPORT2
03199 u_forDigit(int32_t digit, int8_t radix);
03200 
03215 U_STABLE void U_EXPORT2
03216 u_charAge(UChar32 c, UVersionInfo versionArray);
03217 
03229 U_STABLE void U_EXPORT2
03230 u_getUnicodeVersion(UVersionInfo versionArray);
03231 
03232 #if !UCONFIG_NO_NORMALIZATION
03233 
03254 U_STABLE int32_t U_EXPORT2
03255 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
03256 
03257 #endif
03258 
03259 
03260 U_CDECL_END
03261 
03262 #endif /*_UCHAR*/
03263 /*eof*/

Generated on 27 Oct 2013 for ICU 50.1.2 by  doxygen 1.4.7