uchar.h

Go to the documentation of this file.

/*
**********************************************************************
*   Copyright (C) 1997-2016, International Business Machines
*   Corporation and others.  All Rights Reserved.
**********************************************************************
*
* File UCHAR.H
*
* Modification History:
*
*   Date        Name        Description
*   04/02/97    aliu        Creation.
*   03/29/99    helena      Updated for C APIs.
*   4/15/99     Madhu       Updated for C Implementation and Javadoc
*   5/20/99     Madhu       Added the function u_getVersion()
*   8/19/1999   srl         Upgraded scripts to Unicode 3.0
*   8/27/1999   schererm    UCharDirection constants: U_...
*   11/11/1999  weiv        added u_isalnum(), cleaned comments
*   01/11/2000  helena      Renamed u_getVersion to u_getUnicodeVersion().
******************************************************************************
*/

#ifndef UCHAR_H
#define UCHAR_H

#include "unicode/utypes.h"

U_CDECL_BEGIN

/*==========================================================================*/
/* Unicode version number                                                   */
/*==========================================================================*/
#define U_UNICODE_VERSION "8.0"

#define UCHAR_MIN_VALUE 0

#define UCHAR_MAX_VALUE 0x10ffff

#define U_MASK(x) ((uint32_t)1<<(x))

typedef enum UProperty {
    /*
     * Note: UProperty constants are parsed by preparseucd.py.
     * It matches lines like
     *     UCHAR_<Unicode property name>=<integer>,
     */

    /*  Note: Place UCHAR_ALPHABETIC before UCHAR_BINARY_START so that
    debuggers display UCHAR_ALPHABETIC as the symbolic name for 0,
    rather than UCHAR_BINARY_START.  Likewise for other *_START
    identifiers. */

    UCHAR_ALPHABETIC=0,
    UCHAR_BINARY_START=UCHAR_ALPHABETIC,
    UCHAR_ASCII_HEX_DIGIT=1,
    UCHAR_BIDI_CONTROL=2,
    UCHAR_BIDI_MIRRORED=3,
    UCHAR_DASH=4,
    UCHAR_DEFAULT_IGNORABLE_CODE_POINT=5,
    UCHAR_DEPRECATED=6,
    UCHAR_DIACRITIC=7,
    UCHAR_EXTENDER=8,
    UCHAR_FULL_COMPOSITION_EXCLUSION=9,
    UCHAR_GRAPHEME_BASE=10,
    UCHAR_GRAPHEME_EXTEND=11,
    UCHAR_GRAPHEME_LINK=12,
    UCHAR_HEX_DIGIT=13,
    UCHAR_HYPHEN=14,
    UCHAR_ID_CONTINUE=15,
    UCHAR_ID_START=16,
    UCHAR_IDEOGRAPHIC=17,
    UCHAR_IDS_BINARY_OPERATOR=18,
    UCHAR_IDS_TRINARY_OPERATOR=19,
    UCHAR_JOIN_CONTROL=20,
    UCHAR_LOGICAL_ORDER_EXCEPTION=21,
    UCHAR_LOWERCASE=22,
    UCHAR_MATH=23,
    UCHAR_NONCHARACTER_CODE_POINT=24,
    UCHAR_QUOTATION_MARK=25,
    UCHAR_RADICAL=26,
    UCHAR_SOFT_DOTTED=27,
    UCHAR_TERMINAL_PUNCTUATION=28,
    UCHAR_UNIFIED_IDEOGRAPH=29,
    UCHAR_UPPERCASE=30,
    UCHAR_WHITE_SPACE=31,
    UCHAR_XID_CONTINUE=32,
    UCHAR_XID_START=33,
   UCHAR_CASE_SENSITIVE=34,
    UCHAR_S_TERM=35,
    UCHAR_VARIATION_SELECTOR=36,
    UCHAR_NFD_INERT=37,
    UCHAR_NFKD_INERT=38,
    UCHAR_NFC_INERT=39,
    UCHAR_NFKC_INERT=40,
    UCHAR_SEGMENT_STARTER=41,
    UCHAR_PATTERN_SYNTAX=42,
    UCHAR_PATTERN_WHITE_SPACE=43,
    UCHAR_POSIX_ALNUM=44,
    UCHAR_POSIX_BLANK=45,
    UCHAR_POSIX_GRAPH=46,
    UCHAR_POSIX_PRINT=47,
    UCHAR_POSIX_XDIGIT=48,
    UCHAR_CASED=49,
    UCHAR_CASE_IGNORABLE=50,
    UCHAR_CHANGES_WHEN_LOWERCASED=51,
    UCHAR_CHANGES_WHEN_UPPERCASED=52,
    UCHAR_CHANGES_WHEN_TITLECASED=53,
    UCHAR_CHANGES_WHEN_CASEFOLDED=54,
    UCHAR_CHANGES_WHEN_CASEMAPPED=55,
    UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED=56,
#ifndef U_HIDE_DRAFT_API

    UCHAR_EMOJI=57,
    UCHAR_EMOJI_PRESENTATION=58,
    UCHAR_EMOJI_MODIFIER=59,
    UCHAR_EMOJI_MODIFIER_BASE=60,
#endif /* U_HIDE_DRAFT_API */

    UCHAR_BINARY_LIMIT=61,

    UCHAR_BIDI_CLASS=0x1000,
    UCHAR_INT_START=UCHAR_BIDI_CLASS,
    UCHAR_BLOCK=0x1001,
    UCHAR_CANONICAL_COMBINING_CLASS=0x1002,
    UCHAR_DECOMPOSITION_TYPE=0x1003,
    UCHAR_EAST_ASIAN_WIDTH=0x1004,
    UCHAR_GENERAL_CATEGORY=0x1005,
    UCHAR_JOINING_GROUP=0x1006,
    UCHAR_JOINING_TYPE=0x1007,
    UCHAR_LINE_BREAK=0x1008,
    UCHAR_NUMERIC_TYPE=0x1009,
    UCHAR_SCRIPT=0x100A,
    UCHAR_HANGUL_SYLLABLE_TYPE=0x100B,
    UCHAR_NFD_QUICK_CHECK=0x100C,
    UCHAR_NFKD_QUICK_CHECK=0x100D,
    UCHAR_NFC_QUICK_CHECK=0x100E,
    UCHAR_NFKC_QUICK_CHECK=0x100F,
    UCHAR_LEAD_CANONICAL_COMBINING_CLASS=0x1010,
    UCHAR_TRAIL_CANONICAL_COMBINING_CLASS=0x1011,
    UCHAR_GRAPHEME_CLUSTER_BREAK=0x1012,
    UCHAR_SENTENCE_BREAK=0x1013,
    UCHAR_WORD_BREAK=0x1014,
    UCHAR_BIDI_PAIRED_BRACKET_TYPE=0x1015,
    UCHAR_INT_LIMIT=0x1016,

    UCHAR_GENERAL_CATEGORY_MASK=0x2000,
    UCHAR_MASK_START=UCHAR_GENERAL_CATEGORY_MASK,
    UCHAR_MASK_LIMIT=0x2001,

    UCHAR_NUMERIC_VALUE=0x3000,
    UCHAR_DOUBLE_START=UCHAR_NUMERIC_VALUE,
    UCHAR_DOUBLE_LIMIT=0x3001,

    UCHAR_AGE=0x4000,
    UCHAR_STRING_START=UCHAR_AGE,
    UCHAR_BIDI_MIRRORING_GLYPH=0x4001,
    UCHAR_CASE_FOLDING=0x4002,
#ifndef U_HIDE_DEPRECATED_API

    UCHAR_ISO_COMMENT=0x4003,
#endif  /* U_HIDE_DEPRECATED_API */

    UCHAR_LOWERCASE_MAPPING=0x4004,
    UCHAR_NAME=0x4005,
    UCHAR_SIMPLE_CASE_FOLDING=0x4006,
    UCHAR_SIMPLE_LOWERCASE_MAPPING=0x4007,
    UCHAR_SIMPLE_TITLECASE_MAPPING=0x4008,
    UCHAR_SIMPLE_UPPERCASE_MAPPING=0x4009,
    UCHAR_TITLECASE_MAPPING=0x400A,
#ifndef U_HIDE_DEPRECATED_API

    UCHAR_UNICODE_1_NAME=0x400B,
#endif  /* U_HIDE_DEPRECATED_API */

    UCHAR_UPPERCASE_MAPPING=0x400C,
    UCHAR_BIDI_PAIRED_BRACKET=0x400D,
    UCHAR_STRING_LIMIT=0x400E,

    UCHAR_SCRIPT_EXTENSIONS=0x7000,
    UCHAR_OTHER_PROPERTY_START=UCHAR_SCRIPT_EXTENSIONS,
    UCHAR_OTHER_PROPERTY_LIMIT=0x7001,
    UCHAR_INVALID_CODE = -1
} UProperty;

typedef enum UCharCategory
{
    /*
     * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
     * It matches pairs of lines like
     *     / ** <Unicode 2-letter General_Category value> comment... * /
     *     U_<[A-Z_]+> = <integer>,
     */

    U_UNASSIGNED              = 0,
    U_GENERAL_OTHER_TYPES     = 0,
    U_UPPERCASE_LETTER        = 1,
    U_LOWERCASE_LETTER        = 2,
    U_TITLECASE_LETTER        = 3,
    U_MODIFIER_LETTER         = 4,
    U_OTHER_LETTER            = 5,
    U_NON_SPACING_MARK        = 6,
    U_ENCLOSING_MARK          = 7,
    U_COMBINING_SPACING_MARK  = 8,
    U_DECIMAL_DIGIT_NUMBER    = 9,
    U_LETTER_NUMBER           = 10,
    U_OTHER_NUMBER            = 11,
    U_SPACE_SEPARATOR         = 12,
    U_LINE_SEPARATOR          = 13,
    U_PARAGRAPH_SEPARATOR     = 14,
    U_CONTROL_CHAR            = 15,
    U_FORMAT_CHAR             = 16,
    U_PRIVATE_USE_CHAR        = 17,
    U_SURROGATE               = 18,
    U_DASH_PUNCTUATION        = 19,
    U_START_PUNCTUATION       = 20,
    U_END_PUNCTUATION         = 21,
    U_CONNECTOR_PUNCTUATION   = 22,
    U_OTHER_PUNCTUATION       = 23,
    U_MATH_SYMBOL             = 24,
    U_CURRENCY_SYMBOL         = 25,
    U_MODIFIER_SYMBOL         = 26,
    U_OTHER_SYMBOL            = 27,
    U_INITIAL_PUNCTUATION     = 28,
    U_FINAL_PUNCTUATION       = 29,
    U_CHAR_CATEGORY_COUNT
} UCharCategory;

#define U_GC_CN_MASK    U_MASK(U_GENERAL_OTHER_TYPES)

#define U_GC_LU_MASK    U_MASK(U_UPPERCASE_LETTER)

#define U_GC_LL_MASK    U_MASK(U_LOWERCASE_LETTER)

#define U_GC_LT_MASK    U_MASK(U_TITLECASE_LETTER)

#define U_GC_LM_MASK    U_MASK(U_MODIFIER_LETTER)

#define U_GC_LO_MASK    U_MASK(U_OTHER_LETTER)

#define U_GC_MN_MASK    U_MASK(U_NON_SPACING_MARK)

#define U_GC_ME_MASK    U_MASK(U_ENCLOSING_MARK)

#define U_GC_MC_MASK    U_MASK(U_COMBINING_SPACING_MARK)

#define U_GC_ND_MASK    U_MASK(U_DECIMAL_DIGIT_NUMBER)

#define U_GC_NL_MASK    U_MASK(U_LETTER_NUMBER)

#define U_GC_NO_MASK    U_MASK(U_OTHER_NUMBER)

#define U_GC_ZS_MASK    U_MASK(U_SPACE_SEPARATOR)

#define U_GC_ZL_MASK    U_MASK(U_LINE_SEPARATOR)

#define U_GC_ZP_MASK    U_MASK(U_PARAGRAPH_SEPARATOR)

#define U_GC_CC_MASK    U_MASK(U_CONTROL_CHAR)

#define U_GC_CF_MASK    U_MASK(U_FORMAT_CHAR)

#define U_GC_CO_MASK    U_MASK(U_PRIVATE_USE_CHAR)

#define U_GC_CS_MASK    U_MASK(U_SURROGATE)

#define U_GC_PD_MASK    U_MASK(U_DASH_PUNCTUATION)

#define U_GC_PS_MASK    U_MASK(U_START_PUNCTUATION)

#define U_GC_PE_MASK    U_MASK(U_END_PUNCTUATION)

#define U_GC_PC_MASK    U_MASK(U_CONNECTOR_PUNCTUATION)

#define U_GC_PO_MASK    U_MASK(U_OTHER_PUNCTUATION)

#define U_GC_SM_MASK    U_MASK(U_MATH_SYMBOL)

#define U_GC_SC_MASK    U_MASK(U_CURRENCY_SYMBOL)

#define U_GC_SK_MASK    U_MASK(U_MODIFIER_SYMBOL)

#define U_GC_SO_MASK    U_MASK(U_OTHER_SYMBOL)

#define U_GC_PI_MASK    U_MASK(U_INITIAL_PUNCTUATION)

#define U_GC_PF_MASK    U_MASK(U_FINAL_PUNCTUATION)


#define U_GC_L_MASK \
            (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)

#define U_GC_LC_MASK \
            (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)

#define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)

#define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)

#define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)

#define U_GC_C_MASK \
            (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)

#define U_GC_P_MASK \
            (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
             U_GC_PI_MASK|U_GC_PF_MASK)

#define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)

typedef enum UCharDirection {
    /*
     * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
     * It matches pairs of lines like
     *     / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
     *     U_<[A-Z_]+> = <integer>,
     */

    U_LEFT_TO_RIGHT               = 0,
    U_RIGHT_TO_LEFT               = 1,
    U_EUROPEAN_NUMBER             = 2,
    U_EUROPEAN_NUMBER_SEPARATOR   = 3,
    U_EUROPEAN_NUMBER_TERMINATOR  = 4,
    U_ARABIC_NUMBER               = 5,
    U_COMMON_NUMBER_SEPARATOR     = 6,
    U_BLOCK_SEPARATOR             = 7,
    U_SEGMENT_SEPARATOR           = 8,
    U_WHITE_SPACE_NEUTRAL         = 9,
    U_OTHER_NEUTRAL               = 10,
    U_LEFT_TO_RIGHT_EMBEDDING     = 11,
    U_LEFT_TO_RIGHT_OVERRIDE      = 12,
    U_RIGHT_TO_LEFT_ARABIC        = 13,
    U_RIGHT_TO_LEFT_EMBEDDING     = 14,
    U_RIGHT_TO_LEFT_OVERRIDE      = 15,
    U_POP_DIRECTIONAL_FORMAT      = 16,
    U_DIR_NON_SPACING_MARK        = 17,
    U_BOUNDARY_NEUTRAL            = 18,
    U_FIRST_STRONG_ISOLATE        = 19,
    U_LEFT_TO_RIGHT_ISOLATE       = 20,
    U_RIGHT_TO_LEFT_ISOLATE       = 21,
    U_POP_DIRECTIONAL_ISOLATE     = 22,
    U_CHAR_DIRECTION_COUNT
} UCharDirection;

typedef enum UBidiPairedBracketType {
    /*
     * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
     */

    U_BPT_NONE,
    U_BPT_OPEN,
    U_BPT_CLOSE,
    U_BPT_COUNT /* 3 */
} UBidiPairedBracketType;

enum UBlockCode {
    /*
     * Note: UBlockCode constants are parsed by preparseucd.py.
     * It matches lines like
     *     UBLOCK_<Unicode Block value name> = <integer>,
     */

    UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */

    UBLOCK_BASIC_LATIN = 1, /*[0000]*/

    UBLOCK_LATIN_1_SUPPLEMENT=2, /*[0080]*/

    UBLOCK_LATIN_EXTENDED_A =3, /*[0100]*/

    UBLOCK_LATIN_EXTENDED_B =4, /*[0180]*/

    UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/

    UBLOCK_SPACING_MODIFIER_LETTERS =6, /*[02B0]*/

    UBLOCK_COMBINING_DIACRITICAL_MARKS =7, /*[0300]*/

    UBLOCK_GREEK =8, /*[0370]*/

    UBLOCK_CYRILLIC =9, /*[0400]*/

    UBLOCK_ARMENIAN =10, /*[0530]*/

    UBLOCK_HEBREW =11, /*[0590]*/

    UBLOCK_ARABIC =12, /*[0600]*/

    UBLOCK_SYRIAC =13, /*[0700]*/

    UBLOCK_THAANA =14, /*[0780]*/

    UBLOCK_DEVANAGARI =15, /*[0900]*/

    UBLOCK_BENGALI =16, /*[0980]*/

    UBLOCK_GURMUKHI =17, /*[0A00]*/

    UBLOCK_GUJARATI =18, /*[0A80]*/

    UBLOCK_ORIYA =19, /*[0B00]*/

    UBLOCK_TAMIL =20, /*[0B80]*/

    UBLOCK_TELUGU =21, /*[0C00]*/

    UBLOCK_KANNADA =22, /*[0C80]*/

    UBLOCK_MALAYALAM =23, /*[0D00]*/

    UBLOCK_SINHALA =24, /*[0D80]*/

    UBLOCK_THAI =25, /*[0E00]*/

    UBLOCK_LAO =26, /*[0E80]*/

    UBLOCK_TIBETAN =27, /*[0F00]*/

    UBLOCK_MYANMAR =28, /*[1000]*/

    UBLOCK_GEORGIAN =29, /*[10A0]*/

    UBLOCK_HANGUL_JAMO =30, /*[1100]*/

    UBLOCK_ETHIOPIC =31, /*[1200]*/

    UBLOCK_CHEROKEE =32, /*[13A0]*/

    UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS =33, /*[1400]*/

    UBLOCK_OGHAM =34, /*[1680]*/

    UBLOCK_RUNIC =35, /*[16A0]*/

    UBLOCK_KHMER =36, /*[1780]*/

    UBLOCK_MONGOLIAN =37, /*[1800]*/

    UBLOCK_LATIN_EXTENDED_ADDITIONAL =38, /*[1E00]*/

    UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/

    UBLOCK_GENERAL_PUNCTUATION =40, /*[2000]*/

    UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS =41, /*[2070]*/

    UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/

    UBLOCK_COMBINING_MARKS_FOR_SYMBOLS =43, /*[20D0]*/

    UBLOCK_LETTERLIKE_SYMBOLS =44, /*[2100]*/

    UBLOCK_NUMBER_FORMS =45, /*[2150]*/

    UBLOCK_ARROWS =46, /*[2190]*/

    UBLOCK_MATHEMATICAL_OPERATORS =47, /*[2200]*/

    UBLOCK_MISCELLANEOUS_TECHNICAL =48, /*[2300]*/

    UBLOCK_CONTROL_PICTURES =49, /*[2400]*/

    UBLOCK_OPTICAL_CHARACTER_RECOGNITION =50, /*[2440]*/

    UBLOCK_ENCLOSED_ALPHANUMERICS =51, /*[2460]*/

    UBLOCK_BOX_DRAWING =52, /*[2500]*/

    UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/

    UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/

    UBLOCK_MISCELLANEOUS_SYMBOLS =55, /*[2600]*/

    UBLOCK_DINGBATS =56, /*[2700]*/

    UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/

    UBLOCK_CJK_RADICALS_SUPPLEMENT =58, /*[2E80]*/

    UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/

    UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS =60, /*[2FF0]*/

    UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION =61, /*[3000]*/

    UBLOCK_HIRAGANA =62, /*[3040]*/

    UBLOCK_KATAKANA =63, /*[30A0]*/

    UBLOCK_BOPOMOFO =64, /*[3100]*/

    UBLOCK_HANGUL_COMPATIBILITY_JAMO =65, /*[3130]*/

    UBLOCK_KANBUN =66, /*[3190]*/

    UBLOCK_BOPOMOFO_EXTENDED =67, /*[31A0]*/

    UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS =68, /*[3200]*/

    UBLOCK_CJK_COMPATIBILITY =69, /*[3300]*/

    UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A =70, /*[3400]*/

    UBLOCK_CJK_UNIFIED_IDEOGRAPHS =71, /*[4E00]*/

    UBLOCK_YI_SYLLABLES =72, /*[A000]*/

    UBLOCK_YI_RADICALS =73, /*[A490]*/

    UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/

    UBLOCK_HIGH_SURROGATES =75, /*[D800]*/

    UBLOCK_HIGH_PRIVATE_USE_SURROGATES =76, /*[DB80]*/

    UBLOCK_LOW_SURROGATES =77, /*[DC00]*/

    UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
    UBLOCK_PRIVATE_USE = UBLOCK_PRIVATE_USE_AREA,

    UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS =79, /*[F900]*/

    UBLOCK_ALPHABETIC_PRESENTATION_FORMS =80, /*[FB00]*/

    UBLOCK_ARABIC_PRESENTATION_FORMS_A =81, /*[FB50]*/

    UBLOCK_COMBINING_HALF_MARKS =82, /*[FE20]*/

    UBLOCK_CJK_COMPATIBILITY_FORMS =83, /*[FE30]*/

    UBLOCK_SMALL_FORM_VARIANTS =84, /*[FE50]*/

    UBLOCK_ARABIC_PRESENTATION_FORMS_B =85, /*[FE70]*/

    UBLOCK_SPECIALS =86, /*[FFF0]*/

    UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS =87, /*[FF00]*/

    /* New blocks in Unicode 3.1 */

    UBLOCK_OLD_ITALIC = 88, /*[10300]*/
    UBLOCK_GOTHIC = 89, /*[10330]*/
    UBLOCK_DESERET = 90, /*[10400]*/
    UBLOCK_BYZANTINE_MUSICAL_SYMBOLS = 91, /*[1D000]*/
    UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
    UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS = 93, /*[1D400]*/
    UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B  = 94, /*[20000]*/
    UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT = 95, /*[2F800]*/
    UBLOCK_TAGS = 96, /*[E0000]*/

    /* New blocks in Unicode 3.2 */

    UBLOCK_CYRILLIC_SUPPLEMENT = 97, /*[0500]*/
    UBLOCK_CYRILLIC_SUPPLEMENTARY = UBLOCK_CYRILLIC_SUPPLEMENT,
    UBLOCK_TAGALOG = 98, /*[1700]*/
    UBLOCK_HANUNOO = 99, /*[1720]*/
    UBLOCK_BUHID = 100, /*[1740]*/
    UBLOCK_TAGBANWA = 101, /*[1760]*/
    UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A = 102, /*[27C0]*/
    UBLOCK_SUPPLEMENTAL_ARROWS_A = 103, /*[27F0]*/
    UBLOCK_SUPPLEMENTAL_ARROWS_B = 104, /*[2900]*/
    UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B = 105, /*[2980]*/
    UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS = 106, /*[2A00]*/
    UBLOCK_KATAKANA_PHONETIC_EXTENSIONS = 107, /*[31F0]*/
    UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
    UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A = 109, /*[F0000]*/
    UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B = 110, /*[100000]*/

    /* New blocks in Unicode 4 */

    UBLOCK_LIMBU = 111, /*[1900]*/
    UBLOCK_TAI_LE = 112, /*[1950]*/
    UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
    UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
    UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS = 115, /*[2B00]*/
    UBLOCK_YIJING_HEXAGRAM_SYMBOLS = 116, /*[4DC0]*/
    UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
    UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
    UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
    UBLOCK_UGARITIC = 120, /*[10380]*/
    UBLOCK_SHAVIAN = 121, /*[10450]*/
    UBLOCK_OSMANYA = 122, /*[10480]*/
    UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
    UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
    UBLOCK_VARIATION_SELECTORS_SUPPLEMENT = 125, /*[E0100]*/

    /* New blocks in Unicode 4.1 */

    UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION = 126, /*[1D200]*/
    UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
    UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
    UBLOCK_BUGINESE = 129, /*[1A00]*/
    UBLOCK_CJK_STROKES = 130, /*[31C0]*/
    UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT = 131, /*[1DC0]*/
    UBLOCK_COPTIC = 132, /*[2C80]*/
    UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
    UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
    UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
    UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
    UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
    UBLOCK_MODIFIER_TONE_LETTERS = 138, /*[A700]*/
    UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
    UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
    UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT = 141, /*[1D80]*/
    UBLOCK_SUPPLEMENTAL_PUNCTUATION = 142, /*[2E00]*/
    UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
    UBLOCK_TIFINAGH = 144, /*[2D30]*/
    UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/

    /* New blocks in Unicode 5.0 */

    UBLOCK_NKO = 146, /*[07C0]*/
    UBLOCK_BALINESE = 147, /*[1B00]*/
    UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
    UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
    UBLOCK_PHAGS_PA = 150, /*[A840]*/
    UBLOCK_PHOENICIAN = 151, /*[10900]*/
    UBLOCK_CUNEIFORM = 152, /*[12000]*/
    UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION = 153, /*[12400]*/
    UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/

    /* New blocks in Unicode 5.1 */

    UBLOCK_SUNDANESE = 155, /*[1B80]*/
    UBLOCK_LEPCHA = 156, /*[1C00]*/
    UBLOCK_OL_CHIKI = 157, /*[1C50]*/
    UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
    UBLOCK_VAI = 159, /*[A500]*/
    UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
    UBLOCK_SAURASHTRA = 161, /*[A880]*/
    UBLOCK_KAYAH_LI = 162, /*[A900]*/
    UBLOCK_REJANG = 163, /*[A930]*/
    UBLOCK_CHAM = 164, /*[AA00]*/
    UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
    UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
    UBLOCK_LYCIAN = 167, /*[10280]*/
    UBLOCK_CARIAN = 168, /*[102A0]*/
    UBLOCK_LYDIAN = 169, /*[10920]*/
    UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
    UBLOCK_DOMINO_TILES = 171, /*[1F030]*/

    /* New blocks in Unicode 5.2 */

    UBLOCK_SAMARITAN = 172, /*[0800]*/
    UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED = 173, /*[18B0]*/
    UBLOCK_TAI_THAM = 174, /*[1A20]*/
    UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
    UBLOCK_LISU = 176, /*[A4D0]*/
    UBLOCK_BAMUM = 177, /*[A6A0]*/
    UBLOCK_COMMON_INDIC_NUMBER_FORMS = 178, /*[A830]*/
    UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
    UBLOCK_HANGUL_JAMO_EXTENDED_A = 180, /*[A960]*/
    UBLOCK_JAVANESE = 181, /*[A980]*/
    UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
    UBLOCK_TAI_VIET = 183, /*[AA80]*/
    UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
    UBLOCK_HANGUL_JAMO_EXTENDED_B = 185, /*[D7B0]*/
    UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
    UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
    UBLOCK_AVESTAN = 188, /*[10B00]*/
    UBLOCK_INSCRIPTIONAL_PARTHIAN = 189, /*[10B40]*/
    UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
    UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
    UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
    UBLOCK_KAITHI = 193, /*[11080]*/
    UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
    UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT = 195, /*[1F100]*/
    UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT = 196, /*[1F200]*/
    UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C = 197, /*[2A700]*/

    /* New blocks in Unicode 6.0 */

    UBLOCK_MANDAIC = 198, /*[0840]*/
    UBLOCK_BATAK = 199, /*[1BC0]*/
    UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
    UBLOCK_BRAHMI = 201, /*[11000]*/
    UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
    UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
    UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
    UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS = 205, /*[1F300]*/
    UBLOCK_EMOTICONS = 206, /*[1F600]*/
    UBLOCK_TRANSPORT_AND_MAP_SYMBOLS = 207, /*[1F680]*/
    UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
    UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D = 209, /*[2B740]*/

    /* New blocks in Unicode 6.1 */

    UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
    UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS = 211, /*[1EE00]*/
    UBLOCK_CHAKMA = 212, /*[11100]*/
    UBLOCK_MEETEI_MAYEK_EXTENSIONS = 213, /*[AAE0]*/
    UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
    UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
    UBLOCK_MIAO = 216, /*[16F00]*/
    UBLOCK_SHARADA = 217, /*[11180]*/
    UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
    UBLOCK_SUNDANESE_SUPPLEMENT = 219, /*[1CC0]*/
    UBLOCK_TAKRI = 220, /*[11680]*/

    /* New blocks in Unicode 7.0 */

    UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
    UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
    UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
    UBLOCK_COMBINING_DIACRITICAL_MARKS_EXTENDED = 224, /*[1AB0]*/
    UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
    UBLOCK_ELBASAN = 226, /*[10500]*/
    UBLOCK_GEOMETRIC_SHAPES_EXTENDED = 227, /*[1F780]*/
    UBLOCK_GRANTHA = 228, /*[11300]*/
    UBLOCK_KHOJKI = 229, /*[11200]*/
    UBLOCK_KHUDAWADI = 230, /*[112B0]*/
    UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
    UBLOCK_LINEAR_A = 232, /*[10600]*/
    UBLOCK_MAHAJANI = 233, /*[11150]*/
    UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
    UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
    UBLOCK_MODI = 236, /*[11600]*/
    UBLOCK_MRO = 237, /*[16A40]*/
    UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
    UBLOCK_NABATAEAN = 239, /*[10880]*/
    UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
    UBLOCK_OLD_PERMIC = 241, /*[10350]*/
    UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
    UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
    UBLOCK_PALMYRENE = 244, /*[10860]*/
    UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
    UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
    UBLOCK_SHORTHAND_FORMAT_CONTROLS = 247, /*[1BCA0]*/
    UBLOCK_SIDDHAM = 248, /*[11580]*/
    UBLOCK_SINHALA_ARCHAIC_NUMBERS = 249, /*[111E0]*/
    UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
    UBLOCK_TIRHUTA = 251, /*[11480]*/
    UBLOCK_WARANG_CITI = 252, /*[118A0]*/

    /* New blocks in Unicode 8.0 */

    UBLOCK_AHOM = 253, /*[11700]*/
    UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
    UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
    UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E = 256, /*[2B820]*/
    UBLOCK_EARLY_DYNASTIC_CUNEIFORM = 257, /*[12480]*/
    UBLOCK_HATRAN = 258, /*[108E0]*/
    UBLOCK_MULTANI = 259, /*[11280]*/
    UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
    UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS = 261, /*[1F900]*/
    UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/

    UBLOCK_COUNT = 263,

    UBLOCK_INVALID_CODE=-1
};

typedef enum UBlockCode UBlockCode;

typedef enum UEastAsianWidth {
    /*
     * Note: UEastAsianWidth constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_EA_<Unicode East_Asian_Width value name>
     */

    U_EA_NEUTRAL,   /*[N]*/
    U_EA_AMBIGUOUS, /*[A]*/
    U_EA_HALFWIDTH, /*[H]*/
    U_EA_FULLWIDTH, /*[F]*/
    U_EA_NARROW,    /*[Na]*/
    U_EA_WIDE,      /*[W]*/
    U_EA_COUNT
} UEastAsianWidth;

typedef enum UCharNameChoice {
    U_UNICODE_CHAR_NAME,
#ifndef U_HIDE_DEPRECATED_API

    U_UNICODE_10_CHAR_NAME,
#endif  /* U_HIDE_DEPRECATED_API */

    U_EXTENDED_CHAR_NAME = U_UNICODE_CHAR_NAME+2,
    U_CHAR_NAME_ALIAS,
    U_CHAR_NAME_CHOICE_COUNT
} UCharNameChoice;

typedef enum UPropertyNameChoice {
    U_SHORT_PROPERTY_NAME,
    U_LONG_PROPERTY_NAME,
    U_PROPERTY_NAME_CHOICE_COUNT
} UPropertyNameChoice;

typedef enum UDecompositionType {
    /*
     * Note: UDecompositionType constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_DT_<Unicode Decomposition_Type value name>
     */

    U_DT_NONE,              /*[none]*/
    U_DT_CANONICAL,         /*[can]*/
    U_DT_COMPAT,            /*[com]*/
    U_DT_CIRCLE,            /*[enc]*/
    U_DT_FINAL,             /*[fin]*/
    U_DT_FONT,              /*[font]*/
    U_DT_FRACTION,          /*[fra]*/
    U_DT_INITIAL,           /*[init]*/
    U_DT_ISOLATED,          /*[iso]*/
    U_DT_MEDIAL,            /*[med]*/
    U_DT_NARROW,            /*[nar]*/
    U_DT_NOBREAK,           /*[nb]*/
    U_DT_SMALL,             /*[sml]*/
    U_DT_SQUARE,            /*[sqr]*/
    U_DT_SUB,               /*[sub]*/
    U_DT_SUPER,             /*[sup]*/
    U_DT_VERTICAL,          /*[vert]*/
    U_DT_WIDE,              /*[wide]*/
    U_DT_COUNT /* 18 */
} UDecompositionType;

typedef enum UJoiningType {
    /*
     * Note: UJoiningType constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_JT_<Unicode Joining_Type value name>
     */

    U_JT_NON_JOINING,       /*[U]*/
    U_JT_JOIN_CAUSING,      /*[C]*/
    U_JT_DUAL_JOINING,      /*[D]*/
    U_JT_LEFT_JOINING,      /*[L]*/
    U_JT_RIGHT_JOINING,     /*[R]*/
    U_JT_TRANSPARENT,       /*[T]*/
    U_JT_COUNT /* 6 */
} UJoiningType;

typedef enum UJoiningGroup {
    /*
     * Note: UJoiningGroup constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_JG_<Unicode Joining_Group value name>
     */

    U_JG_NO_JOINING_GROUP,
    U_JG_AIN,
    U_JG_ALAPH,
    U_JG_ALEF,
    U_JG_BEH,
    U_JG_BETH,
    U_JG_DAL,
    U_JG_DALATH_RISH,
    U_JG_E,
    U_JG_FEH,
    U_JG_FINAL_SEMKATH,
    U_JG_GAF,
    U_JG_GAMAL,
    U_JG_HAH,
    U_JG_TEH_MARBUTA_GOAL,  
    U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
    U_JG_HE,
    U_JG_HEH,
    U_JG_HEH_GOAL,
    U_JG_HETH,
    U_JG_KAF,
    U_JG_KAPH,
    U_JG_KNOTTED_HEH,
    U_JG_LAM,
    U_JG_LAMADH,
    U_JG_MEEM,
    U_JG_MIM,
    U_JG_NOON,
    U_JG_NUN,
    U_JG_PE,
    U_JG_QAF,
    U_JG_QAPH,
    U_JG_REH,
    U_JG_REVERSED_PE,
    U_JG_SAD,
    U_JG_SADHE,
    U_JG_SEEN,
    U_JG_SEMKATH,
    U_JG_SHIN,
    U_JG_SWASH_KAF,
    U_JG_SYRIAC_WAW,
    U_JG_TAH,
    U_JG_TAW,
    U_JG_TEH_MARBUTA,
    U_JG_TETH,
    U_JG_WAW,
    U_JG_YEH,
    U_JG_YEH_BARREE,
    U_JG_YEH_WITH_TAIL,
    U_JG_YUDH,
    U_JG_YUDH_HE,
    U_JG_ZAIN,
    U_JG_FE,        
    U_JG_KHAPH,     
    U_JG_ZHAIN,     
    U_JG_BURUSHASKI_YEH_BARREE, 
    U_JG_FARSI_YEH, 
    U_JG_NYA,       
    U_JG_ROHINGYA_YEH,  
    U_JG_MANICHAEAN_ALEPH,  
    U_JG_MANICHAEAN_AYIN,  
    U_JG_MANICHAEAN_BETH,  
    U_JG_MANICHAEAN_DALETH,  
    U_JG_MANICHAEAN_DHAMEDH,  
    U_JG_MANICHAEAN_FIVE,  
    U_JG_MANICHAEAN_GIMEL,  
    U_JG_MANICHAEAN_HETH,  
    U_JG_MANICHAEAN_HUNDRED,  
    U_JG_MANICHAEAN_KAPH,  
    U_JG_MANICHAEAN_LAMEDH,  
    U_JG_MANICHAEAN_MEM,  
    U_JG_MANICHAEAN_NUN,  
    U_JG_MANICHAEAN_ONE,  
    U_JG_MANICHAEAN_PE,  
    U_JG_MANICHAEAN_QOPH,  
    U_JG_MANICHAEAN_RESH,  
    U_JG_MANICHAEAN_SADHE,  
    U_JG_MANICHAEAN_SAMEKH,  
    U_JG_MANICHAEAN_TAW,  
    U_JG_MANICHAEAN_TEN,  
    U_JG_MANICHAEAN_TETH,  
    U_JG_MANICHAEAN_THAMEDH,  
    U_JG_MANICHAEAN_TWENTY,  
    U_JG_MANICHAEAN_WAW,  
    U_JG_MANICHAEAN_YODH,  
    U_JG_MANICHAEAN_ZAYIN,  
    U_JG_STRAIGHT_WAW,  
    U_JG_COUNT
} UJoiningGroup;

typedef enum UGraphemeClusterBreak {
    /*
     * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_GCB_<Unicode Grapheme_Cluster_Break value name>
     */

    U_GCB_OTHER = 0,            /*[XX]*/
    U_GCB_CONTROL = 1,          /*[CN]*/
    U_GCB_CR = 2,               /*[CR]*/
    U_GCB_EXTEND = 3,           /*[EX]*/
    U_GCB_L = 4,                /*[L]*/
    U_GCB_LF = 5,               /*[LF]*/
    U_GCB_LV = 6,               /*[LV]*/
    U_GCB_LVT = 7,              /*[LVT]*/
    U_GCB_T = 8,                /*[T]*/
    U_GCB_V = 9,                /*[V]*/
    U_GCB_SPACING_MARK = 10,    /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
    U_GCB_PREPEND = 11,         /*[PP]*/
    U_GCB_REGIONAL_INDICATOR = 12,  /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
    U_GCB_COUNT = 13
} UGraphemeClusterBreak;

typedef enum UWordBreakValues {
    /*
     * Note: UWordBreakValues constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_WB_<Unicode Word_Break value name>
     */

    U_WB_OTHER = 0,             /*[XX]*/
    U_WB_ALETTER = 1,           /*[LE]*/
    U_WB_FORMAT = 2,            /*[FO]*/
    U_WB_KATAKANA = 3,          /*[KA]*/
    U_WB_MIDLETTER = 4,         /*[ML]*/
    U_WB_MIDNUM = 5,            /*[MN]*/
    U_WB_NUMERIC = 6,           /*[NU]*/
    U_WB_EXTENDNUMLET = 7,      /*[EX]*/
    U_WB_CR = 8,                /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
    U_WB_EXTEND = 9,            /*[Extend]*/
    U_WB_LF = 10,               /*[LF]*/
    U_WB_MIDNUMLET =11,         /*[MB]*/
    U_WB_NEWLINE =12,           /*[NL]*/
    U_WB_REGIONAL_INDICATOR = 13,   /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
    U_WB_HEBREW_LETTER = 14,    /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
    U_WB_SINGLE_QUOTE = 15,     /*[SQ]*/
    U_WB_DOUBLE_QUOTE = 16,     /*[DQ]*/
    U_WB_COUNT = 17
} UWordBreakValues;

typedef enum USentenceBreak {
    /*
     * Note: USentenceBreak constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_SB_<Unicode Sentence_Break value name>
     */

    U_SB_OTHER = 0,             /*[XX]*/
    U_SB_ATERM = 1,             /*[AT]*/
    U_SB_CLOSE = 2,             /*[CL]*/
    U_SB_FORMAT = 3,            /*[FO]*/
    U_SB_LOWER = 4,             /*[LO]*/
    U_SB_NUMERIC = 5,           /*[NU]*/
    U_SB_OLETTER = 6,           /*[LE]*/
    U_SB_SEP = 7,               /*[SE]*/
    U_SB_SP = 8,                /*[SP]*/
    U_SB_STERM = 9,             /*[ST]*/
    U_SB_UPPER = 10,            /*[UP]*/
    U_SB_CR = 11,               /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
    U_SB_EXTEND = 12,           /*[EX]*/
    U_SB_LF = 13,               /*[LF]*/
    U_SB_SCONTINUE = 14,        /*[SC]*/
    U_SB_COUNT = 15
} USentenceBreak;

typedef enum ULineBreak {
    /*
     * Note: ULineBreak constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_LB_<Unicode Line_Break value name>
     */

    U_LB_UNKNOWN = 0,           /*[XX]*/
    U_LB_AMBIGUOUS = 1,         /*[AI]*/
    U_LB_ALPHABETIC = 2,        /*[AL]*/
    U_LB_BREAK_BOTH = 3,        /*[B2]*/
    U_LB_BREAK_AFTER = 4,       /*[BA]*/
    U_LB_BREAK_BEFORE = 5,      /*[BB]*/
    U_LB_MANDATORY_BREAK = 6,   /*[BK]*/
    U_LB_CONTINGENT_BREAK = 7,  /*[CB]*/
    U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
    U_LB_COMBINING_MARK = 9,    /*[CM]*/
    U_LB_CARRIAGE_RETURN = 10,   /*[CR]*/
    U_LB_EXCLAMATION = 11,       /*[EX]*/
    U_LB_GLUE = 12,              /*[GL]*/
    U_LB_HYPHEN = 13,            /*[HY]*/
    U_LB_IDEOGRAPHIC = 14,       /*[ID]*/
    U_LB_INSEPARABLE = 15,       /*[IN]*/
    U_LB_INSEPERABLE = U_LB_INSEPARABLE,
    U_LB_INFIX_NUMERIC = 16,     /*[IS]*/
    U_LB_LINE_FEED = 17,         /*[LF]*/
    U_LB_NONSTARTER = 18,        /*[NS]*/
    U_LB_NUMERIC = 19,           /*[NU]*/
    U_LB_OPEN_PUNCTUATION = 20,  /*[OP]*/
    U_LB_POSTFIX_NUMERIC = 21,   /*[PO]*/
    U_LB_PREFIX_NUMERIC = 22,    /*[PR]*/
    U_LB_QUOTATION = 23,         /*[QU]*/
    U_LB_COMPLEX_CONTEXT = 24,   /*[SA]*/
    U_LB_SURROGATE = 25,         /*[SG]*/
    U_LB_SPACE = 26,             /*[SP]*/
    U_LB_BREAK_SYMBOLS = 27,     /*[SY]*/
    U_LB_ZWSPACE = 28,           /*[ZW]*/
    U_LB_NEXT_LINE = 29,         /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
    U_LB_WORD_JOINER = 30,       /*[WJ]*/
    U_LB_H2 = 31,                /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
    U_LB_H3 = 32,                /*[H3]*/
    U_LB_JL = 33,                /*[JL]*/
    U_LB_JT = 34,                /*[JT]*/
    U_LB_JV = 35,                /*[JV]*/
    U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
    U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
    U_LB_HEBREW_LETTER = 38,     /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
    U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
    U_LB_COUNT = 40
} ULineBreak;

typedef enum UNumericType {
    /*
     * Note: UNumericType constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_NT_<Unicode Numeric_Type value name>
     */

    U_NT_NONE,              /*[None]*/
    U_NT_DECIMAL,           /*[de]*/
    U_NT_DIGIT,             /*[di]*/
    U_NT_NUMERIC,           /*[nu]*/
    U_NT_COUNT
} UNumericType;

typedef enum UHangulSyllableType {
    /*
     * Note: UHangulSyllableType constants are parsed by preparseucd.py.
     * It matches lines like
     *     U_HST_<Unicode Hangul_Syllable_Type value name>
     */

    U_HST_NOT_APPLICABLE,   /*[NA]*/
    U_HST_LEADING_JAMO,     /*[L]*/
    U_HST_VOWEL_JAMO,       /*[V]*/
    U_HST_TRAILING_JAMO,    /*[T]*/
    U_HST_LV_SYLLABLE,      /*[LV]*/
    U_HST_LVT_SYLLABLE,     /*[LVT]*/
    U_HST_COUNT
} UHangulSyllableType;

U_STABLE UBool U_EXPORT2
u_hasBinaryProperty(UChar32 c, UProperty which);

U_STABLE UBool U_EXPORT2
u_isUAlphabetic(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isULowercase(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isUUppercase(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isUWhiteSpace(UChar32 c);

U_STABLE int32_t U_EXPORT2
u_getIntPropertyValue(UChar32 c, UProperty which);

U_STABLE int32_t U_EXPORT2
u_getIntPropertyMinValue(UProperty which);

U_STABLE int32_t U_EXPORT2
u_getIntPropertyMaxValue(UProperty which);

U_STABLE double U_EXPORT2
u_getNumericValue(UChar32 c);

#define U_NO_NUMERIC_VALUE ((double)-123456789.)

U_STABLE UBool U_EXPORT2
u_islower(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isupper(UChar32 c);

U_STABLE UBool U_EXPORT2
u_istitle(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isdigit(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isalpha(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isalnum(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isxdigit(UChar32 c);

U_STABLE UBool U_EXPORT2
u_ispunct(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isgraph(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isblank(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isdefined(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isspace(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isJavaSpaceChar(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isWhitespace(UChar32 c);

U_STABLE UBool U_EXPORT2
u_iscntrl(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isISOControl(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isprint(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isbase(UChar32 c);

U_STABLE UCharDirection U_EXPORT2
u_charDirection(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isMirrored(UChar32 c);

U_STABLE UChar32 U_EXPORT2
u_charMirror(UChar32 c);

U_STABLE UChar32 U_EXPORT2
u_getBidiPairedBracket(UChar32 c);

U_STABLE int8_t U_EXPORT2
u_charType(UChar32 c);

#define U_GET_GC_MASK(c) U_MASK(u_charType(c))

typedef UBool U_CALLCONV
UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);

U_STABLE void U_EXPORT2
u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);

#if !UCONFIG_NO_NORMALIZATION

U_STABLE uint8_t U_EXPORT2
u_getCombiningClass(UChar32 c);

#endif

U_STABLE int32_t U_EXPORT2
u_charDigitValue(UChar32 c);

U_STABLE UBlockCode U_EXPORT2
ublock_getCode(UChar32 c);

U_STABLE int32_t U_EXPORT2
u_charName(UChar32 code, UCharNameChoice nameChoice,
           char *buffer, int32_t bufferLength,
           UErrorCode *pErrorCode);

#ifndef U_HIDE_DEPRECATED_API

U_DEPRECATED int32_t U_EXPORT2
u_getISOComment(UChar32 c,
                char *dest, int32_t destCapacity,
                UErrorCode *pErrorCode);
#endif  /* U_HIDE_DEPRECATED_API */

U_STABLE UChar32 U_EXPORT2
u_charFromName(UCharNameChoice nameChoice,
               const char *name,
               UErrorCode *pErrorCode);

typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
                               UChar32 code,
                               UCharNameChoice nameChoice,
                               const char *name,
                               int32_t length);

U_STABLE void U_EXPORT2
u_enumCharNames(UChar32 start, UChar32 limit,
                UEnumCharNamesFn *fn,
                void *context,
                UCharNameChoice nameChoice,
                UErrorCode *pErrorCode);

U_STABLE const char* U_EXPORT2
u_getPropertyName(UProperty property,
                  UPropertyNameChoice nameChoice);

U_STABLE UProperty U_EXPORT2
u_getPropertyEnum(const char* alias);

U_STABLE const char* U_EXPORT2
u_getPropertyValueName(UProperty property,
                       int32_t value,
                       UPropertyNameChoice nameChoice);

U_STABLE int32_t U_EXPORT2
u_getPropertyValueEnum(UProperty property,
                       const char* alias);

U_STABLE UBool U_EXPORT2
u_isIDStart(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isIDPart(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isIDIgnorable(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isJavaIDStart(UChar32 c);

U_STABLE UBool U_EXPORT2
u_isJavaIDPart(UChar32 c);

U_STABLE UChar32 U_EXPORT2
u_tolower(UChar32 c);

U_STABLE UChar32 U_EXPORT2
u_toupper(UChar32 c);

U_STABLE UChar32 U_EXPORT2
u_totitle(UChar32 c);

#define U_FOLD_CASE_DEFAULT 0

#define U_FOLD_CASE_EXCLUDE_SPECIAL_I 1

U_STABLE UChar32 U_EXPORT2
u_foldCase(UChar32 c, uint32_t options);

U_STABLE int32_t U_EXPORT2
u_digit(UChar32 ch, int8_t radix);

U_STABLE UChar32 U_EXPORT2
u_forDigit(int32_t digit, int8_t radix);

U_STABLE void U_EXPORT2
u_charAge(UChar32 c, UVersionInfo versionArray);

U_STABLE void U_EXPORT2
u_getUnicodeVersion(UVersionInfo versionArray);

#if !UCONFIG_NO_NORMALIZATION

U_STABLE int32_t U_EXPORT2
u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);

#endif


U_CDECL_END

#endif /*_UCHAR*/
/*eof*/