uchar.h

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

    /** @stable ICU 2.6 */
    UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
    /** @stable ICU 2.6 */
    UBLOCK_VARIATION_SELECTORS_SUPPLEMENT = 125, /*[E0100]*/

    /** @stable ICU 2.0 */
    UBLOCK_COUNT,

    /** @stable ICU 2.0 */
    UBLOCK_INVALID_CODE=-1
};

/** @stable ICU 2.0 */
typedef enum UBlockCode UBlockCode;

/**
 * East Asian Width constants.
 *
 * @see UCHAR_EAST_ASIAN_WIDTH
 * @see u_getIntPropertyValue
 * @stable ICU 2.2
 */
typedef enum UEastAsianWidth {
    U_EA_NEUTRAL,   /*[N]*/ /*See note !!*/
    U_EA_AMBIGUOUS, /*[A]*/
    U_EA_HALFWIDTH, /*[H]*/
    U_EA_FULLWIDTH, /*[F]*/
    U_EA_NARROW,    /*[Na]*/
    U_EA_WIDE,      /*[W]*/
    U_EA_COUNT
} UEastAsianWidth;
/*
 * Implementation note:
 * Keep UEastAsianWidth constant values in sync with names list in genprops/props2.c.
 */

/**
 * Selector constants for u_charName().
 * u_charName() returns the "modern" name of a
 * Unicode character; or the name that was defined in
 * Unicode version 1.0, before the Unicode standard merged
 * with ISO-10646; or an "extended" name that gives each
 * Unicode code point a unique name.
 *
 * @see u_charName
 * @stable ICU 2.0
 */
typedef enum UCharNameChoice {
    U_UNICODE_CHAR_NAME,
    U_UNICODE_10_CHAR_NAME,
    U_EXTENDED_CHAR_NAME,
    U_CHAR_NAME_CHOICE_COUNT
} UCharNameChoice;

/**
 * Selector constants for u_getPropertyName() and
 * u_getPropertyValueName().  These selectors are used to choose which
 * name is returned for a given property or value.  All properties and
 * values have a long name.  Most have a short name, but some do not.
 * Unicode allows for additional names, beyond the long and short
 * name, which would be indicated by U_LONG_PROPERTY_NAME + i, where
 * i=1, 2,...
 *
 * @see u_getPropertyName()
 * @see u_getPropertyValueName()
 * @stable ICU 2.4
 */
typedef enum UPropertyNameChoice {
    U_SHORT_PROPERTY_NAME,
    U_LONG_PROPERTY_NAME,
    U_PROPERTY_NAME_CHOICE_COUNT
} UPropertyNameChoice;

/**
 * Decomposition Type constants.
 *
 * @see UCHAR_DECOMPOSITION_TYPE
 * @stable ICU 2.2
 */
typedef enum UDecompositionType {
    U_DT_NONE,              /*[none]*/ /*See note !!*/
    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;

/**
 * Joining Type constants.
 *
 * @see UCHAR_JOINING_TYPE
 * @stable ICU 2.2
 */
typedef enum UJoiningType {
    U_JT_NON_JOINING,       /*[U]*/ /*See note !!*/
    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;

/**
 * Joining Group constants.
 *
 * @see UCHAR_JOINING_GROUP
 * @stable ICU 2.2
 */
typedef enum UJoiningGroup {
    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_HAMZA_ON_HEH_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,        /**< @stable ICU 2.6 */
    U_JG_KHAPH,     /**< @stable ICU 2.6 */
    U_JG_ZHAIN,     /**< @stable ICU 2.6 */
    U_JG_COUNT
} UJoiningGroup;

/**
 * Line Break constants.
 *
 * @see UCHAR_LINE_BREAK
 * @stable ICU 2.2
 */
typedef enum ULineBreak {
    U_LB_UNKNOWN,           /*[XX]*/ /*See note !!*/
    U_LB_AMBIGUOUS,         /*[AI]*/
    U_LB_ALPHABETIC,        /*[AL]*/
    U_LB_BREAK_BOTH,        /*[B2]*/
    U_LB_BREAK_AFTER,       /*[BA]*/
    U_LB_BREAK_BEFORE,      /*[BB]*/
    U_LB_MANDATORY_BREAK,   /*[BK]*/
    U_LB_CONTINGENT_BREAK,  /*[CB]*/
    U_LB_CLOSE_PUNCTUATION, /*[CL]*/
    U_LB_COMBINING_MARK,    /*[CM]*/
    U_LB_CARRIAGE_RETURN,   /*[CR]*/
    U_LB_EXCLAMATION,       /*[EX]*/
    U_LB_GLUE,              /*[GL]*/
    U_LB_HYPHEN,            /*[HY]*/
    U_LB_IDEOGRAPHIC,       /*[ID]*/
    U_LB_INSEPERABLE,
    /** Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0 @draft ICU 3.0 */
    U_LB_INSEPARABLE=U_LB_INSEPERABLE,/*[IN]*/
    U_LB_INFIX_NUMERIC,     /*[IS]*/
    U_LB_LINE_FEED,         /*[LF]*/
    U_LB_NONSTARTER,        /*[NS]*/
    U_LB_NUMERIC,           /*[NU]*/
    U_LB_OPEN_PUNCTUATION,  /*[OP]*/
    U_LB_POSTFIX_NUMERIC,   /*[PO]*/
    U_LB_PREFIX_NUMERIC,    /*[PR]*/
    U_LB_QUOTATION,         /*[QU]*/
    U_LB_COMPLEX_CONTEXT,   /*[SA]*/
    U_LB_SURROGATE,         /*[SG]*/
    U_LB_SPACE,             /*[SP]*/
    U_LB_BREAK_SYMBOLS,     /*[SY]*/
    U_LB_ZWSPACE,           /*[ZW]*/
    U_LB_NEXT_LINE,         /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
    U_LB_WORD_JOINER,       /*[WJ]*/
    U_LB_COUNT
} ULineBreak;

/**
 * Numeric Type constants.
 *
 * @see UCHAR_NUMERIC_TYPE
 * @stable ICU 2.2
 */
typedef enum UNumericType {
    U_NT_NONE,              /*[None]*/ /*See note !!*/
    U_NT_DECIMAL,           /*[de]*/
    U_NT_DIGIT,             /*[di]*/
    U_NT_NUMERIC,           /*[nu]*/
    U_NT_COUNT
} UNumericType;

/**
 * Hangul Syllable Type constants.
 *
 * @see UCHAR_HANGUL_SYLLABLE_TYPE
 * @stable ICU 2.6
 */
typedef enum UHangulSyllableType {
    U_HST_NOT_APPLICABLE,   /*[NA]*/ /*See note !!*/
    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;

/**
 * Check a binary Unicode property for a code point.
 *
 * Unicode, especially in version 3.2, defines many more properties than the
 * original set in UnicodeData.txt.
 *
 * The properties APIs are intended to reflect Unicode properties as defined
 * in the Unicode Character Database (UCD) and Unicode Technical Reports (UTR).
 * For details about the properties see http://www.unicode.org/ucd/ .
 * For names of Unicode properties see the UCD file PropertyAliases.txt.
 *
 * Important: If ICU is built with UCD files from Unicode versions below 3.2,
 * then properties marked with "new in Unicode 3.2" are not or not fully available.
 *
 * @param c Code point to test.
 * @param which UProperty selector constant, identifies which binary property to check.
 *        Must be UCHAR_BINARY_START<=which<UCHAR_BINARY_LIMIT.
 * @return TRUE or FALSE according to the binary Unicode property value for c.
 *         Also FALSE if 'which' is out of bounds or if the Unicode version
 *         does not have data for the property at all, or not for this code point.
 *
 * @see UProperty
 * @see u_getIntPropertyValue
 * @see u_getUnicodeVersion
 * @stable ICU 2.1
 */
U_STABLE UBool U_EXPORT2
u_hasBinaryProperty(UChar32 c, UProperty which);

/**
 * Check if a code point has the Alphabetic Unicode property.
 * Same as u_hasBinaryProperty(c, UCHAR_ALPHABETIC).
 * This is different from u_isalpha!
 * @param c Code point to test
 * @return true if the code point has the Alphabetic Unicode property, false otherwise
 *
 * @see UCHAR_ALPHABETIC
 * @see u_isalpha
 * @see u_hasBinaryProperty
 * @stable ICU 2.1
 */
U_STABLE UBool U_EXPORT2
u_isUAlphabetic(UChar32 c);

/**
 * Check if a code point has the Lowercase Unicode property.
 * Same as u_hasBinaryProperty(c, UCHAR_LOWERCASE).
 * This is different from u_islower!
 * @param c Code point to test
 * @return true if the code point has the Lowercase Unicode property, false otherwise
 *
 * @see UCHAR_LOWERCASE
 * @see u_islower
 * @see u_hasBinaryProperty
 * @stable ICU 2.1
 */
U_STABLE UBool U_EXPORT2
u_isULowercase(UChar32 c);

/**
 * Check if a code point has the Uppercase Unicode property.
 * Same as u_hasBinaryProperty(c, UCHAR_UPPERCASE).
 * This is different from u_isupper!
 * @param c Code point to test
 * @return true if the code point has the Uppercase Unicode property, false otherwise
 *
 * @see UCHAR_UPPERCASE
 * @see u_isupper
 * @see u_hasBinaryProperty
 * @stable ICU 2.1
 */
U_STABLE UBool U_EXPORT2
u_isUUppercase(UChar32 c);

/**
 * Check if a code point has the White_Space Unicode property.
 * Same as u_hasBinaryProperty(c, UCHAR_WHITE_SPACE).
 * This is different from both u_isspace and u_isWhitespace!
 *
 * Note: There are several ICU whitespace functions; please see the uchar.h
 * file documentation for a detailed comparison.
 *
 * @param c Code point to test
 * @return true if the code point has the White_Space Unicode property, false otherwise.
 *
 * @see UCHAR_WHITE_SPACE
 * @see u_isWhitespace
 * @see u_isspace
 * @see u_isJavaSpaceChar
 * @see u_hasBinaryProperty
 * @stable ICU 2.1
 */
U_STABLE UBool U_EXPORT2
u_isUWhiteSpace(UChar32 c);

/**
 * Get the property value for an enumerated or integer Unicode property for a code point.
 * Also returns binary and mask property values.
 *
 * Unicode, especially in version 3.2, defines many more properties than the
 * original set in UnicodeData.txt.
 *
 * The properties APIs are intended to reflect Unicode properties as defined
 * in the Unicode Character Database (UCD) and Unicode Technical Reports (UTR).
 * For details about the properties see http://www.unicode.org/ .
 * For names of Unicode properties see the UCD file PropertyAliases.txt.
 *
 * Sample usage:
 * UEastAsianWidth ea=(UEastAsianWidth)u_getIntPropertyValue(c, UCHAR_EAST_ASIAN_WIDTH);
 * UBool b=(UBool)u_getIntPropertyValue(c, UCHAR_IDEOGRAPHIC);
 *
 * @param c Code point to test.
 * @param which UProperty selector constant, identifies which property to check.
 *        Must be UCHAR_BINARY_START<=which<UCHAR_BINARY_LIMIT
 *        or UCHAR_INT_START<=which<UCHAR_INT_LIMIT
 *        or UCHAR_MASK_START<=which<UCHAR_MASK_LIMIT.
 * @return Numeric value that is directly the property value or,
 *         for enumerated properties, corresponds to the numeric value of the enumerated