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

        Same as u_getCombiningClass, returns 8-bit numeric values. @stable ICU 2.2 */
    UCHAR_CANONICAL_COMBINING_CLASS,
    /** Enumerated property Decomposition_Type.
        Returns UDecompositionType values. @stable ICU 2.2 */
    UCHAR_DECOMPOSITION_TYPE,
    /** Enumerated property East_Asian_Width.
        See http://www.unicode.org/reports/tr11/
        Returns UEastAsianWidth values. @stable ICU 2.2 */
    UCHAR_EAST_ASIAN_WIDTH,
    /** Enumerated property General_Category.
        Same as u_charType, returns UCharCategory values. @stable ICU 2.2 */
    UCHAR_GENERAL_CATEGORY,
    /** Enumerated property Joining_Group.
        Returns UJoiningGroup values. @stable ICU 2.2 */
    UCHAR_JOINING_GROUP,
    /** Enumerated property Joining_Type.
        Returns UJoiningType values. @stable ICU 2.2 */
    UCHAR_JOINING_TYPE,
    /** Enumerated property Line_Break.
        Returns ULineBreak values. @stable ICU 2.2 */
    UCHAR_LINE_BREAK,
    /** Enumerated property Numeric_Type.
        Returns UNumericType values. @stable ICU 2.2 */
    UCHAR_NUMERIC_TYPE,
    /** Enumerated property Script.
        Same as uscript_getScript, returns UScriptCode values. @stable ICU 2.2 */
    UCHAR_SCRIPT,
    /** Enumerated property Hangul_Syllable_Type, new in Unicode 4.
        Returns UHangulSyllableType values. @stable ICU 2.6 */
    UCHAR_HANGUL_SYLLABLE_TYPE,
    /** Enumerated property NFD_Quick_Check.
        Returns UNormalizationCheckResult values. @draft ICU 3.0 */
    UCHAR_NFD_QUICK_CHECK,
    /** Enumerated property NFKD_Quick_Check.
        Returns UNormalizationCheckResult values. @draft ICU 3.0 */
    UCHAR_NFKD_QUICK_CHECK,
    /** Enumerated property NFC_Quick_Check.
        Returns UNormalizationCheckResult values. @draft ICU 3.0 */
    UCHAR_NFC_QUICK_CHECK,
    /** Enumerated property NFKC_Quick_Check.
        Returns UNormalizationCheckResult values. @draft ICU 3.0 */
    UCHAR_NFKC_QUICK_CHECK,
    /** Enumerated property Lead_Canonical_Combining_Class.
        ICU-specific property for the ccc of the first code point
        of the decomposition, or lccc(c)=ccc(NFD(c)[0]).
        Useful for checking for canonically ordered text;
        see UNORM_FCD and http://www.unicode.org/notes/tn5/#FCD .
        Returns 8-bit numeric values like UCHAR_CANONICAL_COMBINING_CLASS. @draft ICU 3.0 */
    UCHAR_LEAD_CANONICAL_COMBINING_CLASS,
    /** Enumerated property Trail_Canonical_Combining_Class.
        ICU-specific property for the ccc of the last code point
        of the decomposition, or tccc(c)=ccc(NFD(c)[last]).
        Useful for checking for canonically ordered text;
        see UNORM_FCD and http://www.unicode.org/notes/tn5/#FCD .
        Returns 8-bit numeric values like UCHAR_CANONICAL_COMBINING_CLASS. @draft ICU 3.0 */
    UCHAR_TRAIL_CANONICAL_COMBINING_CLASS,
    /** One more than the last constant for enumerated/integer Unicode properties. @stable ICU 2.2 */
    UCHAR_INT_LIMIT,

    /** Bitmask property General_Category_Mask.
        This is the General_Category property returned as a bit mask.
        When used in u_getIntPropertyValue(c), same as U_MASK(u_charType(c)),
        returns bit masks for UCharCategory values where exactly one bit is set.
        When used with u_getPropertyValueName() and u_getPropertyValueEnum(),
        a multi-bit mask is used for sets of categories like "Letters".
        Mask values should be cast to uint32_t.
        @stable ICU 2.4 */
    UCHAR_GENERAL_CATEGORY_MASK=0x2000,
    /** First constant for bit-mask Unicode properties. @stable ICU 2.4 */
    UCHAR_MASK_START=UCHAR_GENERAL_CATEGORY_MASK,
    /** One more than the last constant for bit-mask Unicode properties. @stable ICU 2.4 */
    UCHAR_MASK_LIMIT,

    /** Double property Numeric_Value.
        Corresponds to u_getNumericValue. @stable ICU 2.4 */
    UCHAR_NUMERIC_VALUE=0x3000,
    /** First constant for double Unicode properties. @stable ICU 2.4 */
    UCHAR_DOUBLE_START=UCHAR_NUMERIC_VALUE,
    /** One more than the last constant for double Unicode properties. @stable ICU 2.4 */
    UCHAR_DOUBLE_LIMIT,

    /** String property Age.
        Corresponds to u_charAge. @stable ICU 2.4 */
    UCHAR_AGE=0x4000,
    /** First constant for string Unicode properties. @stable ICU 2.4 */
    UCHAR_STRING_START=UCHAR_AGE,
    /** String property Bidi_Mirroring_Glyph.
        Corresponds to u_charMirror. @stable ICU 2.4 */
    UCHAR_BIDI_MIRRORING_GLYPH,
    /** String property Case_Folding.
        Corresponds to u_strFoldCase in ustring.h. @stable ICU 2.4 */
    UCHAR_CASE_FOLDING,
    /** String property ISO_Comment.
        Corresponds to u_getISOComment. @stable ICU 2.4 */
    UCHAR_ISO_COMMENT,
    /** String property Lowercase_Mapping.
        Corresponds to u_strToLower in ustring.h. @stable ICU 2.4 */
    UCHAR_LOWERCASE_MAPPING,
    /** String property Name.
        Corresponds to u_charName. @stable ICU 2.4 */
    UCHAR_NAME,
    /** String property Simple_Case_Folding.
        Corresponds to u_foldCase. @stable ICU 2.4 */
    UCHAR_SIMPLE_CASE_FOLDING,
    /** String property Simple_Lowercase_Mapping.
        Corresponds to u_tolower. @stable ICU 2.4 */
    UCHAR_SIMPLE_LOWERCASE_MAPPING,
    /** String property Simple_Titlecase_Mapping.
        Corresponds to u_totitle. @stable ICU 2.4 */
    UCHAR_SIMPLE_TITLECASE_MAPPING,
    /** String property Simple_Uppercase_Mapping.
        Corresponds to u_toupper. @stable ICU 2.4 */
    UCHAR_SIMPLE_UPPERCASE_MAPPING,
    /** String property Titlecase_Mapping.
        Corresponds to u_strToTitle in ustring.h. @stable ICU 2.4 */
    UCHAR_TITLECASE_MAPPING,
    /** String property Unicode_1_Name.
        Corresponds to u_charName. @stable ICU 2.4 */
    UCHAR_UNICODE_1_NAME,
    /** String property Uppercase_Mapping.
        Corresponds to u_strToUpper in ustring.h. @stable ICU 2.4 */
    UCHAR_UPPERCASE_MAPPING,
    /** One more than the last constant for string Unicode properties. @stable ICU 2.4 */
    UCHAR_STRING_LIMIT,

    /** Represents a nonexistent or invalid property or property value. @stable ICU 2.4 */
    UCHAR_INVALID_CODE = -1
} UProperty;

/**
 * Data for enumerated Unicode general category types.
 * See http://www.unicode.org/Public/UNIDATA/UnicodeData.html .
 * @stable ICU 2.0
 */
typedef enum UCharCategory
{
    /** See note !!.  Comments of the form "Cn" are read by genpname. */

    /** Non-category for unassigned and non-character code points. @stable ICU 2.0 */
    U_UNASSIGNED              = 0,
    /** Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNED!) @stable ICU 2.0 */
    U_GENERAL_OTHER_TYPES     = 0,
    /** Lu @stable ICU 2.0 */
    U_UPPERCASE_LETTER        = 1,
    /** Ll @stable ICU 2.0 */
    U_LOWERCASE_LETTER        = 2,
    /** Lt @stable ICU 2.0 */
    U_TITLECASE_LETTER        = 3,
    /** Lm @stable ICU 2.0 */
    U_MODIFIER_LETTER         = 4,
    /** Lo @stable ICU 2.0 */
    U_OTHER_LETTER            = 5,
    /** Mn @stable ICU 2.0 */
    U_NON_SPACING_MARK        = 6,
    /** Me @stable ICU 2.0 */
    U_ENCLOSING_MARK          = 7,
    /** Mc @stable ICU 2.0 */
    U_COMBINING_SPACING_MARK  = 8,
    /** Nd @stable ICU 2.0 */
    U_DECIMAL_DIGIT_NUMBER    = 9,
    /** Nl @stable ICU 2.0 */
    U_LETTER_NUMBER           = 10,
    /** No @stable ICU 2.0 */
    U_OTHER_NUMBER            = 11,
    /** Zs @stable ICU 2.0 */
    U_SPACE_SEPARATOR         = 12,
    /** Zl @stable ICU 2.0 */
    U_LINE_SEPARATOR          = 13,
    /** Zp @stable ICU 2.0 */
    U_PARAGRAPH_SEPARATOR     = 14,
    /** Cc @stable ICU 2.0 */
    U_CONTROL_CHAR            = 15,
    /** Cf @stable ICU 2.0 */
    U_FORMAT_CHAR             = 16,
    /** Co @stable ICU 2.0 */
    U_PRIVATE_USE_CHAR        = 17,
    /** Cs @stable ICU 2.0 */
    U_SURROGATE               = 18,
    /** Pd @stable ICU 2.0 */
    U_DASH_PUNCTUATION        = 19,
    /** Ps @stable ICU 2.0 */
    U_START_PUNCTUATION       = 20,
    /** Pe @stable ICU 2.0 */
    U_END_PUNCTUATION         = 21,
    /** Pc @stable ICU 2.0 */
    U_CONNECTOR_PUNCTUATION   = 22,
    /** Po @stable ICU 2.0 */
    U_OTHER_PUNCTUATION       = 23,
    /** Sm @stable ICU 2.0 */
    U_MATH_SYMBOL             = 24,
    /** Sc @stable ICU 2.0 */
    U_CURRENCY_SYMBOL         = 25,
    /** Sk @stable ICU 2.0 */
    U_MODIFIER_SYMBOL         = 26,
    /** So @stable ICU 2.0 */
    U_OTHER_SYMBOL            = 27,
    /** Pi @stable ICU 2.0 */
    U_INITIAL_PUNCTUATION     = 28,
    /** Pf @stable ICU 2.0 */
    U_FINAL_PUNCTUATION       = 29,
    /** One higher than the last enum UCharCategory constant. @stable ICU 2.0 */
    U_CHAR_CATEGORY_COUNT
} UCharCategory;

/**
 * U_GC_XX_MASK constants are bit flags corresponding to Unicode
 * general category values.
 * For each category, the nth bit is set if the numeric value of the
 * corresponding UCharCategory constant is n.
 *
 * There are also some U_GC_Y_MASK constants for groups of general categories
 * like L for all letter categories.
 *
 * @see u_charType
 * @see U_GET_GC_MASK
 * @see UCharCategory
 * @stable ICU 2.1
 */
#define U_GC_CN_MASK    U_MASK(U_GENERAL_OTHER_TYPES)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_LU_MASK    U_MASK(U_UPPERCASE_LETTER)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_LL_MASK    U_MASK(U_LOWERCASE_LETTER)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_LT_MASK    U_MASK(U_TITLECASE_LETTER)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_LM_MASK    U_MASK(U_MODIFIER_LETTER)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_LO_MASK    U_MASK(U_OTHER_LETTER)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_MN_MASK    U_MASK(U_NON_SPACING_MARK)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_ME_MASK    U_MASK(U_ENCLOSING_MARK)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_MC_MASK    U_MASK(U_COMBINING_SPACING_MARK)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_ND_MASK    U_MASK(U_DECIMAL_DIGIT_NUMBER)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_NL_MASK    U_MASK(U_LETTER_NUMBER)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_NO_MASK    U_MASK(U_OTHER_NUMBER)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_ZS_MASK    U_MASK(U_SPACE_SEPARATOR)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_ZL_MASK    U_MASK(U_LINE_SEPARATOR)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_ZP_MASK    U_MASK(U_PARAGRAPH_SEPARATOR)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_CC_MASK    U_MASK(U_CONTROL_CHAR)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_CF_MASK    U_MASK(U_FORMAT_CHAR)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_CO_MASK    U_MASK(U_PRIVATE_USE_CHAR)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_CS_MASK    U_MASK(U_SURROGATE)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_PD_MASK    U_MASK(U_DASH_PUNCTUATION)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_PS_MASK    U_MASK(U_START_PUNCTUATION)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_PE_MASK    U_MASK(U_END_PUNCTUATION)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_PC_MASK    U_MASK(U_CONNECTOR_PUNCTUATION)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_PO_MASK    U_MASK(U_OTHER_PUNCTUATION)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_SM_MASK    U_MASK(U_MATH_SYMBOL)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_SC_MASK    U_MASK(U_CURRENCY_SYMBOL)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_SK_MASK    U_MASK(U_MODIFIER_SYMBOL)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_SO_MASK    U_MASK(U_OTHER_SYMBOL)

/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_PI_MASK    U_MASK(U_INITIAL_PUNCTUATION)
/** Mask constant for a UCharCategory. @stable ICU 2.1 */
#define U_GC_PF_MASK    U_MASK(U_FINAL_PUNCTUATION)


/** Mask constant for multiple UCharCategory bits (L Letters). @stable ICU 2.1 */
#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)

/** Mask constant for multiple UCharCategory bits (LC Cased Letters). @stable ICU 2.1 */
#define U_GC_LC_MASK \
            (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)

/** Mask constant for multiple UCharCategory bits (M Marks). @stable ICU 2.1 */
#define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)

/** Mask constant for multiple UCharCategory bits (N Numbers). @stable ICU 2.1 */
#define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)

/** Mask constant for multiple UCharCategory bits (Z Separators). @stable ICU 2.1 */
#define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)

/** Mask constant for multiple UCharCategory bits (C Others). @stable ICU 2.1 */
#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)

/** Mask constant for multiple UCharCategory bits (P Punctuation). @stable ICU 2.1 */
#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)

/** Mask constant for multiple UCharCategory bits (S Symbols). @stable ICU 2.1 */
#define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)

/**
 * This specifies the language directional property of a character set.
 * @stable ICU 2.0
 */
typedef enum UCharDirection {
    /** See note !!.  Comments of the form "EN" are read by genpname. */

    /** L @stable ICU 2.0 */
    U_LEFT_TO_RIGHT               = 0,
    /** R @stable ICU 2.0 */
    U_RIGHT_TO_LEFT               = 1,
    /** EN @stable ICU 2.0 */
    U_EUROPEAN_NUMBER             = 2,
    /** ES @stable ICU 2.0 */
    U_EUROPEAN_NUMBER_SEPARATOR   = 3,
    /** ET @stable ICU 2.0 */
    U_EUROPEAN_NUMBER_TERMINATOR  = 4,
    /** AN @stable ICU 2.0 */
    U_ARABIC_NUMBER               = 5,
    /** CS @stable ICU 2.0 */
    U_COMMON_NUMBER_SEPARATOR     = 6,
    /** B @stable ICU 2.0 */
    U_BLOCK_SEPARATOR             = 7,
    /** S @stable ICU 2.0 */
    U_SEGMENT_SEPARATOR           = 8,
    /** WS @stable ICU 2.0 */
    U_WHITE_SPACE_NEUTRAL         = 9,
    /** ON @stable ICU 2.0 */
    U_OTHER_NEUTRAL               = 10,
    /** LRE @stable ICU 2.0 */
    U_LEFT_TO_RIGHT_EMBEDDING     = 11,
    /** LRO @stable ICU 2.0 */
    U_LEFT_TO_RIGHT_OVERRIDE      = 12,
    /** AL @stable ICU 2.0 */
    U_RIGHT_TO_LEFT_ARABIC        = 13,
    /** RLE @stable ICU 2.0 */
    U_RIGHT_TO_LEFT_EMBEDDING     = 14,
    /** RLO @stable ICU 2.0 */
    U_RIGHT_TO_LEFT_OVERRIDE      = 15,
    /** PDF @stable ICU 2.0 */
    U_POP_DIRECTIONAL_FORMAT      = 16,
    /** NSM @stable ICU 2.0 */
    U_DIR_NON_SPACING_MARK        = 17,
    /** BN @stable ICU 2.0 */
    U_BOUNDARY_NEUTRAL            = 18,
    /** @stable ICU 2.0 */
    U_CHAR_DIRECTION_COUNT
} UCharDirection;

/**
 * Constants for Unicode blocks, see the Unicode Data file Blocks.txt
 * @stable ICU 2.0
 */
enum UBlockCode {

    /** New No_Block value in Unicode 4. @stable ICU 2.6 */