com.ibm.icu.lang.UCharacterCategory Java Examples

/**
 * Computes the set of numerics for a string, according to UTS 39 section 5.3.
 */
private void getNumerics(String input, UnicodeSet result) {
    result.clear();

    for (int utf16Offset = 0; utf16Offset < input.length();) {
        int codePoint = Character.codePointAt(input, utf16Offset);
        utf16Offset += Character.charCount(codePoint);

        // Store a representative character for each kind of decimal digit
        if (UCharacter.getType(codePoint) == UCharacterCategory.DECIMAL_DIGIT_NUMBER) {
            // Store the zero character as a representative for comparison.
            // Unicode guarantees it is codePoint - value
            result.add(codePoint - UCharacter.getNumericValue(codePoint));
        }
    }
}
 

/**
* Gets the character extended type
* @param ch character to be tested
* @return extended type it is associated with
*/
private static int getType(int ch)
{
    if (UCharacterUtility.isNonCharacter(ch)) {
        // not a character we return a invalid category count
        return NON_CHARACTER_;
    }
    int result = UCharacter.getType(ch);
    if (result == UCharacterCategory.SURROGATE) {
        if (ch <= UTF16.LEAD_SURROGATE_MAX_VALUE) {
            result = LEAD_SURROGATE_;
        }
        else {
            result = TRAIL_SURROGATE_;
        }
    }
    return result;
}
 

/**
* Gets the character extended type
* @param ch character to be tested
* @return extended type it is associated with
*/
private static int getType(int ch)
{
    if (UCharacterUtility.isNonCharacter(ch)) {
        // not a character we return a invalid category count
        return NON_CHARACTER_;
    }
    int result = UCharacter.getType(ch);
    if (result == UCharacterCategory.SURROGATE) {
        if (ch <= UTF16.LEAD_SURROGATE_MAX_VALUE) {
            result = LEAD_SURROGATE_;
        }
        else {
            result = TRAIL_SURROGATE_;
        }
    }
    return result;
}
 

private static boolean isLNS(int c) {
    // Letter, number, symbol,
    // or a private use code point because those are typically used as letters or numbers.
    // Consider modifier letters only if they are cased.
    int gc = UCharacterProperty.INSTANCE.getType(c);
    return ((1 << gc) & LNS) != 0 ||
            (gc == UCharacterCategory.MODIFIER_LETTER &&
                UCaseProps.INSTANCE.getType(c) != UCaseProps.NONE);
}
 

@Override
int getMaxValue(int which) {
    return UCharacterCategory.CHAR_CATEGORY_COUNT-1;
}
 

@Override
int getMaxValue(int which) {
    return UCharacterCategory.CHAR_CATEGORY_COUNT-1;
}
 

/**
 * Unicode category "Zs" (space separator)
 * 
 * @param c
 *            the character
 * @return {@code true} if the character is space separator
 */
public static boolean isSpaceSeparator(int c) {
    return UCharacter.getType(c) == UCharacterCategory.SPACE_SEPARATOR;
}