Java Code Examples for com.ibm.icu.lang.UScript#COMMON

/**
 * Iterates to the next script run, returning true if one exists.
 * 
 * @return true if there is another script run, false otherwise.
 */
boolean next() {
  if (scriptLimit >= limit)
    return false;

  scriptCode = UScript.COMMON;
  scriptStart = scriptLimit;

  while (index < limit) {
    final int ch = UTF16.charAt(text, start, limit, index - start);
    final int sc = getScript(ch);

    /*
     * From UTR #24: Implementations that determine the boundaries between
     * characters of given scripts should never break between a non-spacing
     * mark and its base character. Thus for boundary determinations and
     * similar sorts of processing, a non-spacing mark — whatever its script
     * value — should inherit the script value of its base character.
     */
    if (isSameScript(scriptCode, sc)
        || UCharacter.getType(ch) == ECharacterCategory.NON_SPACING_MARK) {
      index += UTF16.getCharCount(ch);

      /*
       * Inherited or Common becomes the script code of the surrounding text.
       */
      if (scriptCode <= UScript.INHERITED && sc > UScript.INHERITED) {
        scriptCode = sc;
      }

    } else {
      break;
    }
  }

  scriptLimit = index;
  return true;
}
 

/**
 * Iterates to the next script run, returning true if one exists.
 *
 * @return true if there is another script run, false otherwise.
 */
boolean next() {
    if (scriptLimit >= limit) {
        return false;
    }
    scriptCode = UScript.COMMON;
    scriptStart = scriptLimit;
    while (index < limit) {
        final int ch = UTF16.charAt(text, start, limit, index - start);
        final int sc = getScript(ch);
        /*
         * From UTR #24: Implementations that determine the boundaries between
         * characters of given scripts should never break between a non-spacing
         * mark and its base character. Thus for boundary determinations and
         * similar sorts of processing, a non-spacing mark — whatever its script
         * value — should inherit the script value of its base character.
         */
        if (isSameScript(scriptCode, sc)
                || UCharacter.getType(ch) == ECharacterCategory.NON_SPACING_MARK) {
            index += UTF16.getCharCount(ch);
            /*
             * Inherited or Common becomes the script code of the surrounding text.
             */
            if (scriptCode <= UScript.INHERITED && sc > UScript.INHERITED) {
                scriptCode = sc;
            }
        } else {
            break;
        }
    }
    scriptLimit = index;
    return true;
}
 

@Override
public boolean has(int codePoint, int value) {
    // See https://ssl.icu-project.org/trac/ticket/13462
    switch (codePoint) {
    case 0x3000:
    case 0x3004:
    case 0x3012:
    case 0x3020:
    case 0x3036:
        return value == UScript.COMMON;
    }
    return super.has(codePoint, value);
}
 

@Override
public boolean has(int codePoint, int value) {
    // See https://ssl.icu-project.org/trac/ticket/13462
    switch (codePoint) {
    case 0x3000:
    case 0x3004:
    case 0x3012:
    case 0x3020:
    case 0x3036:
        return value == UScript.COMMON;
    }
    return UScript.hasScript(codePoint, value);
}
 

/**
 * Returns TRUE if there are any more runs.  TRUE is always
 * returned at least once.  Upon return, the caller should
 * examine scriptCode, start, and limit.
 */
public boolean next() {
    int ch;
    int s;

    scriptCode = UScript.INVALID_CODE; // don't know script yet
    start = limit;

    // Are we done?
    if (start == textLimit) {
        return false;
    }

    // Move start back to include adjacent COMMON or INHERITED
    // characters
    while (start > textStart) {
        ch = text.char32At(start - 1); // look back
        s = UScript.getScript(ch);
        if (s == UScript.COMMON || s == UScript.INHERITED) {
            --start;
        } else {
            break;
        }
    }

    // Move limit ahead to include COMMON, INHERITED, and characters
    // of the current script.
    while (limit < textLimit) {
        ch = text.char32At(limit); // look ahead
        s = UScript.getScript(ch);
        if (s != UScript.COMMON && s != UScript.INHERITED) {
            if (scriptCode == UScript.INVALID_CODE) {
                scriptCode = s;
            } else if (s != scriptCode) {
                break;
            }
        }
        ++limit;
    }

    // Return TRUE even if the entire text is COMMON / INHERITED, in
    // which case scriptCode will be UScript.INVALID_CODE.
    return true;
}
 

@Override
public void clear() {
  code = UScript.COMMON;
}
 

public boolean includesCharacter(int codePoint)
{
	if (includedScripts == null && excludedScripts == null)
	{
		return true;
	}
	
	int codeScript = UScript.getScript(codePoint);
	if (codeScript == UScript.UNKNOWN)
	{
		//include by default
		return true;
	}
	
	if (codeScript == UScript.COMMON)
	{
		//COMMON is included unless explicitly excluded
		return !excludedCommon;
	}
	
	if (codeScript == UScript.INHERITED)
	{
		//INHERITED is included unless explicitly excluded
		return !excludedInherited;
	}
	
	if (includedScripts != null && includedScripts.contains(codeScript))
	{
		//the codepoint script is explicitly included
		return true;
	}
	
	if (excludedScripts != null && excludedScripts.contains(codeScript))
	{
		//the codepoint script is explicitly excluded
		return false;
	}
	
	if (includedScripts == null)
	{
		//not excluded
		return true;
	}
	
	for (Integer script : includedScripts)
	{
		if (UScript.hasScript(codePoint, script))
		{
			//included as a secondary/extension script
			return true;
		}
	}
	
	//not included
	return false;
}
 

@Override
public void clear() {
    code = UScript.COMMON;
}