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

/**
 * Return the script code for a given name, or
 * UScript.INVALID_CODE if not found.
 */
private static int scriptNameToCode(String name) {
    try{
        int[] codes = UScript.getCode(name);
        return codes != null ? codes[0] : UScript.INVALID_CODE;
    }catch( MissingResourceException e){
        ///CLOVER:OFF
        return UScript.INVALID_CODE;
        ///CLOVER:ON
    }
}
 

/**
 * Set a new region of text to be examined by this iterator
 * 
 * @param text text buffer to examine
 * @param start offset into buffer
 * @param length maximum length to examine
 */
void setText(char text[], int start, int length) {
  this.text = text;
  this.start = start;
  this.index = start;
  this.limit = start + length;
  this.scriptStart = start;
  this.scriptLimit = start;
  this.scriptCode = UScript.INVALID_CODE;
}
 

/**
 * Set a new region of text to be examined by this iterator.
 *
 * @param text   text buffer to examine
 * @param start  offset into buffer
 * @param length maximum length to examine
 */
void setText(char[] text, int start, int length) {
    this.text = text;
    this.start = start;
    this.index = start;
    this.limit = start + length;
    this.scriptStart = start;
    this.scriptLimit = start;
    this.scriptCode = UScript.INVALID_CODE;
}
 

/**
 * Registers standard transliterators with the system.  Called by
 * Transliterator during initialization.  Scan all current targets
 * and register those that are scripts T as Any-T/V.
 */
static void register() {

    HashMap<String, Set<String>> seen = new HashMap<String, Set<String>>(); // old code used set, but was dependent on order

    for (Enumeration<String> s = Transliterator.getAvailableSources(); s.hasMoreElements(); ) {
        String source = s.nextElement();

        // Ignore the "Any" source
        if (source.equalsIgnoreCase(ANY)) continue;

        for (Enumeration<String> t = Transliterator.getAvailableTargets(source);
             t.hasMoreElements(); ) {
            String target = t.nextElement();

            // Get the script code for the target.  If not a script, ignore.
            int targetScript = scriptNameToCode(target);
            if (targetScript == UScript.INVALID_CODE) {
                continue;
            }

            Set<String> seenVariants = seen.get(target);
            if (seenVariants == null) {
                seen.put(target, seenVariants = new HashSet<String>());
            }

            for (Enumeration<String> v = Transliterator.getAvailableVariants(source, target);
                 v.hasMoreElements(); ) {
                String variant = v.nextElement();

                // Only process each target/variant pair once
                if (seenVariants.contains(variant)) {
                    continue;
                }
                seenVariants.add(variant);

                String id;
                id = TransliteratorIDParser.STVtoID(ANY, target, variant);
                AnyTransliterator trans = new AnyTransliterator(id, target, variant,
                                                                targetScript);
                Transliterator.registerInstance(trans);
                Transliterator.registerSpecialInverse(target, NULL_ID, false);
            }
        }
    }
}
 

/**
 * 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;
}