Java Code Examples for com.ibm.icu.text.BreakIterator#getWordInstance()

public static BreakIterator getTitleBreakIterator(
        Locale locale, int options, BreakIterator iter) {
    options &= TITLECASE_ITERATOR_MASK;
    if (options != 0 && iter != null) {
        throw new IllegalArgumentException(
                "titlecasing iterator option together with an explicit iterator");
    }
    if (iter == null) {
        switch (options) {
        case 0:
            iter = BreakIterator.getWordInstance(locale);
            break;
        case TITLECASE_WHOLE_STRING:
            iter = new WholeStringBreakIterator();
            break;
        case TITLECASE_SENTENCES:
            iter = BreakIterator.getSentenceInstance(locale);
            break;
        default:
            throw new IllegalArgumentException("unknown titlecasing iterator option");
        }
    }
    return iter;
}
 

public static BreakIterator getTitleBreakIterator(
        ULocale locale, int options, BreakIterator iter) {
    options &= TITLECASE_ITERATOR_MASK;
    if (options != 0 && iter != null) {
        throw new IllegalArgumentException(
                "titlecasing iterator option together with an explicit iterator");
    }
    if (iter == null) {
        switch (options) {
        case 0:
            iter = BreakIterator.getWordInstance(locale);
            break;
        case TITLECASE_WHOLE_STRING:
            iter = new WholeStringBreakIterator();
            break;
        case TITLECASE_SENTENCES:
            iter = BreakIterator.getSentenceInstance(locale);
            break;
        default:
            throw new IllegalArgumentException("unknown titlecasing iterator option");
        }
    }
    return iter;
}
 

private BreakIterator createBreakIterator() {
    ULocale locale = ULocale.forLanguageTag(this.locale);
    if ("line".equals(granularity)) {
        // "strictness" cannot be set through unicode extensions (u-lb-strict), handle here:
        locale = locale.setKeywordValue("lb", strictness);
    }
    BreakIterator breakIterator;
    switch (granularity) {
    case "grapheme":
        breakIterator = BreakIterator.getCharacterInstance(locale);
        break;
    case "word":
        breakIterator = BreakIterator.getWordInstance(locale);
        break;
    case "sentence":
        breakIterator = BreakIterator.getSentenceInstance(locale);
        break;
    case "line":
        breakIterator = BreakIterator.getLineInstance(locale);
        break;
    default:
        throw new AssertionError();
    }
    return breakIterator;
}
 

/**
 * This function can be overridden by subclasses to use different heuristics.
 * <b>It MUST return a 'safe' value,
 * one whose modification will not affect this object.</b>
 *
 * @param type
 * @draft ICU 3.6
 * @provisional This API might change or be removed in a future release.
 */
protected BreakIterator guessBreakIterator(int type) {
    BreakIterator bitr = null;
    ULocale brkLocale = getAvailableLocale(TYPE_BREAKITERATOR);
    if (brkLocale == null) {
        brkLocale = ULocale.ROOT;
    }
    switch (type) {
    case BI_CHARACTER:
        bitr = BreakIterator.getCharacterInstance(brkLocale);
        break;
    case BI_TITLE:
        bitr = BreakIterator.getTitleInstance(brkLocale);
        break;
    case BI_WORD:
        bitr = BreakIterator.getWordInstance(brkLocale);
        break;
    case BI_LINE:
        bitr = BreakIterator.getLineInstance(brkLocale);
        break;
    case BI_SENTENCE:
        bitr = BreakIterator.getSentenceInstance(brkLocale);
        break;
    default:
        throw new IllegalArgumentException("Unknown break iterator type");
    }
    return bitr;
}
 

public ICUWordRecognizer( String text, Locale locale )
{
	if(locale!=null)
	{
		wordBreaker = BreakIterator.getWordInstance(locale);
	}
	else
	{
		wordBreaker = BreakIterator.getWordInstance(Locale.getDefault( ));
	}
	this.text = text;
	wordBreaker.setText( text);

}
 

/**
 * Creates a new break iterator.
 */
public JavaBreakIterator() {
	fIterator= BreakIterator.getWordInstance();
	fIndex= fIterator.current();
}
 

/**
 * Creates a new spell check iterator.
 *
 * @param document the document containing the specified partition
 * @param region the region to spell check
 * @param locale the locale to use for spell checking
 */
public SpellCheckIterator(IDocument document, IRegion region, Locale locale) {
	this(document, region, locale, BreakIterator.getWordInstance(locale));
}