Examples with BreakIterator java.text.BreakIterator used on opensource projects

Example 31 with BreakIterator

use of java.text.BreakIterator in project jdk8u_jdk by JetBrains.

the class TextLayoutStrategy method sync.

/**
     * Synchronize the strategy with its FlowView.  Allows the strategy
     * to update its state to account for changes in that portion of the
     * model represented by the FlowView.  Also allows the strategy
     * to update the FlowView in response to these changes.
     */
void sync(FlowView fv) {
    View lv = getLogicalView(fv);
    text.setView(lv);
    Container container = fv.getContainer();
    FontRenderContext frc = sun.swing.SwingUtilities2.getFontRenderContext(container);
    frcTx = frc.getTransform();
    BreakIterator iter;
    Container c = fv.getContainer();
    if (c != null) {
        iter = BreakIterator.getLineInstance(c.getLocale());
    } else {
        iter = BreakIterator.getLineInstance();
    }
    Object shaper = null;
    if (c instanceof JComponent) {
        shaper = ((JComponent) c).getClientProperty(TextAttribute.NUMERIC_SHAPING);
    }
    text.setShaper(shaper);
    measurer = new LineBreakMeasurer(text, iter, frc);
    // If the children of the FlowView's logical view are GlyphViews, they
    // need to have their painters updated.
    int n = lv.getViewCount();
    for (int i = 0; i < n; i++) {
        View child = lv.getView(i);
        if (child instanceof GlyphView) {
            int p0 = child.getStartOffset();
            int p1 = child.getEndOffset();
            measurer.setPosition(text.toIteratorIndex(p0));
            TextLayout layout = measurer.nextLayout(Float.MAX_VALUE, text.toIteratorIndex(p1), false);
            ((GlyphView) child).setGlyphPainter(new GlyphPainter2(layout));
        }
    }
    // Reset measurer.
    measurer.setPosition(text.getBeginIndex());
}

Example 32 with BreakIterator

use of java.text.BreakIterator in project cogcomp-nlp by CogComp.

the class ThaiTokenizer method getTextAnnotation.

public TextAnnotation getTextAnnotation(String text) {
    List<IntPair> offsets = new ArrayList<>();
    List<String> surfaces = new ArrayList<>();
    List<Integer> sen_ends = new ArrayList<>();
    BreakIterator boundary = BreakIterator.getWordInstance(new Locale("th", "TH", "TH"));
    boundary.setText(text);
    int start = boundary.first();
    for (int end = boundary.next(); end != BreakIterator.DONE; start = end, end = boundary.next()) {
        //            System.out.println(start+" "+end+" "+text.length());
        String sur = text.substring(start, end);
        if (sur.trim().isEmpty()) {
            //                    sen_ends.add(surfaces.size());
            continue;
        }
        surfaces.add(sur);
        offsets.add(new IntPair(start, end));
    }
    if (surfaces.size() > 0 && (sen_ends.size() == 0 || sen_ends.get(sen_ends.size() - 1) != surfaces.size()))
        sen_ends.add(surfaces.size());
    IntPair[] offs = new IntPair[offsets.size()];
    offs = offsets.toArray(offs);
    String[] surfs = new String[surfaces.size()];
    surfs = surfaces.toArray(surfs);
    int[] ends = new int[sen_ends.size()];
    for (int i = 0; i < sen_ends.size(); i++) ends[i] = sen_ends.get(i);
    //        System.out.println(text);
    //        System.out.println(offsets);
    //        System.out.println(sen_ends);
    TextAnnotation ta = new TextAnnotation("", "", text, offs, surfs, ends);
    return ta;
}

Example 33 with BreakIterator

use of java.text.BreakIterator in project chefly_android by chef-ly.

the class GetCookingActivity method onStart.

// TODO - add onPause and onResume where the speech and recognizer are stopped
@Override
protected void onStart() {
    super.onStart();
    Intent i = getIntent();
    //  Get ingredient list for popup
    ArrayList<String> ingredients = i.getStringArrayListExtra("ingredients");
    if (ingredients != null) {
        ingredientsPopup = DialogPopUp.newInstance(getResources().getString(R.string.ingredients), ingredients);
        ingredientsPopup.showTitle(false);
    } else {
        ingredientsPopup = DialogPopUp.newInstance(getResources().getString(R.string.ingredients), new ArrayList<String>());
    }
    // Split up the directions to more "digestible" (wink wink) pieces
    String[] rawDirections = i.getStringArrayExtra("directions");
    directions = new ArrayList<>();
    for (int x = 0; x < rawDirections.length; x++) {
        BreakIterator iterator = BreakIterator.getSentenceInstance(Locale.US);
        String source = rawDirections[x];
        iterator.setText(source);
        int start = iterator.first();
        for (int end = iterator.next(); end != BreakIterator.DONE; start = end, end = iterator.next()) {
            directions.add(source.substring(start, end));
        }
    }
    numberSteps = directions.size();
    if (numberSteps > 0) {
        directionsPopup = DialogPopUp.newInstance(getResources().getString(R.string.directions), directions);
        directionsPopup.showTitle(false);
        hasDirections = true;
        updateStepText();
        ArrayList<GetCookingFragment> frags = new ArrayList<>();
        for (int j = 0; j < numberSteps; j++) {
            frags.add(GetCookingFragment.newInstance(String.valueOf(j), directions.get(j)));
        }
        pager.setAdapter(new CookingPagerAdapter(getSupportFragmentManager(), frags));
        Log.d(TAG, "Step ----> " + step);
        pager.setCurrentItem(step);
    }
    // register with EventBus to get events from VoiceRec
    EventBus.getDefault().register(this);
}

Example 34 with BreakIterator

use of java.text.BreakIterator in project hive by apache.

the class GenericUDFSentences method evaluate.

@Override
public Object evaluate(DeferredObject[] arguments) throws HiveException {
    assert (arguments.length >= 1 && arguments.length <= 3);
    if (arguments[0].get() == null) {
        return null;
    }
    // if there is more than 1 argument specified, a different natural language
    // locale is being specified
    Locale locale = null;
    if (arguments.length > 1 && arguments[1].get() != null) {
        Text language = (Text) converters[1].convert(arguments[1].get());
        Text country = null;
        if (arguments.length > 2 && arguments[2].get() != null) {
            country = (Text) converters[2].convert(arguments[2].get());
        }
        if (country != null) {
            locale = new Locale(language.toString().toLowerCase(), country.toString().toUpperCase());
        } else {
            locale = new Locale(language.toString().toLowerCase());
        }
    } else {
        locale = Locale.getDefault();
    }
    // get the input and prepare the output
    Text chunk = (Text) converters[0].convert(arguments[0].get());
    String text = chunk.toString();
    ArrayList<ArrayList<Text>> result = new ArrayList<ArrayList<Text>>();
    // Parse out sentences using Java's text-handling API
    BreakIterator bi = BreakIterator.getSentenceInstance(locale);
    bi.setText(text);
    int idx = 0;
    while (bi.next() != BreakIterator.DONE) {
        String sentence = text.substring(idx, bi.current());
        idx = bi.current();
        result.add(new ArrayList<Text>());
        // Parse out words in the sentence
        BreakIterator wi = BreakIterator.getWordInstance(locale);
        wi.setText(sentence);
        int widx = 0;
        ArrayList<Text> sent_array = result.get(result.size() - 1);
        while (wi.next() != BreakIterator.DONE) {
            String word = sentence.substring(widx, wi.current());
            widx = wi.current();
            if (Character.isLetterOrDigit(word.charAt(0))) {
                sent_array.add(new Text(word));
            }
        }
    }
    return result;
}

Example 35 with BreakIterator

use of java.text.BreakIterator in project elasticsearch by elastic.

the class CustomSeparatorBreakIteratorTests method testSliceMiddle.

public void testSliceMiddle() throws Exception {
    BreakIterator expected = BreakIterator.getSentenceInstance(Locale.ROOT);
    BreakIterator actual = new CustomSeparatorBreakIterator(randomSeparator());
    assertSameBreaks("000a000", 3, 1, expected, actual);
    assertSameBreaks("000ab000", 3, 2, expected, actual);
    assertSameBreaks("000abc000", 3, 3, expected, actual);
    assertSameBreaks("000000", 3, 0, expected, actual);
}