Examples with Metaphone org.apache.commons.codec.language.Metaphone used on opensource projects

Example 1 with Metaphone

use of org.apache.commons.codec.language.Metaphone in project lucene-solr by apache.

the class TestPhoneticFilter method testRandomStrings.

/** blast some random strings through the analyzer */
public void testRandomStrings() throws IOException {
    Encoder[] encoders = new Encoder[] { new Metaphone(), new DoubleMetaphone(), new Soundex(), new RefinedSoundex(), new Caverphone2() };
    for (final Encoder e : encoders) {
        Analyzer a = new Analyzer() {

            @Override
            protected TokenStreamComponents createComponents(String fieldName) {
                Tokenizer tokenizer = new MockTokenizer(MockTokenizer.WHITESPACE, false);
                return new TokenStreamComponents(tokenizer, new PhoneticFilter(tokenizer, e, false));
            }
        };
        checkRandomData(random(), a, 1000 * RANDOM_MULTIPLIER);
        a.close();
        Analyzer b = new Analyzer() {

            @Override
            protected TokenStreamComponents createComponents(String fieldName) {
                Tokenizer tokenizer = new MockTokenizer(MockTokenizer.WHITESPACE, false);
                return new TokenStreamComponents(tokenizer, new PhoneticFilter(tokenizer, e, false));
            }
        };
        checkRandomData(random(), b, 1000 * RANDOM_MULTIPLIER);
        b.close();
    }
}

Example 2 with Metaphone

use of org.apache.commons.codec.language.Metaphone in project lucene-solr by apache.

the class TestPhoneticFilterFactory method testFactoryDefaults.

/**
   * Case: default
   */
public void testFactoryDefaults() throws IOException {
    Map<String, String> args = new HashMap<>();
    args.put(PhoneticFilterFactory.ENCODER, "Metaphone");
    PhoneticFilterFactory factory = new PhoneticFilterFactory(args);
    factory.inform(new ClasspathResourceLoader(factory.getClass()));
    assertTrue(factory.getEncoder() instanceof Metaphone);
    // default
    assertTrue(factory.inject);
}

Example 3 with Metaphone

use of org.apache.commons.codec.language.Metaphone in project Saiy-PS by brandall76.

the class MetaphoneHelper method executeGeneric.

/**
 * Method to iterate through the given input data and attempt to match the given String data
 * using the {@link Metaphone} within ranges applied by the associated thresholds constants.
 *
 * @return an {@link AlgorithmicContainer} or null if thresholds aren't satisfied
 */
public AlgorithmicContainer executeGeneric() {
    long then = System.nanoTime();
    final double jwdLowerThreshold = SPH.getJaroWinklerLower(mContext);
    final ArrayList<AlgorithmicContainer> toKeep = new ArrayList<>();
    final Metaphone metaphone = new Metaphone();
    final JaroWinklerDistance jwd = new JaroWinklerDistance();
    String generic;
    String genericLower;
    AlgorithmicContainer container = null;
    double score;
    boolean matches;
    int size = genericData.size();
    outer: for (int i = 0; i < size; i++) {
        generic = (String) genericData.get(i);
        genericLower = generic.toLowerCase(loc).trim();
        for (String vd : inputData) {
            vd = vd.toLowerCase(loc).trim();
            matches = metaphone.isMetaphoneEqual(genericLower, vd);
            if (matches && Algorithm.checkLength(genericLower, vd)) {
                score = jwd.apply(genericLower, vd);
                if (score > jwdLowerThreshold) {
                    container = new AlgorithmicContainer();
                    container.setInput(vd);
                    container.setGenericMatch(generic);
                    container.setScore(score);
                    container.setAlgorithm(Algorithm.METAPHONE);
                    container.setParentPosition(i);
                    container.setExactMatch(true);
                    toKeep.add(container);
                    break outer;
                } else {
                    if (DEBUG) {
                        MyLog.i(CLS_NAME, "Matches: double check JW: rejected");
                    }
                }
            }
        }
    }
    if (UtilsList.notNaked(toKeep)) {
        if (DEBUG) {
            MyLog.i(CLS_NAME, "Have a match");
        }
        container = toKeep.get(0);
    } else {
        if (DEBUG) {
            MyLog.i(CLS_NAME, "no matches");
        }
    }
    if (DEBUG) {
        MyLog.getElapsed(CLS_NAME, then);
    }
    return container;
}

Example 4 with Metaphone

use of org.apache.commons.codec.language.Metaphone in project Saiy-PS by brandall76.

the class MetaphoneHelper method executeCustomCommand.

/**
 * Method to iterate through the voice data and attempt to match the user's custom commands
 * using the {@link Metaphone} within ranges applied by the associated thresholds constants.
 *
 * @return the highest scoring {@link CustomCommand} or null if thresholds aren't satisfied
 */
public CustomCommand executeCustomCommand() {
    long then = System.nanoTime();
    final double jwdLowerThreshold = SPH.getJaroWinklerLower(mContext);
    CustomCommand customCommand = null;
    final ArrayList<CustomCommandContainer> toKeep = new ArrayList<>();
    final Metaphone metaphone = new Metaphone();
    final JaroWinklerDistance jwd = new JaroWinklerDistance();
    String phrase;
    CustomCommandContainer container;
    double score;
    boolean matches;
    int size = genericData.size();
    outer: for (int i = 0; i < size; i++) {
        container = (CustomCommandContainer) genericData.get(i);
        phrase = container.getKeyphrase().toLowerCase(loc).trim();
        for (String vd : inputData) {
            vd = vd.toLowerCase(loc).trim();
            matches = metaphone.isMetaphoneEqual(phrase, vd);
            if (matches && Algorithm.checkLength(phrase, vd)) {
                score = jwd.apply(phrase, vd);
                if (score > jwdLowerThreshold) {
                    container.setScore(score);
                    container.setUtterance(vd);
                    container.setExactMatch(true);
                    toKeep.add(SerializationUtils.clone(container));
                    break outer;
                } else {
                    if (DEBUG) {
                        MyLog.i(CLS_NAME, "Matches: double check JW: rejected");
                    }
                }
            }
        }
    }
    if (UtilsList.notNaked(toKeep)) {
        if (DEBUG) {
            MyLog.i(CLS_NAME, "Have a match");
        }
        final CustomCommandContainer ccc = toKeep.get(0);
        final Gson gson = new GsonBuilder().disableHtmlEscaping().create();
        customCommand = gson.fromJson(ccc.getSerialised(), CustomCommand.class);
        customCommand.setExactMatch(ccc.isExactMatch());
        customCommand.setUtterance(ccc.getUtterance());
        customCommand.setAlgorithm(Algorithm.METAPHONE);
    } else {
        if (DEBUG) {
            MyLog.i(CLS_NAME, "no custom phrases matched");
        }
    }
    if (DEBUG) {
        MyLog.getElapsed(CLS_NAME, then);
    }
    return customCommand;
}

Example 5 with Metaphone

use of org.apache.commons.codec.language.Metaphone in project lucene-solr by apache.

the class TestPhoneticFilter method testAlgorithms.

public void testAlgorithms() throws Exception {
    assertAlgorithm(new Metaphone(), true, "aaa bbb ccc easgasg", new String[] { "A", "aaa", "B", "bbb", "KKK", "ccc", "ESKS", "easgasg" });
    assertAlgorithm(new Metaphone(), false, "aaa bbb ccc easgasg", new String[] { "A", "B", "KKK", "ESKS" });
    assertAlgorithm(new DoubleMetaphone(), true, "aaa bbb ccc easgasg", new String[] { "A", "aaa", "PP", "bbb", "KK", "ccc", "ASKS", "easgasg" });
    assertAlgorithm(new DoubleMetaphone(), false, "aaa bbb ccc easgasg", new String[] { "A", "PP", "KK", "ASKS" });
    assertAlgorithm(new Soundex(), true, "aaa bbb ccc easgasg", new String[] { "A000", "aaa", "B000", "bbb", "C000", "ccc", "E220", "easgasg" });
    assertAlgorithm(new Soundex(), false, "aaa bbb ccc easgasg", new String[] { "A000", "B000", "C000", "E220" });
    assertAlgorithm(new RefinedSoundex(), true, "aaa bbb ccc easgasg", new String[] { "A0", "aaa", "B1", "bbb", "C3", "ccc", "E034034", "easgasg" });
    assertAlgorithm(new RefinedSoundex(), false, "aaa bbb ccc easgasg", new String[] { "A0", "B1", "C3", "E034034" });
    assertAlgorithm(new Caverphone2(), true, "Darda Karleen Datha Carlene", new String[] { "TTA1111111", "Darda", "KLN1111111", "Karleen", "TTA1111111", "Datha", "KLN1111111", "Carlene" });
    assertAlgorithm(new Caverphone2(), false, "Darda Karleen Datha Carlene", new String[] { "TTA1111111", "KLN1111111", "TTA1111111", "KLN1111111" });
    assertAlgorithm(new Nysiis(), true, "aaa bbb ccc easgasg", new String[] { "A", "aaa", "B", "bbb", "C", "ccc", "EASGAS", "easgasg" });
    assertAlgorithm(new Nysiis(), false, "aaa bbb ccc easgasg", new String[] { "A", "B", "C", "EASGAS" });
}