Examples with LookupResult org.apache.lucene.search.suggest.Lookup.LookupResult used on opensource projects

Example 31 with LookupResult

use of org.apache.lucene.search.suggest.Lookup.LookupResult in project lucene-solr by apache.

the class FuzzySuggesterTest method testRandomEdits.

public void testRandomEdits() throws IOException {
    List<Input> keys = new ArrayList<>();
    int numTerms = atLeast(100);
    for (int i = 0; i < numTerms; i++) {
        keys.add(new Input("boo" + TestUtil.randomSimpleString(random()), 1 + random().nextInt(100)));
    }
    keys.add(new Input("foo bar boo far", 12));
    MockAnalyzer analyzer = new MockAnalyzer(random(), MockTokenizer.KEYWORD, false);
    Directory tempDir = getDirectory();
    FuzzySuggester suggester = new FuzzySuggester(tempDir, "fuzzy", analyzer, analyzer, FuzzySuggester.EXACT_FIRST | FuzzySuggester.PRESERVE_SEP, 256, -1, true, FuzzySuggester.DEFAULT_MAX_EDITS, FuzzySuggester.DEFAULT_TRANSPOSITIONS, 0, FuzzySuggester.DEFAULT_MIN_FUZZY_LENGTH, FuzzySuggester.DEFAULT_UNICODE_AWARE);
    suggester.build(new InputArrayIterator(keys));
    int numIters = atLeast(10);
    for (int i = 0; i < numIters; i++) {
        String addRandomEdit = addRandomEdit("foo bar boo", FuzzySuggester.DEFAULT_NON_FUZZY_PREFIX);
        List<LookupResult> results = suggester.lookup(TestUtil.stringToCharSequence(addRandomEdit, random()), false, 2);
        assertEquals(addRandomEdit, 1, results.size());
        assertEquals("foo bar boo far", results.get(0).key.toString());
        assertEquals(12, results.get(0).value, 0.01F);
    }
    IOUtils.close(analyzer, tempDir);
}

Example 32 with LookupResult

use of org.apache.lucene.search.suggest.Lookup.LookupResult in project lucene-solr by apache.

the class FuzzySuggesterTest method slowFuzzyMatch.

private List<LookupResult> slowFuzzyMatch(int prefixLen, int maxEdits, boolean allowTransposition, List<Input> answers, String frag) {
    final List<LookupResult> results = new ArrayList<>();
    final int fragLen = frag.length();
    for (Input tf : answers) {
        //System.out.println("  check s=" + tf.term.utf8ToString());
        boolean prefixMatches = true;
        for (int i = 0; i < prefixLen; i++) {
            if (i == fragLen) {
                // Prefix still matches:
                break;
            }
            if (i == tf.term.length || tf.term.bytes[i] != (byte) frag.charAt(i)) {
                prefixMatches = false;
                break;
            }
        }
        if (prefixMatches) {
            final int len = tf.term.length;
            if (len >= fragLen - maxEdits) {
                // OK it's possible:
                //System.out.println("    possible");
                int d;
                final String s = tf.term.utf8ToString();
                if (fragLen == prefixLen) {
                    d = 0;
                } else if (false && len < fragLen) {
                    d = getDistance(frag, s, allowTransposition);
                } else {
                    //System.out.println("    try loop");
                    d = maxEdits + 1;
                    //for(int ed=-maxEdits;ed<=maxEdits;ed++) {
                    for (int ed = -maxEdits; ed <= maxEdits; ed++) {
                        if (s.length() < fragLen - ed) {
                            continue;
                        }
                        String check = s.substring(0, fragLen - ed);
                        d = getDistance(frag, check, allowTransposition);
                        //System.out.println("    sub check s=" + check + " d=" + d);
                        if (d <= maxEdits) {
                            break;
                        }
                    }
                }
                if (d <= maxEdits) {
                    results.add(new LookupResult(tf.term.utf8ToString(), tf.v));
                }
            }
        }
        Collections.sort(results, new CompareByCostThenAlpha());
    }
    return results;
}

Example 33 with LookupResult

use of org.apache.lucene.search.suggest.Lookup.LookupResult in project lucene-solr by apache.

the class FuzzySuggesterTest method testNoSeps.

public void testNoSeps() throws Exception {
    Input[] keys = new Input[] { new Input("ab cd", 0), new Input("abcd", 1) };
    int options = 0;
    Analyzer a = new MockAnalyzer(random());
    Directory tempDir = getDirectory();
    FuzzySuggester suggester = new FuzzySuggester(tempDir, "fuzzy", a, a, options, 256, -1, true, 1, true, 1, 3, false);
    suggester.build(new InputArrayIterator(keys));
    // TODO: would be nice if "ab " would allow the test to
    // pass, and more generally if the analyzer can know
    // that the user's current query has ended at a word, 
    // but, analyzers don't produce SEP tokens!
    List<LookupResult> r = suggester.lookup(TestUtil.stringToCharSequence("ab c", random()), false, 2);
    assertEquals(2, r.size());
    // With no PRESERVE_SEPS specified, "ab c" should also
    // complete to "abcd", which has higher weight so should
    // appear first:
    assertEquals("abcd", r.get(0).key.toString());
    IOUtils.close(a, tempDir);
}

Example 34 with LookupResult

use of org.apache.lucene.search.suggest.Lookup.LookupResult in project lucene-solr by apache.

the class FuzzySuggesterTest method testStandard.

/**
   * basic "standardanalyzer" test with stopword removal
   */
public void testStandard() throws Exception {
    Input[] keys = new Input[] { new Input("the ghost of christmas past", 50) };
    Analyzer standard = new MockAnalyzer(random(), MockTokenizer.WHITESPACE, true, MockTokenFilter.ENGLISH_STOPSET);
    Directory tempDir = getDirectory();
    FuzzySuggester suggester = new FuzzySuggester(tempDir, "fuzzy", standard, standard, AnalyzingSuggester.EXACT_FIRST | AnalyzingSuggester.PRESERVE_SEP, 256, -1, false, FuzzySuggester.DEFAULT_MAX_EDITS, FuzzySuggester.DEFAULT_TRANSPOSITIONS, FuzzySuggester.DEFAULT_NON_FUZZY_PREFIX, FuzzySuggester.DEFAULT_MIN_FUZZY_LENGTH, FuzzySuggester.DEFAULT_UNICODE_AWARE);
    suggester.build(new InputArrayIterator(keys));
    List<LookupResult> results = suggester.lookup(TestUtil.stringToCharSequence("the ghost of chris", random()), false, 1);
    assertEquals(1, results.size());
    assertEquals("the ghost of christmas past", results.get(0).key.toString());
    assertEquals(50, results.get(0).value, 0.01F);
    // omit the 'the' since it's a stopword, it's suggested anyway
    results = suggester.lookup(TestUtil.stringToCharSequence("ghost of chris", random()), false, 1);
    assertEquals(1, results.size());
    assertEquals("the ghost of christmas past", results.get(0).key.toString());
    assertEquals(50, results.get(0).value, 0.01F);
    // omit the 'the' and 'of' since they are stopwords, it's suggested anyway
    results = suggester.lookup(TestUtil.stringToCharSequence("ghost chris", random()), false, 1);
    assertEquals(1, results.size());
    assertEquals("the ghost of christmas past", results.get(0).key.toString());
    assertEquals(50, results.get(0).value, 0.01F);
    IOUtils.close(standard, tempDir);
}

Example 35 with LookupResult

use of org.apache.lucene.search.suggest.Lookup.LookupResult in project lucene-solr by apache.

the class FuzzySuggesterTest method testNonLatinRandomEdits.

public void testNonLatinRandomEdits() throws IOException {
    List<Input> keys = new ArrayList<>();
    int numTerms = atLeast(100);
    for (int i = 0; i < numTerms; i++) {
        keys.add(new Input("буу" + TestUtil.randomSimpleString(random()), 1 + random().nextInt(100)));
    }
    keys.add(new Input("фуу бар буу фар", 12));
    MockAnalyzer analyzer = new MockAnalyzer(random(), MockTokenizer.KEYWORD, false);
    Directory tempDir = getDirectory();
    FuzzySuggester suggester = new FuzzySuggester(tempDir, "fuzzy", analyzer, analyzer, FuzzySuggester.EXACT_FIRST | FuzzySuggester.PRESERVE_SEP, 256, -1, true, FuzzySuggester.DEFAULT_MAX_EDITS, FuzzySuggester.DEFAULT_TRANSPOSITIONS, 0, FuzzySuggester.DEFAULT_MIN_FUZZY_LENGTH, true);
    suggester.build(new InputArrayIterator(keys));
    int numIters = atLeast(10);
    for (int i = 0; i < numIters; i++) {
        String addRandomEdit = addRandomEdit("фуу бар буу", 0);
        List<LookupResult> results = suggester.lookup(TestUtil.stringToCharSequence(addRandomEdit, random()), false, 2);
        assertEquals(addRandomEdit, 1, results.size());
        assertEquals("фуу бар буу фар", results.get(0).key.toString());
        assertEquals(12, results.get(0).value, 0.01F);
    }
    IOUtils.close(analyzer, tempDir);
}