FileDocCategorySizeDatePackage
TestFuzzyQuery.javaAPI DocApache Lucene 1.911058Mon Feb 20 09:19:32 GMT 2006org.apache.lucene.search

TestFuzzyQuery

public class TestFuzzyQuery extends TestCase
Tests {@link FuzzyQuery}.
author
Daniel Naber

Fields Summary
Constructors Summary
Methods Summary
private voidaddDoc(java.lang.String text, org.apache.lucene.index.IndexWriter writer)

    Document doc = new Document();
    doc.add(new Field("field", text, Field.Store.YES, Field.Index.TOKENIZED));
    writer.addDocument(doc);
  
public voidtestFuzziness()

    RAMDirectory directory = new RAMDirectory();
    IndexWriter writer = new IndexWriter(directory, new WhitespaceAnalyzer(), true);
    addDoc("aaaaa", writer);
    addDoc("aaaab", writer);
    addDoc("aaabb", writer);
    addDoc("aabbb", writer);
    addDoc("abbbb", writer);
    addDoc("bbbbb", writer);
    addDoc("ddddd", writer);
    writer.optimize();
    writer.close();
    IndexSearcher searcher = new IndexSearcher(directory);

    FuzzyQuery query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 0);   
    Hits hits = searcher.search(query);
    assertEquals(3, hits.length());
    
    // same with prefix
    query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 1);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 2);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 3);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 4);   
    hits = searcher.search(query);
    assertEquals(2, hits.length());
    query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 5);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 6);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());

    // not similar enough:
    query = new FuzzyQuery(new Term("field", "xxxxx"), FuzzyQuery.defaultMinSimilarity, 0);  	
    hits = searcher.search(query);
    assertEquals(0, hits.length());
    query = new FuzzyQuery(new Term("field", "aaccc"), FuzzyQuery.defaultMinSimilarity, 0);   // edit distance to "aaaaa" = 3
    hits = searcher.search(query);
    assertEquals(0, hits.length());

    // query identical to a word in the index:
    query = new FuzzyQuery(new Term("field", "aaaaa"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaa"));
    // default allows for up to two edits:
    assertEquals(hits.doc(1).get("field"), ("aaaab"));
    assertEquals(hits.doc(2).get("field"), ("aaabb"));

    // query similar to a word in the index:
    query = new FuzzyQuery(new Term("field", "aaaac"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaa"));
    assertEquals(hits.doc(1).get("field"), ("aaaab"));
    assertEquals(hits.doc(2).get("field"), ("aaabb"));
    
    // now with prefix
    query = new FuzzyQuery(new Term("field", "aaaac"), FuzzyQuery.defaultMinSimilarity, 1);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaa"));
    assertEquals(hits.doc(1).get("field"), ("aaaab"));
    assertEquals(hits.doc(2).get("field"), ("aaabb"));
    query = new FuzzyQuery(new Term("field", "aaaac"), FuzzyQuery.defaultMinSimilarity, 2);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaa"));
    assertEquals(hits.doc(1).get("field"), ("aaaab"));
    assertEquals(hits.doc(2).get("field"), ("aaabb"));
    query = new FuzzyQuery(new Term("field", "aaaac"), FuzzyQuery.defaultMinSimilarity, 3);   
    hits = searcher.search(query);
    assertEquals(3, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaa"));
    assertEquals(hits.doc(1).get("field"), ("aaaab"));
    assertEquals(hits.doc(2).get("field"), ("aaabb"));
    query = new FuzzyQuery(new Term("field", "aaaac"), FuzzyQuery.defaultMinSimilarity, 4);   
    hits = searcher.search(query);
    assertEquals(2, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaa"));
    assertEquals(hits.doc(1).get("field"), ("aaaab"));
    query = new FuzzyQuery(new Term("field", "aaaac"), FuzzyQuery.defaultMinSimilarity, 5);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());
    

    query = new FuzzyQuery(new Term("field", "ddddX"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("ddddd"));
    
    // now with prefix
    query = new FuzzyQuery(new Term("field", "ddddX"), FuzzyQuery.defaultMinSimilarity, 1);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("ddddd"));
    query = new FuzzyQuery(new Term("field", "ddddX"), FuzzyQuery.defaultMinSimilarity, 2);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("ddddd"));
    query = new FuzzyQuery(new Term("field", "ddddX"), FuzzyQuery.defaultMinSimilarity, 3);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("ddddd"));
    query = new FuzzyQuery(new Term("field", "ddddX"), FuzzyQuery.defaultMinSimilarity, 4);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("ddddd"));
    query = new FuzzyQuery(new Term("field", "ddddX"), FuzzyQuery.defaultMinSimilarity, 5);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());
    

    // different field = no match:
    query = new FuzzyQuery(new Term("anotherfield", "ddddX"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());

    searcher.close();
    directory.close();
  
public voidtestFuzzinessLong()

    RAMDirectory directory = new RAMDirectory();
    IndexWriter writer = new IndexWriter(directory, new WhitespaceAnalyzer(), true);
    addDoc("aaaaaaa", writer);
    addDoc("segment", writer);
    writer.optimize();
    writer.close();
    IndexSearcher searcher = new IndexSearcher(directory);

    FuzzyQuery query;
    // not similar enough:
    query = new FuzzyQuery(new Term("field", "xxxxx"), FuzzyQuery.defaultMinSimilarity, 0);   
    Hits hits = searcher.search(query);
    assertEquals(0, hits.length());
    // edit distance to "aaaaaaa" = 3, this matches because the string is longer than
    // in testDefaultFuzziness so a bigger difference is allowed:
    query = new FuzzyQuery(new Term("field", "aaaaccc"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaaaa"));
    
    // now with prefix
    query = new FuzzyQuery(new Term("field", "aaaaccc"), FuzzyQuery.defaultMinSimilarity, 1);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaaaa"));
    query = new FuzzyQuery(new Term("field", "aaaaccc"), FuzzyQuery.defaultMinSimilarity, 4);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    assertEquals(hits.doc(0).get("field"), ("aaaaaaa"));
    query = new FuzzyQuery(new Term("field", "aaaaccc"), FuzzyQuery.defaultMinSimilarity, 5);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());

    // no match, more than half of the characters is wrong:
    query = new FuzzyQuery(new Term("field", "aaacccc"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());
    
    // now with prefix
    query = new FuzzyQuery(new Term("field", "aaacccc"), FuzzyQuery.defaultMinSimilarity, 2);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());

    // "student" and "stellent" are indeed similar to "segment" by default:
    query = new FuzzyQuery(new Term("field", "student"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    query = new FuzzyQuery(new Term("field", "stellent"), FuzzyQuery.defaultMinSimilarity, 0);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    
    // now with prefix
    query = new FuzzyQuery(new Term("field", "student"), FuzzyQuery.defaultMinSimilarity, 1);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    query = new FuzzyQuery(new Term("field", "stellent"), FuzzyQuery.defaultMinSimilarity, 1);   
    hits = searcher.search(query);
    assertEquals(1, hits.length());
    query = new FuzzyQuery(new Term("field", "student"), FuzzyQuery.defaultMinSimilarity, 2);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());
    query = new FuzzyQuery(new Term("field", "stellent"), FuzzyQuery.defaultMinSimilarity, 2);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());
    
    // "student" doesn't match anymore thanks to increased minimum similarity:
    query = new FuzzyQuery(new Term("field", "student"), 0.6f, 0);   
    hits = searcher.search(query);
    assertEquals(0, hits.length());

    try {
      query = new FuzzyQuery(new Term("field", "student"), 1.1f);
      fail("Expected IllegalArgumentException");
    } catch (IllegalArgumentException e) {
      // expecting exception
    }
    try {
      query = new FuzzyQuery(new Term("field", "student"), -0.1f);
      fail("Expected IllegalArgumentException");
    } catch (IllegalArgumentException e) {
      // expecting exception
    }

    searcher.close();
    directory.close();