TestFuzzyQuerypublic class TestFuzzyQuery extends TestCase | Tests {@link FuzzyQuery}. |
| Methods Summary |
|---|
| private void | addDoc(java.lang.String text, org.apache.lucene.index.IndexWriter writer)
Document doc = new Document();
doc.add(Field.Text("field", text));
writer.addDocument(doc);
| | public void | testDefaultFuzziness()
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"));
Hits hits = searcher.search(query);
assertEquals(3, hits.length());
// not similar enough:
query = new FuzzyQuery(new Term("field", "xxxxx"));
hits = searcher.search(query);
assertEquals(0, hits.length());
query = new FuzzyQuery(new Term("field", "aaccc")); // 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"));
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"));
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", "ddddX"));
hits = searcher.search(query);
assertEquals(1, hits.length());
assertEquals(hits.doc(0).get("field"), ("ddddd"));
// different field = no match:
query = new FuzzyQuery(new Term("anotherfield", "ddddX"));
hits = searcher.search(query);
assertEquals(0, hits.length());
searcher.close();
directory.close();
| | public void | testDefaultFuzzinessLong()
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"));
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"));
hits = searcher.search(query);
assertEquals(1, hits.length());
assertEquals(hits.doc(0).get("field"), ("aaaaaaa"));
// no match, more than half of the characters is wrong:
query = new FuzzyQuery(new Term("field", "aaacccc"));
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"));
hits = searcher.search(query);
assertEquals(1, hits.length());
query = new FuzzyQuery(new Term("field", "stellent"));
hits = searcher.search(query);
assertEquals(1, hits.length());
searcher.close();
directory.close();
|
|
