public void TestClosestEditDistanceMatchComesFirst()
{
FuzzyLikeThisQuery flt = new FuzzyLikeThisQuery(10, analyzer);
flt.AddTerms("smith", "name", 0.3f, 1);
Query q = flt.Rewrite(searcher.IndexReader);
HashSet <Term> queryTerms = new HashSet <Term>();
q.ExtractTerms(queryTerms);
assertTrue("Should have variant smythe", queryTerms.contains(new Term("name", "smythe")));
assertTrue("Should have variant smith", queryTerms.contains(new Term("name", "smith")));
assertTrue("Should have variant smyth", queryTerms.contains(new Term("name", "smyth")));
TopDocs topDocs = searcher.Search(flt, 1);
ScoreDoc[] sd = topDocs.ScoreDocs;
assertTrue("score docs must match 1 doc", (sd != null) && (sd.Length > 0));
Document doc = searcher.Doc(sd[0].Doc);
assertEquals("Should match most similar not most rare variant", "2", doc.Get("id"));
}
public void TestNonExistingField()
{
FuzzyLikeThisQuery flt = new FuzzyLikeThisQuery(10, analyzer);
flt.AddTerms("jonathin smoth", "name", 0.3f, 1);
flt.AddTerms("jonathin smoth", "this field does not exist", 0.3f, 1);
// don't fail here just because the field doesn't exits
Query q = flt.Rewrite(searcher.IndexReader);
HashSet <Term> queryTerms = new HashSet <Term>();
q.ExtractTerms(queryTerms);
assertTrue("Should have variant jonathan", queryTerms.contains(new Term("name", "jonathan")));
assertTrue("Should have variant smith", queryTerms.contains(new Term("name", "smith")));
TopDocs topDocs = searcher.Search(flt, 1);
ScoreDoc[] sd = topDocs.ScoreDocs;
assertTrue("score docs must match 1 doc", (sd != null) && (sd.Length > 0));
Document doc = searcher.Doc(sd[0].Doc);
assertEquals("Should match most similar when using 2 words", "2", doc.Get("id"));
}
