public void TestKeepsLastFilter()
{
DuplicateFilter df = new DuplicateFilter(KEY_FIELD);
df.KeepMode = (KeepMode.KM_USE_LAST_OCCURRENCE);
ScoreDoc[] hits = searcher.Search(tq, df, 1000).ScoreDocs;
assertTrue("Filtered searching should have found some matches", hits.Length > 0);
foreach (ScoreDoc hit in hits)
{
Document d = searcher.Doc(hit.Doc);
string url = d.Get(KEY_FIELD);
DocsEnum td = TestUtil.Docs(Random(), reader,
KEY_FIELD,
new BytesRef(url),
MultiFields.GetLiveDocs(reader),
null,
0);
int lastDoc = 0;
while (td.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
lastDoc = td.DocID;
}
assertEquals("Duplicate urls should return last doc", lastDoc, hit.Doc);
}
}
public void TestDefaultFilter()
{
DuplicateFilter df = new DuplicateFilter(KEY_FIELD);
HashSet <string> results = new HashSet <string>();
ScoreDoc[] hits = searcher.Search(tq, df, 1000).ScoreDocs;
foreach (ScoreDoc hit in hits)
{
Document d = searcher.Doc(hit.Doc);
string url = d.Get(KEY_FIELD);
assertFalse("No duplicate urls should be returned", results.contains(url));
results.add(url);
}
}
public override bool Equals(object obj)
{
if (this == obj)
{
return(true);
}
if ((obj == null) || (obj.GetType() != this.GetType()))
{
return(false);
}
DuplicateFilter other = (DuplicateFilter)obj;
return(keepMode == other.keepMode &&
processingMode == other.processingMode &&
fieldName != null && fieldName.Equals(other.fieldName, StringComparison.Ordinal));
}
public void TestFastFilter()
{
DuplicateFilter df = new DuplicateFilter(KEY_FIELD);
df.ProcessingMode = (ProcessingMode.PM_FAST_INVALIDATION);
HashSet <string> results = new HashSet <string>();
ScoreDoc[] hits = searcher.Search(tq, df, 1000).ScoreDocs;
assertTrue("Filtered searching should have found some matches", hits.Length > 0);
foreach (ScoreDoc hit in hits)
{
Document d = searcher.Doc(hit.Doc);
string url = d.Get(KEY_FIELD);
assertFalse("No duplicate urls should be returned", results.contains(url));
results.add(url);
}
assertEquals("Two urls found", 2, results.size());
}