public virtual void TestDocCount()
{
Directory dir = new RAMDirectory();
IndexWriter writer = null;
IndexReader reader = null;
int i;
try
{
writer = new IndexWriter(dir, new WhitespaceAnalyzer(), true);
// add 100 documents
for (i = 0; i < 100; i++)
{
AddDoc(writer);
}
Assert.AreEqual(100, writer.DocCount());
writer.Close();
// delete 40 documents
reader = IndexReader.Open(dir);
for (i = 0; i < 40; i++)
{
reader.Delete(i);
}
reader.Close();
// test doc count before segments are merged/index is optimized
writer = new IndexWriter(dir, new WhitespaceAnalyzer(), false);
Assert.AreEqual(100, writer.DocCount());
writer.Close();
reader = IndexReader.Open(dir);
Assert.AreEqual(100, reader.MaxDoc());
Assert.AreEqual(60, reader.NumDocs());
reader.Close();
// optimize the index and check that the new doc count is correct
writer = new IndexWriter(dir, new WhitespaceAnalyzer(), false);
writer.Optimize();
Assert.AreEqual(60, writer.DocCount());
writer.Close();
// check that the index reader gives the same numbers.
reader = IndexReader.Open(dir);
Assert.AreEqual(60, reader.MaxDoc());
Assert.AreEqual(60, reader.NumDocs());
reader.Close();
}
catch (System.IO.IOException e)
{
System.Console.Error.WriteLine(e.StackTrace);
}
}
public virtual void TestBasicDelete()
{
Directory dir = new RAMDirectory();
IndexWriter writer = null;
IndexReader reader = null;
Term searchTerm = new Term("content", "aaa");
// add 100 documents with term : aaa
writer = new IndexWriter(dir, new WhitespaceAnalyzer(), true);
for (int i = 0; i < 100; i++)
{
AddDoc(writer, searchTerm.Text());
}
writer.Close();
// OPEN READER AT THIS POINT - this should fix the view of the
// index at the point of having 100 "aaa" documents and 0 "bbb"
reader = IndexReader.Open(dir);
Assert.AreEqual(100, reader.DocFreq(searchTerm), "first docFreq");
AssertTermDocsCount("first reader", reader, searchTerm, 100);
// DELETE DOCUMENTS CONTAINING TERM: aaa
int deleted = 0;
reader = IndexReader.Open(dir);
deleted = reader.Delete(searchTerm);
Assert.AreEqual(100, deleted, "deleted count");
Assert.AreEqual(100, reader.DocFreq(searchTerm), "deleted docFreq");
AssertTermDocsCount("deleted termDocs", reader, searchTerm, 0);
reader.Close();
// CREATE A NEW READER and re-test
reader = IndexReader.Open(dir);
Assert.AreEqual(100, reader.DocFreq(searchTerm), "deleted docFreq");
AssertTermDocsCount("deleted termDocs", reader, searchTerm, 0);
reader.Close();
}
private void DeleteReaderReaderConflict(bool optimize)
{
Directory dir = GetDirectory(true);
Term searchTerm1 = new Term("content", "aaa");
Term searchTerm2 = new Term("content", "bbb");
Term searchTerm3 = new Term("content", "ccc");
// add 100 documents with term : aaa
// add 100 documents with term : bbb
// add 100 documents with term : ccc
IndexWriter writer = new IndexWriter(dir, new WhitespaceAnalyzer(), true);
for (int i = 0; i < 100; i++)
{
AddDoc(writer, searchTerm1.Text());
AddDoc(writer, searchTerm2.Text());
AddDoc(writer, searchTerm3.Text());
}
if (optimize)
{
writer.Optimize();
}
writer.Close();
// OPEN TWO READERS
// Both readers get segment info as exists at this time
IndexReader reader1 = IndexReader.Open(dir);
Assert.AreEqual(100, reader1.DocFreq(searchTerm1), "first opened");
Assert.AreEqual(100, reader1.DocFreq(searchTerm2), "first opened");
Assert.AreEqual(100, reader1.DocFreq(searchTerm3), "first opened");
AssertTermDocsCount("first opened", reader1, searchTerm1, 100);
AssertTermDocsCount("first opened", reader1, searchTerm2, 100);
AssertTermDocsCount("first opened", reader1, searchTerm3, 100);
IndexReader reader2 = IndexReader.Open(dir);
Assert.AreEqual(100, reader2.DocFreq(searchTerm1), "first opened");
Assert.AreEqual(100, reader2.DocFreq(searchTerm2), "first opened");
Assert.AreEqual(100, reader2.DocFreq(searchTerm3), "first opened");
AssertTermDocsCount("first opened", reader2, searchTerm1, 100);
AssertTermDocsCount("first opened", reader2, searchTerm2, 100);
AssertTermDocsCount("first opened", reader2, searchTerm3, 100);
// DELETE DOCS FROM READER 2 and CLOSE IT
// delete documents containing term: aaa
// when the reader is closed, the segment info is updated and
// the first reader is now stale
reader2.Delete(searchTerm1);
Assert.AreEqual(100, reader2.DocFreq(searchTerm1), "after delete 1");
Assert.AreEqual(100, reader2.DocFreq(searchTerm2), "after delete 1");
Assert.AreEqual(100, reader2.DocFreq(searchTerm3), "after delete 1");
AssertTermDocsCount("after delete 1", reader2, searchTerm1, 0);
AssertTermDocsCount("after delete 1", reader2, searchTerm2, 100);
AssertTermDocsCount("after delete 1", reader2, searchTerm3, 100);
reader2.Close();
// Make sure reader 1 is unchanged since it was open earlier
Assert.AreEqual(100, reader1.DocFreq(searchTerm1), "after delete 1");
Assert.AreEqual(100, reader1.DocFreq(searchTerm2), "after delete 1");
Assert.AreEqual(100, reader1.DocFreq(searchTerm3), "after delete 1");
AssertTermDocsCount("after delete 1", reader1, searchTerm1, 100);
AssertTermDocsCount("after delete 1", reader1, searchTerm2, 100);
AssertTermDocsCount("after delete 1", reader1, searchTerm3, 100);
// ATTEMPT TO DELETE FROM STALE READER
// delete documents containing term: bbb
try
{
reader1.Delete(searchTerm2);
Assert.Fail("Delete allowed from a stale index reader");
}
catch (System.IO.IOException e)
{
/* success */
}
// RECREATE READER AND TRY AGAIN
reader1.Close();
reader1 = IndexReader.Open(dir);
Assert.AreEqual(100, reader1.DocFreq(searchTerm1), "reopened");
Assert.AreEqual(100, reader1.DocFreq(searchTerm2), "reopened");
Assert.AreEqual(100, reader1.DocFreq(searchTerm3), "reopened");
AssertTermDocsCount("reopened", reader1, searchTerm1, 0);
AssertTermDocsCount("reopened", reader1, searchTerm2, 100);
AssertTermDocsCount("reopened", reader1, searchTerm3, 100);
reader1.Delete(searchTerm2);
Assert.AreEqual(100, reader1.DocFreq(searchTerm1), "deleted 2");
Assert.AreEqual(100, reader1.DocFreq(searchTerm2), "deleted 2");
Assert.AreEqual(100, reader1.DocFreq(searchTerm3), "deleted 2");
AssertTermDocsCount("deleted 2", reader1, searchTerm1, 0);
AssertTermDocsCount("deleted 2", reader1, searchTerm2, 0);
AssertTermDocsCount("deleted 2", reader1, searchTerm3, 100);
reader1.Close();
// Open another reader to confirm that everything is deleted
reader2 = IndexReader.Open(dir);
Assert.AreEqual(100, reader2.DocFreq(searchTerm1), "reopened 2");
Assert.AreEqual(100, reader2.DocFreq(searchTerm2), "reopened 2");
Assert.AreEqual(100, reader2.DocFreq(searchTerm3), "reopened 2");
AssertTermDocsCount("reopened 2", reader2, searchTerm1, 0);
AssertTermDocsCount("reopened 2", reader2, searchTerm2, 0);
AssertTermDocsCount("reopened 2", reader2, searchTerm3, 100);
reader2.Close();
dir.Close();
}
private void DeleteReaderWriterConflict(bool optimize)
{
//Directory dir = new RAMDirectory();
Directory dir = GetDirectory(true);
Term searchTerm = new Term("content", "aaa");
Term searchTerm2 = new Term("content", "bbb");
// add 100 documents with term : aaa
IndexWriter writer = new IndexWriter(dir, new WhitespaceAnalyzer(), true);
for (int i = 0; i < 100; i++)
{
AddDoc(writer, searchTerm.Text());
}
writer.Close();
// OPEN READER AT THIS POINT - this should fix the view of the
// index at the point of having 100 "aaa" documents and 0 "bbb"
IndexReader reader = IndexReader.Open(dir);
Assert.AreEqual(100, reader.DocFreq(searchTerm), "first docFreq");
Assert.AreEqual(0, reader.DocFreq(searchTerm2), "first docFreq");
AssertTermDocsCount("first reader", reader, searchTerm, 100);
AssertTermDocsCount("first reader", reader, searchTerm2, 0);
// add 100 documents with term : bbb
writer = new IndexWriter(dir, new WhitespaceAnalyzer(), false);
for (int i = 0; i < 100; i++)
{
AddDoc(writer, searchTerm2.Text());
}
// REQUEST OPTIMIZATION
// This causes a new segment to become current for all subsequent
// searchers. Because of this, deletions made via a previously open
// reader, which would be applied to that reader's segment, are lost
// for subsequent searchers/readers
if (optimize)
{
writer.Optimize();
}
writer.Close();
// The reader should not see the new data
Assert.AreEqual(100, reader.DocFreq(searchTerm), "first docFreq");
Assert.AreEqual(0, reader.DocFreq(searchTerm2), "first docFreq");
AssertTermDocsCount("first reader", reader, searchTerm, 100);
AssertTermDocsCount("first reader", reader, searchTerm2, 0);
// DELETE DOCUMENTS CONTAINING TERM: aaa
// NOTE: the reader was created when only "aaa" documents were in
int deleted = 0;
try
{
deleted = reader.Delete(searchTerm);
Assert.Fail("Delete allowed on an index reader with stale segment information");
}
catch (System.IO.IOException e)
{
/* success */
}
// Re-open index reader and try again. This time it should see
// the new data.
reader.Close();
reader = IndexReader.Open(dir);
Assert.AreEqual(100, reader.DocFreq(searchTerm), "first docFreq");
Assert.AreEqual(100, reader.DocFreq(searchTerm2), "first docFreq");
AssertTermDocsCount("first reader", reader, searchTerm, 100);
AssertTermDocsCount("first reader", reader, searchTerm2, 100);
deleted = reader.Delete(searchTerm);
Assert.AreEqual(100, deleted, "deleted count");
Assert.AreEqual(100, reader.DocFreq(searchTerm), "deleted docFreq");
Assert.AreEqual(100, reader.DocFreq(searchTerm2), "deleted docFreq");
AssertTermDocsCount("deleted termDocs", reader, searchTerm, 0);
AssertTermDocsCount("deleted termDocs", reader, searchTerm2, 100);
reader.Close();
// CREATE A NEW READER and re-test
reader = IndexReader.Open(dir);
Assert.AreEqual(100, reader.DocFreq(searchTerm), "deleted docFreq");
Assert.AreEqual(100, reader.DocFreq(searchTerm2), "deleted docFreq");
AssertTermDocsCount("deleted termDocs", reader, searchTerm, 0);
AssertTermDocsCount("deleted termDocs", reader, searchTerm2, 100);
reader.Close();
}
protected internal override void DoDelete(int n)
{
in_Renamed.Delete(n);
}
