|
public SetRAMBufferSizeMB ( double ramBufferSizeMB ) : |
||
| ramBufferSizeMB | double | |
| return | ||
private IndexWriterConfig NewWriterConfig()
{
IndexWriterConfig conf = NewIndexWriterConfig(TEST_VERSION_CURRENT, null);
conf.SetMaxBufferedDocs(IndexWriterConfig.DISABLE_AUTO_FLUSH);
conf.SetRAMBufferSizeMB(IndexWriterConfig.DEFAULT_RAM_BUFFER_SIZE_MB);
// prevent any merges by default.
conf.SetMergePolicy(NoMergePolicy.COMPOUND_FILES);
return(conf);
}
[Slow] // LUCENENET: occasionally
public virtual void TestStallControl()
{
// LUCENENET specific - disable the test if asserts are not enabled
AssumeTrue("This test requires asserts to be enabled.", Debugging.AssertsEnabled);
int[] numThreads = new int[] { 4 + Random.Next(8), 1 };
int numDocumentsToIndex = 50 + Random.Next(50);
for (int i = 0; i < numThreads.Length; i++)
{
AtomicInt32 numDocs = new AtomicInt32(numDocumentsToIndex);
MockDirectoryWrapper dir = NewMockDirectory();
// mock a very slow harddisk sometimes here so that flushing is very slow
dir.Throttling = Throttling.SOMETIMES;
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random));
iwc.SetMaxBufferedDocs(IndexWriterConfig.DISABLE_AUTO_FLUSH);
iwc.SetMaxBufferedDeleteTerms(IndexWriterConfig.DISABLE_AUTO_FLUSH);
FlushPolicy flushPolicy = new FlushByRamOrCountsPolicy();
iwc.SetFlushPolicy(flushPolicy);
DocumentsWriterPerThreadPool threadPool = new DocumentsWriterPerThreadPool(numThreads[i] == 1 ? 1 : 2);
iwc.SetIndexerThreadPool(threadPool);
// with such a small ram buffer we should be stalled quiet quickly
iwc.SetRAMBufferSizeMB(0.25);
IndexWriter writer = new IndexWriter(dir, iwc);
IndexThread[] threads = new IndexThread[numThreads[i]];
for (int x = 0; x < threads.Length; x++)
{
threads[x] = new IndexThread(numDocs, writer, lineDocFile, false);
threads[x].Start();
}
for (int x = 0; x < threads.Length; x++)
{
threads[x].Join();
}
DocumentsWriter docsWriter = writer.DocsWriter;
Assert.IsNotNull(docsWriter);
DocumentsWriterFlushControl flushControl = docsWriter.flushControl;
Assert.AreEqual(0, flushControl.FlushBytes, " all flushes must be due");
Assert.AreEqual(numDocumentsToIndex, writer.NumDocs);
Assert.AreEqual(numDocumentsToIndex, writer.MaxDoc);
if (numThreads[i] == 1)
{
assertFalse("single thread must not block numThreads: " + numThreads[i], docsWriter.flushControl.stallControl.HasBlocked);
}
if (docsWriter.flushControl.peakNetBytes > (2d * iwc.RAMBufferSizeMB * 1024d * 1024d))
{
Assert.IsTrue(docsWriter.flushControl.stallControl.WasStalled);
}
AssertActiveBytesAfter(flushControl);
writer.Dispose(true);
dir.Dispose();
}
}
public virtual void TestFlushDocCount()
{
// LUCENENET specific - disable the test if asserts are not enabled
AssumeTrue("This test requires asserts to be enabled.", Debugging.AssertsEnabled);
int[] numThreads = new int[] { 2 + AtLeast(1), 1 };
for (int i = 0; i < numThreads.Length; i++)
{
int numDocumentsToIndex = 50 + AtLeast(30);
AtomicInt32 numDocs = new AtomicInt32(numDocumentsToIndex);
Directory dir = NewDirectory();
MockDefaultFlushPolicy flushPolicy = new MockDefaultFlushPolicy();
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random)).SetFlushPolicy(flushPolicy);
int numDWPT = 1 + AtLeast(2);
DocumentsWriterPerThreadPool threadPool = new DocumentsWriterPerThreadPool(numDWPT);
iwc.SetIndexerThreadPool(threadPool);
iwc.SetMaxBufferedDocs(2 + AtLeast(10));
iwc.SetRAMBufferSizeMB(IndexWriterConfig.DISABLE_AUTO_FLUSH);
iwc.SetMaxBufferedDeleteTerms(IndexWriterConfig.DISABLE_AUTO_FLUSH);
IndexWriter writer = new IndexWriter(dir, iwc);
flushPolicy = (MockDefaultFlushPolicy)writer.Config.FlushPolicy;
Assert.IsTrue(flushPolicy.FlushOnDocCount);
Assert.IsFalse(flushPolicy.FlushOnDeleteTerms);
Assert.IsFalse(flushPolicy.FlushOnRAM);
DocumentsWriter docsWriter = writer.DocsWriter;
Assert.IsNotNull(docsWriter);
DocumentsWriterFlushControl flushControl = docsWriter.flushControl;
Assert.AreEqual(0, flushControl.FlushBytes, " bytes must be 0 after init");
IndexThread[] threads = new IndexThread[numThreads[i]];
for (int x = 0; x < threads.Length; x++)
{
threads[x] = new IndexThread(numDocs, writer, lineDocFile, false);
threads[x].Start();
}
for (int x = 0; x < threads.Length; x++)
{
threads[x].Join();
}
Assert.AreEqual(0, flushControl.FlushBytes, " all flushes must be due numThreads=" + numThreads[i]);
Assert.AreEqual(numDocumentsToIndex, writer.NumDocs);
Assert.AreEqual(numDocumentsToIndex, writer.MaxDoc);
Assert.IsTrue(flushPolicy.peakDocCountWithoutFlush <= iwc.MaxBufferedDocs, "peak bytes without flush exceeded watermark");
AssertActiveBytesAfter(flushControl);
writer.Dispose();
Assert.AreEqual(0, flushControl.ActiveBytes);
dir.Dispose();
}
}
public virtual void TestInvalidValues()
{
IndexWriterConfig conf = new IndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
// Test IndexDeletionPolicy
Assert.AreEqual(typeof(KeepOnlyLastCommitDeletionPolicy), conf.DelPolicy.GetType());
conf.SetIndexDeletionPolicy(new SnapshotDeletionPolicy(null));
Assert.AreEqual(typeof(SnapshotDeletionPolicy), conf.DelPolicy.GetType());
try
{
conf.SetIndexDeletionPolicy(null);
Assert.Fail();
}
catch (System.ArgumentException e)
{
// ok
}
// Test MergeScheduler
Assert.AreEqual(typeof(ConcurrentMergeScheduler), conf.MergeScheduler.GetType());
conf.SetMergeScheduler(new SerialMergeScheduler());
Assert.AreEqual(typeof(SerialMergeScheduler), conf.MergeScheduler.GetType());
try
{
conf.SetMergeScheduler(null);
Assert.Fail();
}
catch (System.ArgumentException e)
{
// ok
}
// Test Similarity:
// we shouldnt assert what the default is, just that its not null.
Assert.IsTrue(IndexSearcher.DefaultSimilarity == conf.Similarity);
conf.SetSimilarity(new MySimilarity());
Assert.AreEqual(typeof(MySimilarity), conf.Similarity.GetType());
try
{
conf.SetSimilarity(null);
Assert.Fail();
}
catch (System.ArgumentException e)
{
// ok
}
// Test IndexingChain
Assert.IsTrue(DocumentsWriterPerThread.DefaultIndexingChain == conf.IndexingChain);
conf.SetIndexingChain(new MyIndexingChain());
Assert.AreEqual(typeof(MyIndexingChain), conf.IndexingChain.GetType());
try
{
conf.SetIndexingChain(null);
Assert.Fail();
}
catch (System.ArgumentException e)
{
// ok
}
try
{
conf.SetMaxBufferedDeleteTerms(0);
Assert.Fail("should not have succeeded to set maxBufferedDeleteTerms to 0");
}
catch (System.ArgumentException e)
{
// this is expected
}
try
{
conf.SetMaxBufferedDocs(1);
Assert.Fail("should not have succeeded to set maxBufferedDocs to 1");
}
catch (System.ArgumentException e)
{
// this is expected
}
try
{
// Disable both MAX_BUF_DOCS and RAM_SIZE_MB
conf.SetMaxBufferedDocs(4);
conf.SetRAMBufferSizeMB(IndexWriterConfig.DISABLE_AUTO_FLUSH);
conf.SetMaxBufferedDocs(IndexWriterConfig.DISABLE_AUTO_FLUSH);
Assert.Fail("should not have succeeded to disable maxBufferedDocs when ramBufferSizeMB is disabled as well");
}
catch (System.ArgumentException e)
{
// this is expected
}
conf.SetRAMBufferSizeMB(IndexWriterConfig.DEFAULT_RAM_BUFFER_SIZE_MB);
conf.SetMaxBufferedDocs(IndexWriterConfig.DEFAULT_MAX_BUFFERED_DOCS);
try
{
conf.SetRAMBufferSizeMB(IndexWriterConfig.DISABLE_AUTO_FLUSH);
Assert.Fail("should not have succeeded to disable ramBufferSizeMB when maxBufferedDocs is disabled as well");
}
catch (System.ArgumentException e)
{
// this is expected
}
// Test setReaderTermsIndexDivisor
try
{
conf.SetReaderTermsIndexDivisor(0);
Assert.Fail("should not have succeeded to set termsIndexDivisor to 0");
}
catch (System.ArgumentException e)
{
// this is expected
}
// Setting to -1 is ok
conf.SetReaderTermsIndexDivisor(-1);
try
{
conf.SetReaderTermsIndexDivisor(-2);
Assert.Fail("should not have succeeded to set termsIndexDivisor to < -1");
}
catch (System.ArgumentException e)
{
// this is expected
}
try
{
conf.SetRAMPerThreadHardLimitMB(2048);
Assert.Fail("should not have succeeded to set RAMPerThreadHardLimitMB to >= 2048");
}
catch (System.ArgumentException e)
{
// this is expected
}
try
{
conf.SetRAMPerThreadHardLimitMB(0);
Assert.Fail("should not have succeeded to set RAMPerThreadHardLimitMB to 0");
}
catch (System.ArgumentException e)
{
// this is expected
}
// Test MergePolicy
Assert.AreEqual(typeof(TieredMergePolicy), conf.MergePolicy.GetType());
conf.SetMergePolicy(new LogDocMergePolicy());
Assert.AreEqual(typeof(LogDocMergePolicy), conf.MergePolicy.GetType());
try
{
conf.SetMergePolicy(null);
Assert.Fail();
}
catch (System.ArgumentException e)
{
// ok
}
}
public virtual void TestTonsOfUpdates()
{
// LUCENE-5248: make sure that when there are many updates, we don't use too much RAM
Directory dir = NewDirectory();
Random random = Random;
IndexWriterConfig conf = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random));
conf.SetRAMBufferSizeMB(IndexWriterConfig.DEFAULT_RAM_BUFFER_SIZE_MB);
conf.SetMaxBufferedDocs(IndexWriterConfig.DISABLE_AUTO_FLUSH); // don't flush by doc
IndexWriter writer = new IndexWriter(dir, conf);
// test data: lots of documents (few 10Ks) and lots of update terms (few hundreds)
int numDocs = AtLeast(20000);
int numBinaryFields = AtLeast(5);
int numTerms = TestUtil.NextInt32(random, 10, 100); // terms should affect many docs
ISet <string> updateTerms = new JCG.HashSet <string>();
while (updateTerms.Count < numTerms)
{
updateTerms.Add(TestUtil.RandomSimpleString(random));
}
// System.out.println("numDocs=" + numDocs + " numBinaryFields=" + numBinaryFields + " numTerms=" + numTerms);
// build a large index with many BDV fields and update terms
for (int i = 0; i < numDocs; i++)
{
Document doc = new Document();
int numUpdateTerms = TestUtil.NextInt32(random, 1, numTerms / 10);
for (int j = 0; j < numUpdateTerms; j++)
{
doc.Add(new StringField("upd", RandomPicks.RandomFrom(random, updateTerms), Store.NO));
}
for (int j = 0; j < numBinaryFields; j++)
{
long val = random.Next();
doc.Add(new BinaryDocValuesField("f" + j, TestBinaryDocValuesUpdates.ToBytes(val)));
doc.Add(new NumericDocValuesField("cf" + j, val * 2));
}
writer.AddDocument(doc);
}
writer.Commit(); // commit so there's something to apply to
// set to flush every 2048 bytes (approximately every 12 updates), so we get
// many flushes during binary updates
writer.Config.SetRAMBufferSizeMB(2048.0 / 1024 / 1024);
int numUpdates = AtLeast(100);
// System.out.println("numUpdates=" + numUpdates);
for (int i = 0; i < numUpdates; i++)
{
int field = random.Next(numBinaryFields);
Term updateTerm = new Term("upd", RandomPicks.RandomFrom(random, updateTerms));
long value = random.Next();
writer.UpdateBinaryDocValue(updateTerm, "f" + field, TestBinaryDocValuesUpdates.ToBytes(value));
writer.UpdateNumericDocValue(updateTerm, "cf" + field, value * 2);
}
writer.Dispose();
DirectoryReader reader = DirectoryReader.Open(dir);
BytesRef scratch = new BytesRef();
foreach (AtomicReaderContext context in reader.Leaves)
{
for (int i = 0; i < numBinaryFields; i++)
{
AtomicReader r = context.AtomicReader;
BinaryDocValues f = r.GetBinaryDocValues("f" + i);
NumericDocValues cf = r.GetNumericDocValues("cf" + i);
for (int j = 0; j < r.MaxDoc; j++)
{
Assert.AreEqual(cf.Get(j), TestBinaryDocValuesUpdates.GetValue(f, j, scratch) * 2, "reader=" + r + ", field=f" + i + ", doc=" + j);
}
}
}
reader.Dispose();
dir.Dispose();
}
public virtual void TestDeletes1()
{
//IndexWriter.debug2 = System.out;
Directory dir = new MockDirectoryWrapper(new J2N.Randomizer(Random.NextInt64()), new RAMDirectory());
IndexWriterConfig iwc = new IndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random));
iwc.SetMergeScheduler(new SerialMergeScheduler());
iwc.SetMaxBufferedDocs(5000);
iwc.SetRAMBufferSizeMB(100);
RangeMergePolicy fsmp = new RangeMergePolicy(this, false);
iwc.SetMergePolicy(fsmp);
IndexWriter writer = new IndexWriter(dir, iwc);
for (int x = 0; x < 5; x++)
{
writer.AddDocument(DocHelper.CreateDocument(x, "1", 2));
//System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
//System.out.println("commit1");
writer.Commit();
Assert.AreEqual(1, writer.SegmentCount);
for (int x = 5; x < 10; x++)
{
writer.AddDocument(DocHelper.CreateDocument(x, "2", 2));
//System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
//System.out.println("commit2");
writer.Commit();
Assert.AreEqual(2, writer.SegmentCount);
for (int x = 10; x < 15; x++)
{
writer.AddDocument(DocHelper.CreateDocument(x, "3", 2));
//System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
writer.DeleteDocuments(new Term("id", "1"));
writer.DeleteDocuments(new Term("id", "11"));
// flushing without applying deletes means
// there will still be deletes in the segment infos
writer.Flush(false, false);
Assert.IsTrue(writer.bufferedUpdatesStream.Any());
// get reader flushes pending deletes
// so there should not be anymore
IndexReader r1 = writer.GetReader();
Assert.IsFalse(writer.bufferedUpdatesStream.Any());
r1.Dispose();
// delete id:2 from the first segment
// merge segments 0 and 1
// which should apply the delete id:2
writer.DeleteDocuments(new Term("id", "2"));
writer.Flush(false, false);
fsmp = (RangeMergePolicy)writer.Config.MergePolicy;
fsmp.doMerge = true;
fsmp.start = 0;
fsmp.length = 2;
writer.MaybeMerge();
Assert.AreEqual(2, writer.SegmentCount);
// id:2 shouldn't exist anymore because
// it's been applied in the merge and now it's gone
IndexReader r2 = writer.GetReader();
int[] id2docs = ToDocsArray(new Term("id", "2"), null, r2);
Assert.IsTrue(id2docs is null);
r2.Dispose();
/*
* /// // added docs are in the ram buffer
* /// for (int x = 15; x < 20; x++) {
* /// writer.AddDocument(TestIndexWriterReader.CreateDocument(x, "4", 2));
* /// System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
* /// }
* /// Assert.IsTrue(writer.numRamDocs() > 0);
* /// // delete from the ram buffer
* /// writer.DeleteDocuments(new Term("id", Integer.toString(13)));
* ///
* /// Term id3 = new Term("id", Integer.toString(3));
* ///
* /// // delete from the 1st segment
* /// writer.DeleteDocuments(id3);
* ///
* /// Assert.IsTrue(writer.numRamDocs() > 0);
* ///
* /// //System.out
* /// // .println("segdels1:" + writer.docWriter.deletesToString());
* ///
* /// //Assert.IsTrue(writer.docWriter.segmentDeletes.Size() > 0);
* ///
* /// // we cause a merge to happen
* /// fsmp.doMerge = true;
* /// fsmp.start = 0;
* /// fsmp.length = 2;
* /// System.out.println("maybeMerge "+writer.SegmentInfos);
* ///
* /// SegmentInfo info0 = writer.SegmentInfos[0];
* /// SegmentInfo info1 = writer.SegmentInfos[1];
* ///
* /// writer.MaybeMerge();
* /// System.out.println("maybeMerge after "+writer.SegmentInfos);
* /// // there should be docs in RAM
* /// Assert.IsTrue(writer.numRamDocs() > 0);
* ///
* /// // assert we've merged the 1 and 2 segments
* /// // and still have a segment leftover == 2
* /// Assert.AreEqual(2, writer.SegmentInfos.Size());
* /// Assert.IsFalse(segThere(info0, writer.SegmentInfos));
* /// Assert.IsFalse(segThere(info1, writer.SegmentInfos));
* ///
* /// //System.out.println("segdels2:" + writer.docWriter.deletesToString());
* ///
* /// //Assert.IsTrue(writer.docWriter.segmentDeletes.Size() > 0);
* ///
* /// IndexReader r = writer.GetReader();
* /// IndexReader r1 = r.getSequentialSubReaders()[0];
* /// printDelDocs(r1.GetLiveDocs());
* /// int[] docs = toDocsArray(id3, null, r);
* /// System.out.println("id3 docs:"+Arrays.toString(docs));
* /// // there shouldn't be any docs for id:3
* /// Assert.IsTrue(docs is null);
* /// r.Dispose();
* ///
* /// part2(writer, fsmp);
* ///
*/
// System.out.println("segdels2:"+writer.docWriter.segmentDeletes.toString());
//System.out.println("close");
writer.Dispose();
dir.Dispose();
}
protected internal virtual void RunFlushByRam(int numThreads, double maxRamMB, bool ensureNotStalled)
{
int numDocumentsToIndex = 10 + AtLeast(30);
AtomicInt32 numDocs = new AtomicInt32(numDocumentsToIndex);
Directory dir = NewDirectory();
MockDefaultFlushPolicy flushPolicy = new MockDefaultFlushPolicy();
MockAnalyzer analyzer = new MockAnalyzer(Random);
analyzer.MaxTokenLength = TestUtil.NextInt32(Random, 1, IndexWriter.MAX_TERM_LENGTH);
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, analyzer).SetFlushPolicy(flushPolicy);
int numDWPT = 1 + AtLeast(2);
DocumentsWriterPerThreadPool threadPool = new ThreadAffinityDocumentsWriterThreadPool(numDWPT);
iwc.SetIndexerThreadPool(threadPool);
iwc.SetRAMBufferSizeMB(maxRamMB);
iwc.SetMaxBufferedDocs(IndexWriterConfig.DISABLE_AUTO_FLUSH);
iwc.SetMaxBufferedDeleteTerms(IndexWriterConfig.DISABLE_AUTO_FLUSH);
IndexWriter writer = new IndexWriter(dir, iwc);
flushPolicy = (MockDefaultFlushPolicy)writer.Config.FlushPolicy;
Assert.IsFalse(flushPolicy.FlushOnDocCount);
Assert.IsFalse(flushPolicy.FlushOnDeleteTerms);
Assert.IsTrue(flushPolicy.FlushOnRAM);
DocumentsWriter docsWriter = writer.DocsWriter;
Assert.IsNotNull(docsWriter);
DocumentsWriterFlushControl flushControl = docsWriter.flushControl;
Assert.AreEqual(0, flushControl.FlushBytes, " bytes must be 0 after init");
IndexThread[] threads = new IndexThread[numThreads];
for (int x = 0; x < threads.Length; x++)
{
threads[x] = new IndexThread(this, numDocs, numThreads, writer, LineDocFile, false);
threads[x].Start();
}
for (int x = 0; x < threads.Length; x++)
{
threads[x].Join();
}
long maxRAMBytes = (long)(iwc.RAMBufferSizeMB * 1024.0 * 1024.0);
Assert.AreEqual(0, flushControl.FlushBytes, " all flushes must be due numThreads=" + numThreads);
Assert.AreEqual(numDocumentsToIndex, writer.NumDocs);
Assert.AreEqual(numDocumentsToIndex, writer.MaxDoc);
Assert.IsTrue(flushPolicy.PeakBytesWithoutFlush <= maxRAMBytes, "peak bytes without flush exceeded watermark");
AssertActiveBytesAfter(flushControl);
if (flushPolicy.HasMarkedPending)
{
Assert.IsTrue(maxRAMBytes < flushControl.peakActiveBytes);
}
if (ensureNotStalled)
{
Assert.IsFalse(docsWriter.flushControl.stallControl.WasStalled);
}
writer.Dispose();
Assert.AreEqual(0, flushControl.ActiveBytes);
dir.Dispose();
}
public virtual void DoTestLongPostingsNoPositions(IndexOptions options)
{
// Don't use TestUtil.getTempDir so that we own the
// randomness (ie same seed will point to same dir):
Directory dir = NewFSDirectory(CreateTempDir("longpostings" + "." + Random.NextInt64()));
int NUM_DOCS = AtLeast(2000);
if (VERBOSE)
{
Console.WriteLine("TEST: NUM_DOCS=" + NUM_DOCS);
}
string s1 = GetRandomTerm(null);
string s2 = GetRandomTerm(s1);
if (VERBOSE)
{
Console.WriteLine("\nTEST: s1=" + s1 + " s2=" + s2);
/*
* for(int idx=0;idx<s1.Length();idx++) {
* System.out.println(" s1 ch=0x" + Integer.toHexString(s1.charAt(idx)));
* }
* for(int idx=0;idx<s2.Length();idx++) {
* System.out.println(" s2 ch=0x" + Integer.toHexString(s2.charAt(idx)));
* }
*/
}
FixedBitSet isS1 = new FixedBitSet(NUM_DOCS);
for (int idx = 0; idx < NUM_DOCS; idx++)
{
if (Random.NextBoolean())
{
isS1.Set(idx);
}
}
IndexReader r;
if (true)
{
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random)).SetOpenMode(OpenMode.CREATE).SetMergePolicy(NewLogMergePolicy());
iwc.SetRAMBufferSizeMB(16.0 + 16.0 * Random.NextDouble());
iwc.SetMaxBufferedDocs(-1);
RandomIndexWriter riw = new RandomIndexWriter(Random, dir, iwc);
FieldType ft = new FieldType(TextField.TYPE_NOT_STORED);
ft.IndexOptions = options;
for (int idx = 0; idx < NUM_DOCS; idx++)
{
Document doc = new Document();
string s = isS1.Get(idx) ? s1 : s2;
Field f = NewField("field", s, ft);
int count = TestUtil.NextInt32(Random, 1, 4);
for (int ct = 0; ct < count; ct++)
{
doc.Add(f);
}
riw.AddDocument(doc);
}
r = riw.GetReader();
riw.Dispose();
}
else
{
#pragma warning disable 162
r = DirectoryReader.Open(dir);
#pragma warning restore 162
}
/*
* if (VERBOSE) {
* System.out.println("TEST: terms");
* TermEnum termEnum = r.Terms();
* while(termEnum.Next()) {
* System.out.println(" term=" + termEnum.Term() + " len=" + termEnum.Term().Text().Length());
* Assert.IsTrue(termEnum.DocFreq() > 0);
* System.out.println(" s1?=" + (termEnum.Term().Text().equals(s1)) + " s1len=" + s1.Length());
* System.out.println(" s2?=" + (termEnum.Term().Text().equals(s2)) + " s2len=" + s2.Length());
* final String s = termEnum.Term().Text();
* for(int idx=0;idx<s.Length();idx++) {
* System.out.println(" ch=0x" + Integer.toHexString(s.charAt(idx)));
* }
* }
* }
*/
Assert.AreEqual(NUM_DOCS, r.NumDocs);
Assert.IsTrue(r.DocFreq(new Term("field", s1)) > 0);
Assert.IsTrue(r.DocFreq(new Term("field", s2)) > 0);
int num = AtLeast(1000);
for (int iter = 0; iter < num; iter++)
{
string term;
bool doS1;
if (Random.NextBoolean())
{
term = s1;
doS1 = true;
}
else
{
term = s2;
doS1 = false;
}
if (VERBOSE)
{
Console.WriteLine("\nTEST: iter=" + iter + " doS1=" + doS1 + " term=" + term);
}
DocsEnum docs;
DocsEnum postings;
if (options == IndexOptions.DOCS_ONLY)
{
docs = TestUtil.Docs(Random, r, "field", new BytesRef(term), null, null, DocsFlags.NONE);
postings = null;
}
else
{
docs = postings = TestUtil.Docs(Random, r, "field", new BytesRef(term), null, null, DocsFlags.FREQS);
Debug.Assert(postings != null);
}
Debug.Assert(docs != null);
int docID = -1;
while (docID < DocIdSetIterator.NO_MORE_DOCS)
{
int what = Random.Next(3);
if (what == 0)
{
if (VERBOSE)
{
Console.WriteLine("TEST: docID=" + docID + "; do next()");
}
// nextDoc
int expected = docID + 1;
while (true)
{
if (expected == NUM_DOCS)
{
expected = int.MaxValue;
break;
}
else if (isS1.Get(expected) == doS1)
{
break;
}
else
{
expected++;
}
}
docID = docs.NextDoc();
if (VERBOSE)
{
Console.WriteLine(" got docID=" + docID);
}
Assert.AreEqual(expected, docID);
if (docID == DocIdSetIterator.NO_MORE_DOCS)
{
break;
}
if (Random.Next(6) == 3 && postings != null)
{
int freq = postings.Freq;
Assert.IsTrue(freq >= 1 && freq <= 4);
}
}
else
{
// advance
int targetDocID;
if (docID == -1)
{
targetDocID = Random.Next(NUM_DOCS + 1);
}
else
{
targetDocID = docID + TestUtil.NextInt32(Random, 1, NUM_DOCS - docID);
}
if (VERBOSE)
{
Console.WriteLine("TEST: docID=" + docID + "; do advance(" + targetDocID + ")");
}
int expected = targetDocID;
while (true)
{
if (expected == NUM_DOCS)
{
expected = int.MaxValue;
break;
}
else if (isS1.Get(expected) == doS1)
{
break;
}
else
{
expected++;
}
}
docID = docs.Advance(targetDocID);
if (VERBOSE)
{
Console.WriteLine(" got docID=" + docID);
}
Assert.AreEqual(expected, docID);
if (docID == DocIdSetIterator.NO_MORE_DOCS)
{
break;
}
if (Random.Next(6) == 3 && postings != null)
{
int freq = postings.Freq;
Assert.IsTrue(freq >= 1 && freq <= 4, "got invalid freq=" + freq);
}
}
}
}
r.Dispose();
dir.Dispose();
}
public virtual void TestInvalidValues()
{
IndexWriterConfig conf = new IndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
// Test IndexDeletionPolicy
Assert.AreEqual(typeof(KeepOnlyLastCommitDeletionPolicy), conf.IndexDeletionPolicy.GetType());
conf.SetIndexDeletionPolicy(new SnapshotDeletionPolicy(null));
Assert.AreEqual(typeof(SnapshotDeletionPolicy), conf.IndexDeletionPolicy.GetType());
try
{
conf.SetIndexDeletionPolicy(null);
Assert.Fail();
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// ok
}
// Test MergeScheduler
#if !FEATURE_CONCURRENTMERGESCHEDULER
Assert.AreEqual(typeof(TaskMergeScheduler), conf.MergeScheduler.GetType());
#else
Assert.AreEqual(typeof(ConcurrentMergeScheduler), conf.MergeScheduler.GetType());
#endif
conf.SetMergeScheduler(new SerialMergeScheduler());
Assert.AreEqual(typeof(SerialMergeScheduler), conf.MergeScheduler.GetType());
try
{
conf.SetMergeScheduler(null);
Assert.Fail();
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// ok
}
// Test Similarity:
// we shouldnt assert what the default is, just that its not null.
Assert.IsTrue(IndexSearcher.DefaultSimilarity == conf.Similarity);
conf.SetSimilarity(new MySimilarity());
Assert.AreEqual(typeof(MySimilarity), conf.Similarity.GetType());
try
{
conf.SetSimilarity(null);
Assert.Fail();
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// ok
}
// Test IndexingChain
Assert.IsTrue(DocumentsWriterPerThread.DefaultIndexingChain == conf.IndexingChain);
conf.SetIndexingChain(new MyIndexingChain());
Assert.AreEqual(typeof(MyIndexingChain), conf.IndexingChain.GetType());
try
{
conf.SetIndexingChain(null);
Assert.Fail();
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// ok
}
try
{
conf.SetMaxBufferedDeleteTerms(0);
Assert.Fail("should not have succeeded to set maxBufferedDeleteTerms to 0");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
try
{
conf.SetMaxBufferedDocs(1);
Assert.Fail("should not have succeeded to set maxBufferedDocs to 1");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
try
{
// Disable both MAX_BUF_DOCS and RAM_SIZE_MB
conf.SetMaxBufferedDocs(4);
conf.SetRAMBufferSizeMB(IndexWriterConfig.DISABLE_AUTO_FLUSH);
conf.SetMaxBufferedDocs(IndexWriterConfig.DISABLE_AUTO_FLUSH);
Assert.Fail("should not have succeeded to disable maxBufferedDocs when ramBufferSizeMB is disabled as well");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
conf.SetRAMBufferSizeMB(IndexWriterConfig.DEFAULT_RAM_BUFFER_SIZE_MB);
conf.SetMaxBufferedDocs(IndexWriterConfig.DEFAULT_MAX_BUFFERED_DOCS);
try
{
conf.SetRAMBufferSizeMB(IndexWriterConfig.DISABLE_AUTO_FLUSH);
Assert.Fail("should not have succeeded to disable ramBufferSizeMB when maxBufferedDocs is disabled as well");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
// Test setReaderTermsIndexDivisor
try
{
conf.SetReaderTermsIndexDivisor(0);
Assert.Fail("should not have succeeded to set termsIndexDivisor to 0");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
// Setting to -1 is ok
conf.SetReaderTermsIndexDivisor(-1);
try
{
conf.SetReaderTermsIndexDivisor(-2);
Assert.Fail("should not have succeeded to set termsIndexDivisor to < -1");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
try
{
conf.SetRAMPerThreadHardLimitMB(2048);
Assert.Fail("should not have succeeded to set RAMPerThreadHardLimitMB to >= 2048");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
try
{
conf.SetRAMPerThreadHardLimitMB(0);
Assert.Fail("should not have succeeded to set RAMPerThreadHardLimitMB to 0");
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// this is expected
}
// Test MergePolicy
Assert.AreEqual(typeof(TieredMergePolicy), conf.MergePolicy.GetType());
conf.SetMergePolicy(new LogDocMergePolicy());
Assert.AreEqual(typeof(LogDocMergePolicy), conf.MergePolicy.GetType());
try
{
conf.SetMergePolicy(null);
Assert.Fail();
}
#pragma warning disable 168
catch (System.ArgumentException e)
#pragma warning restore 168
{
// ok
}
}
public virtual void TestLongPostings_Mem()
{
// Don't use TestUtil.getTempDir so that we own the
// randomness (ie same seed will point to same dir):
Directory dir = NewFSDirectory(CreateTempDir("longpostings" + "." + Random.NextInt64()));
int NUM_DOCS = AtLeast(2000);
if (Verbose)
{
Console.WriteLine("TEST: NUM_DOCS=" + NUM_DOCS);
}
string s1 = GetRandomTerm(null);
string s2 = GetRandomTerm(s1);
if (Verbose)
{
Console.WriteLine("\nTEST: s1=" + s1 + " s2=" + s2);
/*
* for(int idx=0;idx<s1.length();idx++) {
* System.out.println(" s1 ch=0x" + Integer.toHexString(s1.charAt(idx)));
* }
* for(int idx=0;idx<s2.length();idx++) {
* System.out.println(" s2 ch=0x" + Integer.toHexString(s2.charAt(idx)));
* }
*/
}
FixedBitSet isS1 = new FixedBitSet(NUM_DOCS);
for (int idx = 0; idx < NUM_DOCS; idx++)
{
if (Random.NextBoolean())
{
isS1.Set(idx);
}
}
IndexReader r;
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random)).SetOpenMode(OpenMode.CREATE).SetMergePolicy(NewLogMergePolicy());
iwc.SetRAMBufferSizeMB(16.0 + 16.0 * Random.NextDouble());
iwc.SetMaxBufferedDocs(-1);
RandomIndexWriter riw = new RandomIndexWriter(Random, dir, iwc);
for (int idx = 0; idx < NUM_DOCS; idx++)
{
Document doc = new Document();
string s = isS1.Get(idx) ? s1 : s2;
Field f = NewTextField("field", s, Field.Store.NO);
int count = TestUtil.NextInt32(Random, 1, 4);
for (int ct = 0; ct < count; ct++)
{
doc.Add(f);
}
riw.AddDocument(doc);
}
r = riw.GetReader();
riw.Dispose();
/*
* if (VERBOSE) {
* System.out.println("TEST: terms");
* TermEnum termEnum = r.Terms();
* while(termEnum.Next()) {
* System.out.println(" term=" + termEnum.Term() + " len=" + termEnum.Term().Text().length());
* Assert.IsTrue(termEnum.DocFreq() > 0);
* System.out.println(" s1?=" + (termEnum.Term().Text().equals(s1)) + " s1len=" + s1.length());
* System.out.println(" s2?=" + (termEnum.Term().Text().equals(s2)) + " s2len=" + s2.length());
* final String s = termEnum.Term().Text();
* for(int idx=0;idx<s.length();idx++) {
* System.out.println(" ch=0x" + Integer.toHexString(s.charAt(idx)));
* }
* }
* }
*/
Assert.AreEqual(NUM_DOCS, r.NumDocs);
Assert.IsTrue(r.DocFreq(new Term("field", s1)) > 0);
Assert.IsTrue(r.DocFreq(new Term("field", s2)) > 0);
int num = AtLeast(1000);
for (int iter = 0; iter < num; iter++)
{
string term;
bool doS1;
if (Random.NextBoolean())
{
term = s1;
doS1 = true;
}
else
{
term = s2;
doS1 = false;
}
if (Verbose)
{
Console.WriteLine("\nTEST: iter=" + iter + " doS1=" + doS1);
}
DocsAndPositionsEnum postings = MultiFields.GetTermPositionsEnum(r, null, "field", new BytesRef(term));
int docID = -1;
while (docID < DocIdSetIterator.NO_MORE_DOCS)
{
int what = Random.Next(3);
if (what == 0)
{
if (Verbose)
{
Console.WriteLine("TEST: docID=" + docID + "; do next()");
}
// nextDoc
int expected = docID + 1;
while (true)
{
if (expected == NUM_DOCS)
{
expected = int.MaxValue;
break;
}
else if (isS1.Get(expected) == doS1)
{
break;
}
else
{
expected++;
}
}
docID = postings.NextDoc();
if (Verbose)
{
Console.WriteLine(" got docID=" + docID);
}
Assert.AreEqual(expected, docID);
if (docID == DocIdSetIterator.NO_MORE_DOCS)
{
break;
}
if (Random.Next(6) == 3)
{
int freq = postings.Freq;
Assert.IsTrue(freq >= 1 && freq <= 4);
for (int pos = 0; pos < freq; pos++)
{
Assert.AreEqual(pos, postings.NextPosition());
if (Random.NextBoolean())
{
var dummy = postings.GetPayload();
if (Random.NextBoolean())
{
dummy = postings.GetPayload(); // get it again
}
}
}
}
}
else
{
// advance
int targetDocID;
if (docID == -1)
{
targetDocID = Random.Next(NUM_DOCS + 1);
}
else
{
targetDocID = docID + TestUtil.NextInt32(Random, 1, NUM_DOCS - docID);
}
if (Verbose)
{
Console.WriteLine("TEST: docID=" + docID + "; do advance(" + targetDocID + ")");
}
int expected = targetDocID;
while (true)
{
if (expected == NUM_DOCS)
{
expected = int.MaxValue;
break;
}
else if (isS1.Get(expected) == doS1)
{
break;
}
else
{
expected++;
}
}
docID = postings.Advance(targetDocID);
if (Verbose)
{
Console.WriteLine(" got docID=" + docID);
}
Assert.AreEqual(expected, docID);
if (docID == DocIdSetIterator.NO_MORE_DOCS)
{
break;
}
if (Random.Next(6) == 3)
{
int freq = postings.Freq;
Assert.IsTrue(freq >= 1 && freq <= 4);
for (int pos = 0; pos < freq; pos++)
{
Assert.AreEqual(pos, postings.NextPosition());
if (Random.NextBoolean())
{
var dummy = postings.GetPayload();
if (Random.NextBoolean())
{
dummy = postings.GetPayload(); // get it again
}
}
}
}
}
}
}
r.Dispose();
dir.Dispose();
}
public virtual void TestDeletes1()
{
//IndexWriter.debug2 = System.out;
Directory dir = new MockDirectoryWrapper(new Random(Random().Next()), new RAMDirectory());
IndexWriterConfig iwc = new IndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
iwc.SetMergeScheduler(new SerialMergeScheduler());
iwc.SetMaxBufferedDocs(5000);
iwc.SetRAMBufferSizeMB(100);
RangeMergePolicy fsmp = new RangeMergePolicy(this, false);
iwc.SetMergePolicy(fsmp);
IndexWriter writer = new IndexWriter(dir, iwc);
for (int x = 0; x < 5; x++)
{
writer.AddDocument(DocHelper.CreateDocument(x, "1", 2));
//System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
//System.out.println("commit1");
writer.Commit();
Assert.AreEqual(1, writer.SegmentCount);
for (int x = 5; x < 10; x++)
{
writer.AddDocument(DocHelper.CreateDocument(x, "2", 2));
//System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
//System.out.println("commit2");
writer.Commit();
Assert.AreEqual(2, writer.SegmentCount);
for (int x = 10; x < 15; x++)
{
writer.AddDocument(DocHelper.CreateDocument(x, "3", 2));
//System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
writer.DeleteDocuments(new Term("id", "1"));
writer.DeleteDocuments(new Term("id", "11"));
// flushing without applying deletes means
// there will still be deletes in the segment infos
writer.Flush(false, false);
Assert.IsTrue(writer.BufferedUpdatesStreamAny);
// get reader flushes pending deletes
// so there should not be anymore
IndexReader r1 = writer.Reader;
Assert.IsFalse(writer.BufferedUpdatesStreamAny);
r1.Dispose();
// delete id:2 from the first segment
// merge segments 0 and 1
// which should apply the delete id:2
writer.DeleteDocuments(new Term("id", "2"));
writer.Flush(false, false);
fsmp = (RangeMergePolicy)writer.Config.MergePolicy;
fsmp.DoMerge = true;
fsmp.Start = 0;
fsmp.Length = 2;
writer.MaybeMerge();
Assert.AreEqual(2, writer.SegmentCount);
// id:2 shouldn't exist anymore because
// it's been applied in the merge and now it's gone
IndexReader r2 = writer.Reader;
int[] id2docs = ToDocsArray(new Term("id", "2"), null, r2);
Assert.IsTrue(id2docs == null);
r2.Dispose();
/*
/// // added docs are in the ram buffer
/// for (int x = 15; x < 20; x++) {
/// writer.AddDocument(TestIndexWriterReader.CreateDocument(x, "4", 2));
/// System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
/// }
/// Assert.IsTrue(writer.numRamDocs() > 0);
/// // delete from the ram buffer
/// writer.DeleteDocuments(new Term("id", Integer.toString(13)));
///
/// Term id3 = new Term("id", Integer.toString(3));
///
/// // delete from the 1st segment
/// writer.DeleteDocuments(id3);
///
/// Assert.IsTrue(writer.numRamDocs() > 0);
///
/// //System.out
/// // .println("segdels1:" + writer.docWriter.deletesToString());
///
/// //Assert.IsTrue(writer.docWriter.segmentDeletes.Size() > 0);
///
/// // we cause a merge to happen
/// fsmp.doMerge = true;
/// fsmp.start = 0;
/// fsmp.Length = 2;
/// System.out.println("maybeMerge "+writer.SegmentInfos);
///
/// SegmentInfo info0 = writer.SegmentInfos.Info(0);
/// SegmentInfo info1 = writer.SegmentInfos.Info(1);
///
/// writer.MaybeMerge();
/// System.out.println("maybeMerge after "+writer.SegmentInfos);
/// // there should be docs in RAM
/// Assert.IsTrue(writer.numRamDocs() > 0);
///
/// // assert we've merged the 1 and 2 segments
/// // and still have a segment leftover == 2
/// Assert.AreEqual(2, writer.SegmentInfos.Size());
/// Assert.IsFalse(segThere(info0, writer.SegmentInfos));
/// Assert.IsFalse(segThere(info1, writer.SegmentInfos));
///
/// //System.out.println("segdels2:" + writer.docWriter.deletesToString());
///
/// //Assert.IsTrue(writer.docWriter.segmentDeletes.Size() > 0);
///
/// IndexReader r = writer.GetReader();
/// IndexReader r1 = r.getSequentialSubReaders()[0];
/// printDelDocs(r1.GetLiveDocs());
/// int[] docs = toDocsArray(id3, null, r);
/// System.out.println("id3 docs:"+Arrays.toString(docs));
/// // there shouldn't be any docs for id:3
/// Assert.IsTrue(docs == null);
/// r.Dispose();
///
/// part2(writer, fsmp);
///
*/
// System.out.println("segdels2:"+writer.docWriter.segmentDeletes.toString());
//System.out.println("close");
writer.Dispose();
dir.Dispose();
}