private int CopyVectorsNoDeletions(MergeState mergeState, Lucene40TermVectorsReader matchingVectorsReader, AtomicReader reader, int[] rawDocLengths, int[] rawDocLengths2)
{
int maxDoc = reader.MaxDoc;
if (matchingVectorsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
int docCount = 0;
while (docCount < maxDoc)
{
int len = Math.Min(MAX_RAW_MERGE_DOCS, maxDoc - docCount);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, docCount, len);
AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, len);
docCount += len;
mergeState.CheckAbort.Work(300 * len);
}
}
else
{
for (int docNum = 0; docNum < maxDoc; docNum++)
{
// NOTE: it's very important to first assign to vectors then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
Fields vectors = reader.GetTermVectors(docNum);
AddAllDocVectors(vectors, mergeState);
mergeState.CheckAbort.Work(300);
}
}
return(maxDoc);
}
public TVTerms(Lucene40TermVectorsReader outerInstance, long tvfFP)
{
this.outerInstance = outerInstance;
outerInstance.tvf.Seek(tvfFP);
numTerms = outerInstance.tvf.ReadVInt32();
byte bits = outerInstance.tvf.ReadByte();
storePositions = (bits & STORE_POSITIONS_WITH_TERMVECTOR) != 0;
storeOffsets = (bits & STORE_OFFSET_WITH_TERMVECTOR) != 0;
storePayloads = (bits & STORE_PAYLOAD_WITH_TERMVECTOR) != 0;
tvfFPStart = outerInstance.tvf.Position; // LUCENENET specific: Renamed from getFilePointer() to match FileStream
}
public TVTerms(Lucene40TermVectorsReader outerInstance, long tvfFP)
{
this.outerInstance = outerInstance;
outerInstance.tvf.Seek(tvfFP);
numTerms = outerInstance.tvf.ReadVInt32();
byte bits = outerInstance.tvf.ReadByte();
storePositions = (bits & STORE_POSITIONS_WITH_TERMVECTOR) != 0;
storeOffsets = (bits & STORE_OFFSET_WITH_TERMVECTOR) != 0;
storePayloads = (bits & STORE_PAYLOAD_WITH_TERMVECTOR) != 0;
tvfFPStart = outerInstance.tvf.GetFilePointer();
}
/// <summary>
/// Do a bulk copy of numDocs documents from reader to our
/// streams. This is used to expedite merging, if the
/// field numbers are congruent.
/// </summary>
private void AddRawDocuments(Lucene40TermVectorsReader reader, int[] tvdLengths, int[] tvfLengths, int numDocs)
{
long tvdPosition = tvd.GetFilePointer();
long tvfPosition = tvf.GetFilePointer();
long tvdStart = tvdPosition;
long tvfStart = tvfPosition;
for (int i = 0; i < numDocs; i++)
{
tvx.WriteInt64(tvdPosition);
tvdPosition += tvdLengths[i];
tvx.WriteInt64(tvfPosition);
tvfPosition += tvfLengths[i];
}
tvd.CopyBytes(reader.TvdStream, tvdPosition - tvdStart);
tvf.CopyBytes(reader.TvfStream, tvfPosition - tvfStart);
Debug.Assert(tvd.GetFilePointer() == tvdPosition);
Debug.Assert(tvf.GetFilePointer() == tvfPosition);
}
public TVFields(Lucene40TermVectorsReader outerInstance, int docID)
{
this.outerInstance = outerInstance;
outerInstance.SeekTvx(docID);
outerInstance.tvd.Seek(outerInstance.tvx.ReadInt64());
int fieldCount = outerInstance.tvd.ReadVInt32();
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fieldCount >= 0);
}
if (fieldCount != 0)
{
fieldNumbers = new int[fieldCount];
fieldFPs = new long[fieldCount];
for (int fieldUpto = 0; fieldUpto < fieldCount; fieldUpto++)
{
int fieldNumber = outerInstance.tvd.ReadVInt32();
fieldNumbers[fieldUpto] = fieldNumber;
fieldNumberToIndex[fieldNumber] = fieldUpto;
}
long position = outerInstance.tvx.ReadInt64();
fieldFPs[0] = position;
for (int fieldUpto = 1; fieldUpto < fieldCount; fieldUpto++)
{
position += outerInstance.tvd.ReadVInt64();
fieldFPs[fieldUpto] = position;
}
}
else
{
// TODO: we can improve writer here, eg write 0 into
// tvx file, so we know on first read from tvx that
// this doc has no TVs
fieldNumbers = null;
fieldFPs = null;
}
}
/// <summary>
/// Do a bulk copy of numDocs documents from reader to our
/// streams. This is used to expedite merging, if the
/// field numbers are congruent.
/// </summary>
private void AddRawDocuments(Lucene40TermVectorsReader reader, int[] tvdLengths, int[] tvfLengths, int numDocs)
{
long tvdPosition = tvd.Position; // LUCENENET specific: Renamed from getFilePointer() to match FileStream
long tvfPosition = tvf.Position; // LUCENENET specific: Renamed from getFilePointer() to match FileStream
long tvdStart = tvdPosition;
long tvfStart = tvfPosition;
for (int i = 0; i < numDocs; i++)
{
tvx.WriteInt64(tvdPosition);
tvdPosition += tvdLengths[i];
tvx.WriteInt64(tvfPosition);
tvfPosition += tvfLengths[i];
}
tvd.CopyBytes(reader.TvdStream, tvdPosition - tvdStart);
tvf.CopyBytes(reader.TvfStream, tvfPosition - tvfStart);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(tvd.Position == tvdPosition); // LUCENENET specific: Renamed from getFilePointer() to match FileStream
Debugging.Assert(tvf.Position == tvfPosition); // LUCENENET specific: Renamed from getFilePointer() to match FileStream
}
}
public override int Merge(MergeState mergeState)
{
// Used for bulk-reading raw bytes for term vectors
int[] rawDocLengths = new int[MAX_RAW_MERGE_DOCS];
int[] rawDocLengths2 = new int[MAX_RAW_MERGE_DOCS];
int idx = 0;
int numDocs = 0;
for (int i = 0; i < mergeState.Readers.Count; i++)
{
AtomicReader reader = mergeState.Readers[i];
SegmentReader matchingSegmentReader = mergeState.MatchingSegmentReaders[idx++];
Lucene40TermVectorsReader matchingVectorsReader = null;
if (matchingSegmentReader != null)
{
TermVectorsReader vectorsReader = matchingSegmentReader.TermVectorsReader;
if (vectorsReader != null && vectorsReader is Lucene40TermVectorsReader)
{
matchingVectorsReader = (Lucene40TermVectorsReader)vectorsReader;
}
}
if (reader.LiveDocs != null)
{
numDocs += CopyVectorsWithDeletions(mergeState, matchingVectorsReader, reader, rawDocLengths, rawDocLengths2);
}
else
{
numDocs += CopyVectorsNoDeletions(mergeState, matchingVectorsReader, reader, rawDocLengths, rawDocLengths2);
}
}
Finish(mergeState.FieldInfos, numDocs);
return(numDocs);
}
// NOTE: tvf is pre-positioned by caller
public TVTermsEnum(Lucene40TermVectorsReader outerInstance)
{
this.outerInstance = outerInstance;
this.origTVF = outerInstance.tvf;
tvf = (IndexInput)origTVF.Clone();
}
private int CopyVectorsWithDeletions(MergeState mergeState, Lucene40TermVectorsReader matchingVectorsReader, AtomicReader reader, int[] rawDocLengths, int[] rawDocLengths2)
{
int maxDoc = reader.MaxDoc;
IBits liveDocs = reader.LiveDocs;
int totalNumDocs = 0;
if (matchingVectorsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
for (int docNum = 0; docNum < maxDoc;)
{
if (!liveDocs.Get(docNum))
{
// skip deleted docs
++docNum;
continue;
}
// We can optimize this case (doing a bulk byte copy) since the field
// numbers are identical
int start = docNum, numDocs = 0;
do
{
docNum++;
numDocs++;
if (docNum >= maxDoc)
{
break;
}
if (!liveDocs.Get(docNum))
{
docNum++;
break;
}
} while (numDocs < MAX_RAW_MERGE_DOCS);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, start, numDocs);
AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, numDocs);
totalNumDocs += numDocs;
mergeState.CheckAbort.Work(300 * numDocs);
}
}
else
{
for (int docNum = 0; docNum < maxDoc; docNum++)
{
if (!liveDocs.Get(docNum))
{
// skip deleted docs
continue;
}
// NOTE: it's very important to first assign to vectors then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
Fields vectors = reader.GetTermVectors(docNum);
AddAllDocVectors(vectors, mergeState);
totalNumDocs++;
mergeState.CheckAbort.Work(300);
}
}
return(totalNumDocs);
}
