private void CopyVectorsNoDeletions(TermVectorsWriter termVectorsWriter, TermVectorsReader matchingVectorsReader, IndexReader reader)
{
int maxDoc = reader.MaxDoc;
if (matchingVectorsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
int docCount = 0;
while (docCount < maxDoc)
{
int len = System.Math.Min(MAX_RAW_MERGE_DOCS, maxDoc - docCount);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, docCount, len);
termVectorsWriter.AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, len);
docCount += len;
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
ITermFreqVector[] vectors = reader.GetTermFreqVectors(docNum);
termVectorsWriter.AddAllDocVectors(vectors);
checkAbort.Work(300);
}
}
}
private void CopyVectorsWithDeletions(TermVectorsWriter termVectorsWriter, TermVectorsReader matchingVectorsReader, IndexReader reader)
{
int maxDoc = reader.MaxDoc;
if (matchingVectorsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
for (int docNum = 0; docNum < maxDoc;)
{
if (reader.IsDeleted(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 (reader.IsDeleted(docNum))
{
docNum++;
break;
}
}while (numDocs < MAX_RAW_MERGE_DOCS);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, start, numDocs);
termVectorsWriter.AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, numDocs);
checkAbort.Work(300 * numDocs);
}
}
else
{
for (int docNum = 0; docNum < maxDoc; docNum++)
{
if (reader.IsDeleted(docNum))
{
// skip deleted docs
continue;
}
// NOTE: it's very important to first assign to vectors then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
ITermFreqVector[] vectors = reader.GetTermFreqVectors(docNum);
termVectorsWriter.AddAllDocVectors(vectors);
checkAbort.Work(300);
}
}
}
private void CopyVectorsNoDeletions(TermVectorsWriter termVectorsWriter, TermVectorsReader matchingVectorsReader, IndexReader reader)
{
int maxDoc = reader.MaxDoc();
if (matchingVectorsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
int docCount = 0;
while (docCount < maxDoc)
{
int len = System.Math.Min(MAX_RAW_MERGE_DOCS, maxDoc - docCount);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, docCount, len);
termVectorsWriter.AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, len);
docCount += len;
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
TermFreqVector[] vectors = reader.GetTermFreqVectors(docNum);
termVectorsWriter.AddAllDocVectors(vectors);
checkAbort.Work(300);
}
}
}
private void CopyVectorsWithDeletions(TermVectorsWriter termVectorsWriter, TermVectorsReader matchingVectorsReader, IndexReader reader)
{
int maxDoc = reader.MaxDoc();
if (matchingVectorsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
for (int docNum = 0; docNum < maxDoc; )
{
if (reader.IsDeleted(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 (reader.IsDeleted(docNum))
{
docNum++;
break;
}
}
while (numDocs < MAX_RAW_MERGE_DOCS);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, start, numDocs);
termVectorsWriter.AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, numDocs);
checkAbort.Work(300 * numDocs);
}
}
else
{
for (int docNum = 0; docNum < maxDoc; docNum++)
{
if (reader.IsDeleted(docNum))
{
// skip deleted docs
continue;
}
// NOTE: it's very important to first assign to vectors then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
TermFreqVector[] vectors = reader.GetTermFreqVectors(docNum);
termVectorsWriter.AddAllDocVectors(vectors);
checkAbort.Work(300);
}
}
}
/// <summary> Merge the TermVectors from each of the segments into the new one.</summary>
/// <throws> IOException </throws>
private void MergeVectors()
{
TermVectorsWriter termVectorsWriter =
new TermVectorsWriter(directory, segment, fieldInfos);
try
{
for (int r = 0; r < readers.Count; r++)
{
SegmentReader matchingSegmentReader = matchingSegmentReaders[r];
TermVectorsReader matchingVectorsReader;
bool hasMatchingReader;
if (matchingSegmentReader != null)
{
matchingVectorsReader = matchingSegmentReader.termVectorsReaderOrig;
// If the TV* files are an older format then they
// cannot read raw docs:
if (matchingVectorsReader != null && !matchingVectorsReader.CanReadRawDocs())
{
matchingVectorsReader = null;
hasMatchingReader = false;
}
else
hasMatchingReader = matchingVectorsReader != null;
}
else
{
hasMatchingReader = false;
matchingVectorsReader = null;
}
IndexReader reader = (IndexReader)readers[r];
bool hasDeletions = reader.HasDeletions();
int maxDoc = reader.MaxDoc();
for (int docNum = 0; docNum < maxDoc; )
{
// skip deleted docs
if (!hasDeletions || !reader.IsDeleted(docNum))
{
if (hasMatchingReader)
{
// We can optimize this case (doing a bulk
// byte copy) since the field numbers are
// identical
int start = docNum;
int numDocs = 0;
do
{
docNum++;
numDocs++;
if (docNum >= maxDoc)
break;
if (hasDeletions && matchingSegmentReader.IsDeleted(docNum))
{
docNum++;
break;
}
} while (numDocs < MAX_RAW_MERGE_DOCS);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, start, numDocs);
termVectorsWriter.AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, numDocs);
if (checkAbort != null)
checkAbort.Work(300 * numDocs);
}
else
{
// NOTE: it's very important to first assign
// to vectors then pass it to
// termVectorsWriter.addAllDocVectors; see
// LUCENE-1282
TermFreqVector[] vectors = reader.GetTermFreqVectors(docNum);
termVectorsWriter.AddAllDocVectors(vectors);
docNum++;
if (checkAbort != null)
checkAbort.Work(300);
}
}
else
docNum++;
}
}
}
finally
{
termVectorsWriter.Close();
}
long tvxSize = directory.FileLength(segment + "." + IndexFileNames.VECTORS_INDEX_EXTENSION);
// {{dougsale-2.4.0}
// this shouldn't be a problem for us - if it is,
// then it's not a JRE bug
//if (4 + mergedDocs * 16 != tvxSize)
// // This is most likely a bug in Sun JRE 1.6.0_04/_05;
// // we detect that the bug has struck, here, and
// // throw an exception to prevent the corruption from
// // entering the index. See LUCENE-1282 for
// // details.
// throw new RuntimeException("mergeVectors produced an invalid result: mergedDocs is " + mergedDocs + " but tvx size is " + tvxSize + "; now aborting this merge to prevent index corruption");
}
/// <summary> Merge the TermVectors from each of the segments into the new one.</summary>
/// <throws> IOException </throws>
private void MergeVectors()
{
TermVectorsWriter termVectorsWriter =
new TermVectorsWriter(directory, segment, fieldInfos);
try
{
for (int r = 0; r < readers.Count; r++)
{
SegmentReader matchingSegmentReader = matchingSegmentReaders[r];
TermVectorsReader matchingVectorsReader;
bool hasMatchingReader;
if (matchingSegmentReader != null)
{
matchingVectorsReader = matchingSegmentReader.termVectorsReaderOrig;
// If the TV* files are an older format then they
// cannot read raw docs:
if (matchingVectorsReader != null && !matchingVectorsReader.CanReadRawDocs())
{
matchingVectorsReader = null;
hasMatchingReader = false;
}
else
{
hasMatchingReader = matchingVectorsReader != null;
}
}
else
{
hasMatchingReader = false;
matchingVectorsReader = null;
}
IndexReader reader = (IndexReader)readers[r];
bool hasDeletions = reader.HasDeletions();
int maxDoc = reader.MaxDoc();
for (int docNum = 0; docNum < maxDoc;)
{
// skip deleted docs
if (!hasDeletions || !reader.IsDeleted(docNum))
{
if (hasMatchingReader)
{
// We can optimize this case (doing a bulk
// byte copy) since the field numbers are
// identical
int start = docNum;
int numDocs = 0;
do
{
docNum++;
numDocs++;
if (docNum >= maxDoc)
{
break;
}
if (hasDeletions && matchingSegmentReader.IsDeleted(docNum))
{
docNum++;
break;
}
} while (numDocs < MAX_RAW_MERGE_DOCS);
matchingVectorsReader.RawDocs(rawDocLengths, rawDocLengths2, start, numDocs);
termVectorsWriter.AddRawDocuments(matchingVectorsReader, rawDocLengths, rawDocLengths2, numDocs);
if (checkAbort != null)
{
checkAbort.Work(300 * numDocs);
}
}
else
{
// NOTE: it's very important to first assign
// to vectors then pass it to
// termVectorsWriter.addAllDocVectors; see
// LUCENE-1282
TermFreqVector[] vectors = reader.GetTermFreqVectors(docNum);
termVectorsWriter.AddAllDocVectors(vectors);
docNum++;
if (checkAbort != null)
{
checkAbort.Work(300);
}
}
}
else
{
docNum++;
}
}
}
}
finally
{
termVectorsWriter.Close();
}
long tvxSize = directory.FileLength(segment + "." + IndexFileNames.VECTORS_INDEX_EXTENSION);
// {{dougsale-2.4.0}
// this shouldn't be a problem for us - if it is,
// then it's not a JRE bug
//if (4 + mergedDocs * 16 != tvxSize)
// // This is most likely a bug in Sun JRE 1.6.0_04/_05;
// // we detect that the bug has struck, here, and
// // throw an exception to prevent the corruption from
// // entering the index. See LUCENE-1282 for
// // details.
// throw new RuntimeException("mergeVectors produced an invalid result: mergedDocs is " + mergedDocs + " but tvx size is " + tvxSize + "; now aborting this merge to prevent index corruption");
}