public void CloseDocStore(SegmentWriteState state)
{
lock (this)
{
int inc = state.numDocsInStore - lastDocID;
if (inc > 0)
{
InitFieldsWriter();
Fill(state.numDocsInStore - docWriter.GetDocStoreOffset());
}
if (fieldsWriter != null)
{
fieldsWriter.Close();
fieldsWriter = null;
lastDocID = 0;
System.Diagnostics.Debug.Assert(state.docStoreSegmentName != null);
SupportClass.CollectionsHelper.AddIfNotContains(state.flushedFiles, state.docStoreSegmentName + "." + IndexFileNames.FIELDS_EXTENSION);
SupportClass.CollectionsHelper.AddIfNotContains(state.flushedFiles, state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
state.docWriter.RemoveOpenFile(state.docStoreSegmentName + "." + IndexFileNames.FIELDS_EXTENSION);
state.docWriter.RemoveOpenFile(state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
System.String fileName = state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION;
if (4 + ((long)state.numDocsInStore) * 8 != state.directory.FileLength(fileName))
{
throw new System.SystemException("after flush: fdx size mismatch: " + state.numDocsInStore + " docs vs " + state.directory.FileLength(fileName) + " length in bytes of " + fileName + " file exists?=" + state.directory.FileExists(fileName));
}
}
}
}
internal override void closeDocStore(DocumentsWriter.FlushState state)
{
lock (this)
{
int inc = state.numDocsInStore - lastDocID;
if (inc > 0)
{
initFieldsWriter();
fill(state.numDocsInStore - docWriter.GetDocStoreOffset());
}
if (fieldsWriter != null)
{
fieldsWriter.Close();
fieldsWriter = null;
lastDocID = 0;
System.Diagnostics.Debug.Assert(state.docStoreSegmentName != null);
string fdtFile = state.docStoreSegmentName + "." + IndexFileNames.FIELDS_EXTENSION;
string fdxFile = state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION;
state.flushedFiles[fdtFile] = fdtFile;
state.flushedFiles[fdxFile] = fdxFile;
state.docWriter.RemoveOpenFile(fdtFile);
state.docWriter.RemoveOpenFile(fdxFile);
if (4 + state.numDocsInStore * 8 != state.directory.FileLength(fdxFile))
{
throw new System.SystemException("after flush: fdx size mismatch: " + state.numDocsInStore + " docs vs " + state.directory.FileLength(fdxFile) + " length in bytes of " + fdxFile);
}
}
}
}
public void CloseDocStore(SegmentWriteState state)
{
lock (this)
{
int inc = state.numDocsInStore - lastDocID;
if (inc > 0)
{
InitFieldsWriter();
Fill(state.numDocsInStore - docWriter.GetDocStoreOffset());
}
if (fieldsWriter != null)
{
fieldsWriter.Close();
fieldsWriter = null;
lastDocID = 0;
System.Diagnostics.Debug.Assert(state.docStoreSegmentName != null);
SupportClass.CollectionsHelper.AddIfNotContains(state.flushedFiles, state.docStoreSegmentName + "." + IndexFileNames.FIELDS_EXTENSION);
SupportClass.CollectionsHelper.AddIfNotContains(state.flushedFiles, state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
state.docWriter.RemoveOpenFile(state.docStoreSegmentName + "." + IndexFileNames.FIELDS_EXTENSION);
state.docWriter.RemoveOpenFile(state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
System.String fileName = state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION;
if (4 + ((long) state.numDocsInStore) * 8 != state.directory.FileLength(fileName))
throw new System.SystemException("after flush: fdx size mismatch: " + state.numDocsInStore + " docs vs " + state.directory.FileLength(fileName) + " length in bytes of " + fileName + " file exists?=" + state.directory.FileExists(fileName));
}
}
}
private int CopyFieldsNoDeletions(FieldsWriter fieldsWriter, IndexReader reader, FieldsReader matchingFieldsReader)
{
int maxDoc = reader.MaxDoc;
int docCount = 0;
if (matchingFieldsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
while (docCount < maxDoc)
{
int len = System.Math.Min(MAX_RAW_MERGE_DOCS, maxDoc - docCount);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, docCount, len);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, len);
docCount += len;
checkAbort.Work(300 * len);
}
}
else
{
for (; docCount < maxDoc; docCount++)
{
// NOTE: it's very important to first assign to doc then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
Document doc = reader.Document(docCount);
fieldsWriter.AddDocument(doc);
checkAbort.Work(300);
}
}
return(docCount);
}
private int CopyFieldsWithDeletions(FieldsWriter fieldsWriter, IndexReader reader, FieldsReader matchingFieldsReader)
{
int docCount = 0;
int maxDoc = reader.MaxDoc;
if (matchingFieldsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
for (int j = 0; j < maxDoc;)
{
if (reader.IsDeleted(j))
{
// skip deleted docs
++j;
continue;
}
// We can optimize this case (doing a bulk byte copy) since the field
// numbers are identical
int start = j, numDocs = 0;
do
{
j++;
numDocs++;
if (j >= maxDoc)
{
break;
}
if (reader.IsDeleted(j))
{
j++;
break;
}
}while (numDocs < MAX_RAW_MERGE_DOCS);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, start, numDocs);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, numDocs);
docCount += numDocs;
checkAbort.Work(300 * numDocs);
}
}
else
{
for (int j = 0; j < maxDoc; j++)
{
if (reader.IsDeleted(j))
{
// skip deleted docs
continue;
}
// NOTE: it's very important to first assign to doc then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
Document doc = reader.Document(j);
fieldsWriter.AddDocument(doc);
docCount++;
checkAbort.Work(300);
}
}
return(docCount);
}
public /*internal*/ void AddDocument(System.String segment, Document doc)
{
// write field names
fieldInfos = new FieldInfos();
fieldInfos.Add(doc);
fieldInfos.Write(directory, segment + ".fnm");
// write field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
fieldsWriter.AddDocument(doc);
}
finally
{
fieldsWriter.Close();
}
// invert doc into postingTable
postingTable.Clear(); // clear postingTable
fieldLengths = new int[fieldInfos.Size()]; // init fieldLengths
fieldPositions = new int[fieldInfos.Size()]; // init fieldPositions
fieldOffsets = new int[fieldInfos.Size()]; // init fieldOffsets
fieldBoosts = new float[fieldInfos.Size()]; // init fieldBoosts
float boost = doc.GetBoost();
for (int i = 0; i < fieldBoosts.Length; i++)
{
fieldBoosts[i] = boost;
}
InvertDocument(doc);
// sort postingTable into an array
Posting[] postings = SortPostingTable();
/*
* for (int i = 0; i < postings.length; i++) {
* Posting posting = postings[i];
* System.out.print(posting.term);
* System.out.print(" freq=" + posting.freq);
* System.out.print(" pos=");
* System.out.print(posting.positions[0]);
* for (int j = 1; j < posting.freq; j++)
* System.out.print("," + posting.positions[j]);
* System.out.println("");
* }
*/
// write postings
WritePostings(postings, segment);
// write norms of indexed fields
WriteNorms(segment);
}
public void AddDocument(System.String segment, Document doc)
{
// write field names
fieldInfos = new FieldInfos();
fieldInfos.Add(doc);
fieldInfos.Write(directory, segment + ".fnm");
// write field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
fieldsWriter.AddDocument(doc);
}
finally
{
fieldsWriter.Close();
}
// invert doc into postingTable
postingTable.Clear(); // clear postingTable
fieldLengths = new int[fieldInfos.Size()]; // init fieldLengths
fieldPositions = new int[fieldInfos.Size()]; // init fieldPositions
fieldOffsets = new int[fieldInfos.Size()]; // init fieldOffsets
fieldBoosts = new float[fieldInfos.Size()]; // init fieldBoosts
float boost = doc.GetBoost();
for (int i = 0; i < fieldBoosts.Length; i++)
{
fieldBoosts[i] = boost;
}
InvertDocument(doc);
// sort postingTable into an array
Posting[] postings = SortPostingTable();
/*
for (int i = 0; i < postings.length; i++) {
Posting posting = postings[i];
System.out.print(posting.term);
System.out.print(" freq=" + posting.freq);
System.out.print(" pos=");
System.out.print(posting.positions[0]);
for (int j = 1; j < posting.freq; j++)
System.out.print("," + posting.positions[j]);
System.out.println("");
}
*/
// write postings
WritePostings(postings, segment);
// write norms of indexed fields
WriteNorms(segment);
}
private void InitFieldsWriter()
{
if (fieldsWriter == null)
{
System.String docStoreSegment = docWriter.DocStoreSegment;
if (docStoreSegment != null)
{
System.Diagnostics.Debug.Assert(docStoreSegment != null);
fieldsWriter = new FieldsWriter(docWriter.directory, docStoreSegment, fieldInfos);
docWriter.AddOpenFile(docStoreSegment + "." + IndexFileNames.FIELDS_EXTENSION);
docWriter.AddOpenFile(docStoreSegment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
lastDocID = 0;
}
}
}
private void InitFieldsWriter()
{
if (fieldsWriter == null)
{
System.String docStoreSegment = docWriter.GetDocStoreSegment();
if (docStoreSegment != null)
{
System.Diagnostics.Debug.Assert(docStoreSegment != null);
fieldsWriter = new FieldsWriter(docWriter.directory, docStoreSegment, fieldInfos);
docWriter.AddOpenFile(docStoreSegment + "." + IndexFileNames.FIELDS_EXTENSION);
docWriter.AddOpenFile(docStoreSegment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
lastDocID = 0;
}
}
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> IOException </throws>
private int MergeFields()
{
fieldInfos = new FieldInfos(); // merge field names
int docCount = 0;
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
fieldInfos.Write(directory, segment + ".fnm");
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
// for merging we don't want to compress/uncompress the data, so to tell the FieldsReader that we're
// in merge mode, we use this FieldSelector
FieldSelector fieldSelectorMerge = new AnonymousClassFieldSelector(this);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
int maxDoc = reader.MaxDoc();
for (int j = 0; j < maxDoc; j++)
{
if (!reader.IsDeleted(j))
{
// skip deleted docs
fieldsWriter.AddDocument(reader.Document(j, fieldSelectorMerge));
docCount++;
}
}
}
}
finally
{
fieldsWriter.Close();
}
return(docCount);
}
internal void Abort()
{
lock (this)
{
if (fieldsWriter != null)
{
try
{
fieldsWriter.Close();
}
catch (System.Exception t)
{
}
fieldsWriter = null;
lastDocID = 0;
}
}
}
internal override void Abort()
{
lock (this)
{
if (fieldsWriter != null)
{
try
{
fieldsWriter.Close();
}
catch (System.Exception)
{
}
fieldsWriter = null;
lastDocID = 0;
}
}
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> IOException </throws>
private int MergeFields()
{
fieldInfos = new FieldInfos(); // merge field names
int docCount = 0;
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
fieldInfos.Write(directory, segment + ".fnm");
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
int maxDoc = reader.MaxDoc();
for (int j = 0; j < maxDoc; j++)
{
if (!reader.IsDeleted(j))
{
// skip deleted docs
fieldsWriter.AddDocument(reader.Document(j));
docCount++;
}
}
}
}
finally
{
fieldsWriter.Close();
}
return(docCount);
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> CorruptIndexException if the index is corrupt </throws>
/// <throws> IOException if there is a low-level IO error </throws>
private int MergeFields()
{
if (!mergeDocStores)
{
// When we are not merging by doc stores, that means
// all segments were written as part of a single
// autoCommit=false IndexWriter session, so their field
// name -> number mapping are the same. So, we start
// with the fieldInfos of the last segment in this
// case, to keep that numbering.
SegmentReader sr = (SegmentReader)readers[readers.Count - 1];
fieldInfos = (FieldInfos)sr.fieldInfos.Clone();
}
else
{
fieldInfos = new FieldInfos(); // merge field names
}
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader)reader;
for (int j = 0; j < segmentReader.GetFieldInfos().Size(); j++)
{
FieldInfo fi = segmentReader.GetFieldInfos().FieldInfo(j);
fieldInfos.Add(fi.name, fi.isIndexed, fi.storeTermVector, fi.storePositionWithTermVector, fi.storeOffsetWithTermVector, !reader.HasNorms(fi.name), fi.storePayloads, fi.omitTf);
}
}
else
{
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.OMIT_TF), false, false, false, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.STORES_PAYLOADS), false, false, false, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
}
fieldInfos.Write(directory, segment + ".fnm");
int docCount = 0;
SetMatchingSegmentReaders();
if (mergeDocStores)
{
// for merging we don't want to compress/uncompress the data, so to tell the FieldsReader that we're
// in merge mode, we use this FieldSelector
FieldSelector fieldSelectorMerge = new AnonymousClassFieldSelector(this);
// merge field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
SegmentReader matchingSegmentReader = matchingSegmentReaders[i];
FieldsReader matchingFieldsReader;
bool hasMatchingReader;
if (matchingSegmentReader != null)
{
FieldsReader fieldsReader = matchingSegmentReader.GetFieldsReader();
if (fieldsReader != null && !fieldsReader.CanReadRawDocs())
{
matchingFieldsReader = null;
hasMatchingReader = false;
}
else
{
matchingFieldsReader = fieldsReader;
hasMatchingReader = true;
}
}
else
{
hasMatchingReader = false;
matchingFieldsReader = null;
}
int maxDoc = reader.MaxDoc();
bool hasDeletions = reader.HasDeletions();
for (int j = 0; j < maxDoc;)
{
if (!hasDeletions || !reader.IsDeleted(j))
{ // skip deleted docs
if (hasMatchingReader)
{
// We can optimize this case (doing a bulk
// byte copy) since the field numbers are
// identical
int start = j;
int numDocs = 0;
do
{
j++;
numDocs++;
if (j >= maxDoc)
{
break;
}
if (hasDeletions && matchingSegmentReader.IsDeleted(j))
{
j++;
break;
}
} while (numDocs < MAX_RAW_MERGE_DOCS);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, start, numDocs);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, numDocs);
docCount += numDocs;
if (checkAbort != null)
{
checkAbort.Work(300 * numDocs);
}
}
else
{
// NOTE: it's very important to first assign
// to doc then pass it to
// termVectorsWriter.addAllDocVectors; see
// LUCENE-1282
Document doc = reader.Document(j, fieldSelectorMerge);
fieldsWriter.AddDocument(doc);
j++;
docCount++;
if (checkAbort != null)
{
checkAbort.Work(300);
}
}
}
else
{
j++;
}
}
}
}
finally
{
fieldsWriter.Close();
}
long fdxFileLength = directory.FileLength(segment + "." + IndexFileNames.FIELDS_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+docCount*8 != fdxFileLength)
// // 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("mergeFields produced an invalid result: docCount is " + docCount + " but fdx file size is " + fdxFileLength + "; now aborting this merge to prevent index corruption");
}
else
{
// If we are skipping the doc stores, that means there
// are no deletions in any of these segments, so we
// just sum numDocs() of each segment to get total docCount
for (int i = 0; i < readers.Count; i++)
{
docCount += ((IndexReader)readers[i]).NumDocs();
}
}
return(docCount);
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> CorruptIndexException if the index is corrupt </throws>
/// <throws> IOException if there is a low-level IO error </throws>
private int MergeFields()
{
if (!mergeDocStores)
{
// When we are not merging by doc stores, that means
// all segments were written as part of a single
// autoCommit=false IndexWriter session, so their field
// name -> number mapping are the same. So, we start
// with the fieldInfos of the last segment in this
// case, to keep that numbering.
SegmentReader sr = (SegmentReader) readers[readers.Count - 1];
fieldInfos = (FieldInfos) sr.fieldInfos.Clone();
}
else
{
fieldInfos = new FieldInfos(); // merge field names
}
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader) reader;
for (int j = 0; j < segmentReader.GetFieldInfos().Size(); j++)
{
FieldInfo fi = segmentReader.GetFieldInfos().FieldInfo(j);
fieldInfos.Add(fi.name, fi.isIndexed, fi.storeTermVector, fi.storePositionWithTermVector, fi.storeOffsetWithTermVector, !reader.HasNorms(fi.name), fi.storePayloads);
}
}
else
{
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.STORES_PAYLOADS), false, false, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
}
fieldInfos.Write(directory, segment + ".fnm");
int docCount = 0;
if (mergeDocStores)
{
// If the i'th reader is a SegmentReader and has
// identical fieldName -> number mapping, then this
// array will be non-null at position i:
SegmentReader[] matchingSegmentReaders = new SegmentReader[readers.Count];
// If this reader is a SegmentReader, and all of its
// field name -> number mappings match the "merged"
// FieldInfos, then we can do a bulk copy of the
// stored fields:
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader) reader;
bool same = true;
FieldInfos segmentFieldInfos = segmentReader.GetFieldInfos();
for (int j = 0; same && j < segmentFieldInfos.Size(); j++)
same = fieldInfos.FieldName(j).Equals(segmentFieldInfos.FieldName(j));
if (same)
{
matchingSegmentReaders[i] = segmentReader;
}
}
}
// Used for bulk-reading raw bytes for stored fields
int[] rawDocLengths = new int[MAX_RAW_MERGE_DOCS];
// for merging we don't want to compress/uncompress the data, so to tell the FieldsReader that we're
// in merge mode, we use this FieldSelector
FieldSelector fieldSelectorMerge = new AnonymousClassFieldSelector(this);
// merge field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
SegmentReader matchingSegmentReader = matchingSegmentReaders[i];
FieldsReader matchingFieldsReader;
if (matchingSegmentReader != null)
matchingFieldsReader = matchingSegmentReader.GetFieldsReader();
else
matchingFieldsReader = null;
int maxDoc = reader.MaxDoc();
for (int j = 0; j < maxDoc; )
{
if (!reader.IsDeleted(j))
{
// skip deleted docs
if (matchingSegmentReader != null)
{
// We can optimize this case (doing a bulk
// byte copy) since the field numbers are
// identical
int start = j;
int numDocs = 0;
do
{
j++;
numDocs++;
}
while (j < maxDoc && !matchingSegmentReader.IsDeleted(j) && numDocs < MAX_RAW_MERGE_DOCS);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, start, numDocs);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, numDocs);
docCount += numDocs;
if (checkAbort != null)
checkAbort.Work(300 * numDocs);
}
else
{
fieldsWriter.AddDocument(reader.Document(j, fieldSelectorMerge));
j++;
docCount++;
if (checkAbort != null)
checkAbort.Work(300);
}
}
else
j++;
}
}
}
finally
{
fieldsWriter.Close();
}
System.Diagnostics.Debug.Assert(docCount*8 == directory.FileLength(segment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION),
"after MergeFields: fdx size mismatch: " + docCount + " docs vs " +
directory.FileLength(segment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION) +
" length in bytes of " + segment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
}
// If we are skipping the doc stores, that means there
// are no deletions in any of these segments, so we
// just sum numDocs() of each segment to get total docCount
else
for (int i = 0; i < readers.Count; i++)
docCount += ((IndexReader) readers[i]).NumDocs();
return docCount;
}
private int CopyFieldsWithDeletions(FieldSelector fieldSelectorMerge, FieldsWriter fieldsWriter, IndexReader reader, FieldsReader matchingFieldsReader)
{
int docCount = 0;
int maxDoc = reader.MaxDoc();
if (matchingFieldsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
for (int j = 0; j < maxDoc; )
{
if (reader.IsDeleted(j))
{
// skip deleted docs
++j;
continue;
}
// We can optimize this case (doing a bulk byte copy) since the field
// numbers are identical
int start = j, numDocs = 0;
do
{
j++;
numDocs++;
if (j >= maxDoc)
break;
if (reader.IsDeleted(j))
{
j++;
break;
}
}
while (numDocs < MAX_RAW_MERGE_DOCS);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, start, numDocs);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, numDocs);
docCount += numDocs;
checkAbort.Work(300 * numDocs);
}
}
else
{
for (int j = 0; j < maxDoc; j++)
{
if (reader.IsDeleted(j))
{
// skip deleted docs
continue;
}
// NOTE: it's very important to first assign to doc then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
Document doc = reader.Document(j, fieldSelectorMerge);
fieldsWriter.AddDocument(doc);
docCount++;
checkAbort.Work(300);
}
}
return docCount;
}
private int CopyFieldsNoDeletions(FieldSelector fieldSelectorMerge, FieldsWriter fieldsWriter, IndexReader reader, FieldsReader matchingFieldsReader)
{
int maxDoc = reader.MaxDoc();
int docCount = 0;
if (matchingFieldsReader != null)
{
// We can bulk-copy because the fieldInfos are "congruent"
while (docCount < maxDoc)
{
int len = System.Math.Min(MAX_RAW_MERGE_DOCS, maxDoc - docCount);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, docCount, len);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, len);
docCount += len;
checkAbort.Work(300 * len);
}
}
else
{
for (; docCount < maxDoc; docCount++)
{
// NOTE: it's very important to first assign to doc then pass it to
// termVectorsWriter.addAllDocVectors; see LUCENE-1282
Document doc = reader.Document(docCount, fieldSelectorMerge);
fieldsWriter.AddDocument(doc);
checkAbort.Work(300);
}
}
return docCount;
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> CorruptIndexException if the index is corrupt </throws>
/// <throws> IOException if there is a low-level IO error </throws>
private int MergeFields()
{
if (!mergeDocStores)
{
// When we are not merging by doc stores, that means
// all segments were written as part of a single
// autoCommit=false IndexWriter session, so their field
// name -> number mapping are the same. So, we start
// with the fieldInfos of the last segment in this
// case, to keep that numbering.
SegmentReader sr = (SegmentReader) readers[readers.Count - 1];
fieldInfos = (FieldInfos) sr.fieldInfos.Clone();
}
else
{
fieldInfos = new FieldInfos(); // merge field names
}
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader) reader;
for (int j = 0; j < segmentReader.GetFieldInfos().Size(); j++)
{
FieldInfo fi = segmentReader.GetFieldInfos().FieldInfo(j);
fieldInfos.Add(fi.name, fi.isIndexed, fi.storeTermVector, fi.storePositionWithTermVector, fi.storeOffsetWithTermVector, !reader.HasNorms(fi.name), fi.storePayloads, fi.omitTf);
}
}
else
{
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.OMIT_TF), false, false, false, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.STORES_PAYLOADS), false, false, false, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
}
fieldInfos.Write(directory, segment + ".fnm");
int docCount = 0;
SetMatchingSegmentReaders();
if (mergeDocStores)
{
// for merging we don't want to compress/uncompress the data, so to tell the FieldsReader that we're
// in merge mode, we use this FieldSelector
FieldSelector fieldSelectorMerge = new AnonymousClassFieldSelector(this);
// merge field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
SegmentReader matchingSegmentReader = matchingSegmentReaders[i];
FieldsReader matchingFieldsReader;
bool hasMatchingReader;
if (matchingSegmentReader != null)
{
FieldsReader fieldsReader = matchingSegmentReader.GetFieldsReader();
if (fieldsReader != null && !fieldsReader.CanReadRawDocs())
{
matchingFieldsReader = null;
hasMatchingReader = false;
}
else
{
matchingFieldsReader = fieldsReader;
hasMatchingReader = true;
}
}
else
{
hasMatchingReader = false;
matchingFieldsReader = null;
}
int maxDoc = reader.MaxDoc();
bool hasDeletions = reader.HasDeletions();
for (int j = 0; j < maxDoc; )
{
if (!hasDeletions || !reader.IsDeleted(j))
{ // skip deleted docs
if (hasMatchingReader)
{
// We can optimize this case (doing a bulk
// byte copy) since the field numbers are
// identical
int start = j;
int numDocs = 0;
do
{
j++;
numDocs++;
if (j >= maxDoc)
break;
if (hasDeletions && matchingSegmentReader.IsDeleted(j))
{
j++;
break;
}
} while (numDocs < MAX_RAW_MERGE_DOCS);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, start, numDocs);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, numDocs);
docCount += numDocs;
if (checkAbort != null)
checkAbort.Work(300 * numDocs);
}
else
{
// NOTE: it's very important to first assign
// to doc then pass it to
// termVectorsWriter.addAllDocVectors; see
// LUCENE-1282
Document doc = reader.Document(j, fieldSelectorMerge);
fieldsWriter.AddDocument(doc);
j++;
docCount++;
if (checkAbort != null)
checkAbort.Work(300);
}
}
else
j++;
}
}
}
finally
{
fieldsWriter.Close();
}
long fdxFileLength = directory.FileLength(segment + "." + IndexFileNames.FIELDS_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+docCount*8 != fdxFileLength)
// // 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("mergeFields produced an invalid result: docCount is " + docCount + " but fdx file size is " + fdxFileLength + "; now aborting this merge to prevent index corruption");
}
else
// If we are skipping the doc stores, that means there
// are no deletions in any of these segments, so we
// just sum numDocs() of each segment to get total docCount
for (int i = 0; i < readers.Count; i++)
docCount += ((IndexReader)readers[i]).NumDocs();
return docCount;
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> CorruptIndexException if the index is corrupt </throws>
/// <throws> IOException if there is a low-level IO error </throws>
private int MergeFields()
{
if (!mergeDocStores)
{
// When we are not merging by doc stores, that means
// all segments were written as part of a single
// autoCommit=false IndexWriter session, so their field
// name -> number mapping are the same. So, we start
// with the fieldInfos of the last segment in this
// case, to keep that numbering.
SegmentReader sr = (SegmentReader) readers[readers.Count - 1];
fieldInfos = (FieldInfos) sr.core.fieldInfos.Clone();
}
else
{
fieldInfos = new FieldInfos(); // merge field names
}
for (System.Collections.IEnumerator iter = readers.GetEnumerator(); iter.MoveNext(); )
{
IndexReader reader = (IndexReader) iter.Current;
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader) reader;
FieldInfos readerFieldInfos = segmentReader.FieldInfos();
int numReaderFieldInfos = readerFieldInfos.Size();
for (int j = 0; j < numReaderFieldInfos; j++)
{
FieldInfo fi = readerFieldInfos.FieldInfo(j);
fieldInfos.Add(fi.name, fi.isIndexed, fi.storeTermVector, fi.storePositionWithTermVector, fi.storeOffsetWithTermVector, !reader.HasNorms(fi.name), fi.storePayloads, fi.omitTermFreqAndPositions);
}
}
else
{
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR_WITH_POSITION), true, true, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR), true, false, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.OMIT_TERM_FREQ_AND_POSITIONS), false, false, false, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.STORES_PAYLOADS), false, false, false, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.INDEXED), false, false, false, false, false);
fieldInfos.Add(reader.GetFieldNames(FieldOption.UNINDEXED), false);
}
}
fieldInfos.Write(directory, segment + ".fnm");
int docCount = 0;
SetMatchingSegmentReaders();
if (mergeDocStores)
{
// for merging we don't want to compress/uncompress the data, so to tell the FieldsReader that we're
// in merge mode, we use this FieldSelector
FieldSelector fieldSelectorMerge = new AnonymousClassFieldSelector(this);
// merge field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
int idx = 0;
for (System.Collections.IEnumerator iter = readers.GetEnumerator(); iter.MoveNext(); )
{
IndexReader reader = (IndexReader) iter.Current;
SegmentReader matchingSegmentReader = matchingSegmentReaders[idx++];
FieldsReader matchingFieldsReader = null;
if (matchingSegmentReader != null)
{
FieldsReader fieldsReader = matchingSegmentReader.GetFieldsReader();
if (fieldsReader != null && fieldsReader.CanReadRawDocs())
{
matchingFieldsReader = fieldsReader;
}
}
if (reader.HasDeletions())
{
docCount += CopyFieldsWithDeletions(fieldSelectorMerge, fieldsWriter, reader, matchingFieldsReader);
}
else
{
docCount += CopyFieldsNoDeletions(fieldSelectorMerge, fieldsWriter, reader, matchingFieldsReader);
}
}
}
finally
{
fieldsWriter.Close();
}
System.String fileName = segment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION;
long fdxFileLength = directory.FileLength(fileName);
if (4 + ((long) docCount) * 8 != fdxFileLength)
// 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 System.SystemException("mergeFields produced an invalid result: docCount is " + docCount + " but fdx file size is " + fdxFileLength + " file=" + fileName + " file exists?=" + directory.FileExists(fileName) + "; now aborting this merge to prevent index corruption");
}
// If we are skipping the doc stores, that means there
// are no deletions in any of these segments, so we
// just sum numDocs() of each segment to get total docCount
else
{
for (System.Collections.IEnumerator iter = readers.GetEnumerator(); iter.MoveNext(); )
{
docCount += ((IndexReader) iter.Current).NumDocs();
}
}
return docCount;
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> CorruptIndexException if the index is corrupt </throws>
/// <throws> IOException if there is a low-level IO error </throws>
private int MergeFields()
{
if (!mergeDocStores)
{
// When we are not merging by doc stores, their field
// name -> number mapping are the same. So, we start
// with the fieldInfos of the last segment in this
// case, to keep that numbering.
SegmentReader sr = (SegmentReader)readers[readers.Count - 1];
fieldInfos = (FieldInfos)sr.core.fieldInfos.Clone();
}
else
{
fieldInfos = new FieldInfos(); // merge field names
}
foreach (IndexReader reader in readers)
{
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader)reader;
FieldInfos readerFieldInfos = segmentReader.FieldInfos();
int numReaderFieldInfos = readerFieldInfos.Size();
for (int j = 0; j < numReaderFieldInfos; j++)
{
FieldInfo fi = readerFieldInfos.FieldInfo(j);
fieldInfos.Add(fi.name, fi.isIndexed, fi.storeTermVector, fi.storePositionWithTermVector, fi.storeOffsetWithTermVector, !reader.HasNorms(fi.name), fi.storePayloads, fi.omitTermFreqAndPositions);
}
}
else
{
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR_WITH_POSITION), true, true, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.TERMVECTOR), true, false, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.OMIT_TERM_FREQ_AND_POSITIONS), false, false, false, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.STORES_PAYLOADS), false, false, false, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(FieldOption.INDEXED), false, false, false, false, false);
fieldInfos.Add(reader.GetFieldNames(FieldOption.UNINDEXED), false);
}
}
fieldInfos.Write(directory, segment + ".fnm");
int docCount = 0;
SetMatchingSegmentReaders();
if (mergeDocStores)
{
// merge field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
int idx = 0;
foreach (IndexReader reader in readers)
{
SegmentReader matchingSegmentReader = matchingSegmentReaders[idx++];
FieldsReader matchingFieldsReader = null;
if (matchingSegmentReader != null)
{
FieldsReader fieldsReader = matchingSegmentReader.GetFieldsReader();
if (fieldsReader != null && fieldsReader.CanReadRawDocs())
{
matchingFieldsReader = fieldsReader;
}
}
if (reader.HasDeletions)
{
docCount += CopyFieldsWithDeletions(fieldsWriter, reader, matchingFieldsReader);
}
else
{
docCount += CopyFieldsNoDeletions(fieldsWriter, reader, matchingFieldsReader);
}
}
}
finally
{
fieldsWriter.Dispose();
}
System.String fileName = segment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION;
long fdxFileLength = directory.FileLength(fileName);
if (4 + ((long)docCount) * 8 != fdxFileLength)
{
// 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 System.SystemException("mergeFields produced an invalid result: docCount is " + docCount + " but fdx file size is " + fdxFileLength + " file=" + fileName + " file exists?=" + directory.FileExists(fileName) + "; now aborting this merge to prevent index corruption");
}
}
// If we are skipping the doc stores, that means there
// are no deletions in any of these segments, so we
// just sum numDocs() of each segment to get total docCount
else
{
foreach (IndexReader reader in readers)
{
docCount += reader.NumDocs();
}
}
return(docCount);
}
/// <summary>Initializes shared state for this new document </summary>
internal void Init(Document doc, int docID)
{
System.Diagnostics.Debug.Assert(!isIdle);
System.Diagnostics.Debug.Assert(Enclosing_Instance.writer.TestPoint("DocumentsWriter.ThreadState.init start"));
this.docID = docID;
docBoost = doc.GetBoost();
numStoredFields = 0;
numFieldData = 0;
numVectorFields = 0;
maxTermPrefix = null;
System.Diagnostics.Debug.Assert(0 == fdtLocal.Length());
System.Diagnostics.Debug.Assert(0 == fdtLocal.GetFilePointer());
System.Diagnostics.Debug.Assert(0 == tvfLocal.Length());
System.Diagnostics.Debug.Assert(0 == tvfLocal.GetFilePointer());
int thisFieldGen = fieldGen++;
System.Collections.IList docFields = doc.GetFields();
int numDocFields = docFields.Count;
bool docHasVectors = false;
// Absorb any new fields first seen in this document.
// Also absorb any changes to fields we had already
// seen before (eg suddenly turning on norms or
// vectors, etc.):
for (int i = 0; i < numDocFields; i++)
{
Fieldable field = (Fieldable) docFields[i];
FieldInfo fi = Enclosing_Instance.fieldInfos.Add(field.Name(), field.IsIndexed(), field.IsTermVectorStored(), field.IsStorePositionWithTermVector(), field.IsStoreOffsetWithTermVector(), field.GetOmitNorms(), false);
if (fi.isIndexed && !fi.omitNorms)
{
// Maybe grow our buffered norms
if (Enclosing_Instance.norms.Length <= fi.number)
{
int newSize = (int) ((1 + fi.number) * 1.25);
BufferedNorms[] newNorms = new BufferedNorms[newSize];
Array.Copy(Enclosing_Instance.norms, 0, newNorms, 0, Enclosing_Instance.norms.Length);
Enclosing_Instance.norms = newNorms;
}
if (Enclosing_Instance.norms[fi.number] == null)
Enclosing_Instance.norms[fi.number] = new BufferedNorms();
Enclosing_Instance.hasNorms = true;
}
// Make sure we have a FieldData allocated
int hashPos = fi.name.GetHashCode() & fieldDataHashMask;
FieldData fp = fieldDataHash[hashPos];
while (fp != null && !fp.fieldInfo.name.Equals(fi.name))
fp = fp.next;
if (fp == null)
{
fp = new FieldData(this, fi);
fp.next = fieldDataHash[hashPos];
fieldDataHash[hashPos] = fp;
if (numAllFieldData == allFieldDataArray.Length)
{
int newSize = (int) (allFieldDataArray.Length * 1.5);
int newHashSize = fieldDataHash.Length * 2;
FieldData[] newArray = new FieldData[newSize];
FieldData[] newHashArray = new FieldData[newHashSize];
Array.Copy(allFieldDataArray, 0, newArray, 0, numAllFieldData);
// Rehash
fieldDataHashMask = newSize - 1;
for (int j = 0; j < fieldDataHash.Length; j++)
{
FieldData fp0 = fieldDataHash[j];
while (fp0 != null)
{
hashPos = fp0.fieldInfo.name.GetHashCode() & fieldDataHashMask;
FieldData nextFP0 = fp0.next;
fp0.next = newHashArray[hashPos];
newHashArray[hashPos] = fp0;
fp0 = nextFP0;
}
}
allFieldDataArray = newArray;
fieldDataHash = newHashArray;
}
allFieldDataArray[numAllFieldData++] = fp;
}
else
{
System.Diagnostics.Debug.Assert(fp.fieldInfo == fi);
}
if (thisFieldGen != fp.lastGen)
{
// First time we're seeing this field for this doc
fp.lastGen = thisFieldGen;
fp.fieldCount = 0;
fp.doVectors = fp.doVectorPositions = fp.doVectorOffsets = false;
fp.doNorms = fi.isIndexed && !fi.omitNorms;
if (numFieldData == fieldDataArray.Length)
{
int newSize = fieldDataArray.Length * 2;
FieldData[] newArray = new FieldData[newSize];
Array.Copy(fieldDataArray, 0, newArray, 0, numFieldData);
fieldDataArray = newArray;
}
fieldDataArray[numFieldData++] = fp;
}
if (field.IsTermVectorStored())
{
if (!fp.doVectors && numVectorFields++ == vectorFieldPointers.Length)
{
int newSize = (int) (numVectorFields * 1.5);
vectorFieldPointers = new long[newSize];
vectorFieldNumbers = new int[newSize];
}
fp.doVectors = true;
docHasVectors = true;
fp.doVectorPositions |= field.IsStorePositionWithTermVector();
fp.doVectorOffsets |= field.IsStoreOffsetWithTermVector();
}
if (fp.fieldCount == fp.docFields.Length)
{
Fieldable[] newArray = new Fieldable[fp.docFields.Length * 2];
Array.Copy(fp.docFields, 0, newArray, 0, fp.docFields.Length);
fp.docFields = newArray;
}
// Lazily allocate arrays for postings:
if (field.IsIndexed() && fp.postingsHash == null)
fp.InitPostingArrays();
fp.docFields[fp.fieldCount++] = field;
}
// Maybe init the local & global fieldsWriter
if (localFieldsWriter == null)
{
if (Enclosing_Instance.fieldsWriter == null)
{
System.Diagnostics.Debug.Assert(Enclosing_Instance.docStoreSegment == null);
System.Diagnostics.Debug.Assert(Enclosing_Instance.segment != null);
Enclosing_Instance.docStoreSegment = Enclosing_Instance.segment;
// If we hit an exception while init'ing the
// fieldsWriter, we must abort this segment
// because those files will be in an unknown
// state:
try
{
Enclosing_Instance.fieldsWriter = new FieldsWriter(Enclosing_Instance.directory, Enclosing_Instance.docStoreSegment, Enclosing_Instance.fieldInfos);
}
catch (System.Exception t)
{
throw new AbortException(t, Enclosing_Instance);
}
Enclosing_Instance.files = null;
}
localFieldsWriter = new FieldsWriter(null, fdtLocal, Enclosing_Instance.fieldInfos);
}
// First time we see a doc that has field(s) with
// stored vectors, we init our tvx writer
if (docHasVectors)
{
if (Enclosing_Instance.tvx == null)
{
System.Diagnostics.Debug.Assert(Enclosing_Instance.docStoreSegment != null);
// If we hit an exception while init'ing the term
// vector output files, we must abort this segment
// because those files will be in an unknown
// state:
try
{
Enclosing_Instance.tvx = Enclosing_Instance.directory.CreateOutput(Enclosing_Instance.docStoreSegment + "." + IndexFileNames.VECTORS_INDEX_EXTENSION);
Enclosing_Instance.tvx.WriteInt(TermVectorsReader.FORMAT_VERSION);
Enclosing_Instance.tvd = Enclosing_Instance.directory.CreateOutput(Enclosing_Instance.docStoreSegment + "." + IndexFileNames.VECTORS_DOCUMENTS_EXTENSION);
Enclosing_Instance.tvd.WriteInt(TermVectorsReader.FORMAT_VERSION);
Enclosing_Instance.tvf = Enclosing_Instance.directory.CreateOutput(Enclosing_Instance.docStoreSegment + "." + IndexFileNames.VECTORS_FIELDS_EXTENSION);
Enclosing_Instance.tvf.WriteInt(TermVectorsReader.FORMAT_VERSION);
// We must "catch up" for all docs before us
// that had no vectors:
for (int i = 0; i < Enclosing_Instance.numDocsInStore; i++)
{
Enclosing_Instance.tvx.WriteLong(Enclosing_Instance.tvd.GetFilePointer());
Enclosing_Instance.tvd.WriteVInt(0);
}
}
catch (System.Exception t)
{
throw new AbortException(t, Enclosing_Instance);
}
Enclosing_Instance.files = null;
}
numVectorFields = 0;
}
}
/// <summary>Called if we hit an exception when adding docs,
/// flushing, etc. This resets our state, discarding any
/// docs added since last flush. If ae is non-null, it
/// contains the root cause exception (which we re-throw
/// after we are done aborting).
/// </summary>
internal void Abort(AbortException ae)
{
lock (this)
{
// Anywhere that throws an AbortException must first
// mark aborting to make sure while the exception is
// unwinding the un-synchronized stack, no thread grabs
// the corrupt ThreadState that hit the aborting
// exception:
System.Diagnostics.Debug.Assert(ae == null || abortCount > 0);
try
{
if (infoStream != null)
infoStream.WriteLine("docWriter: now abort");
// Forcefully remove waiting ThreadStates from line
for (int i = 0; i < numWaiting; i++)
waitingThreadStates[i].isIdle = true;
numWaiting = 0;
// Wait for all other threads to finish with DocumentsWriter:
PauseAllThreads();
System.Diagnostics.Debug.Assert(0 == numWaiting);
try
{
bufferedDeleteTerms.Clear();
bufferedDeleteDocIDs.Clear();
numBufferedDeleteTerms = 0;
try
{
abortedFiles = Files();
}
catch (System.Exception)
{
abortedFiles = null;
}
docStoreSegment = null;
numDocsInStore = 0;
docStoreOffset = 0;
files = null;
// Clear vectors & fields from ThreadStates
for (int i = 0; i < threadStates.Length; i++)
{
ThreadState state = threadStates[i];
state.tvfLocal.Reset();
state.fdtLocal.Reset();
if (state.localFieldsWriter != null)
{
try
{
state.localFieldsWriter.Close();
}
catch (System.Exception)
{
}
state.localFieldsWriter = null;
}
}
// Reset vectors writer
if (tvx != null)
{
try
{
tvx.Close();
}
catch (System.Exception)
{
}
tvx = null;
}
if (tvd != null)
{
try
{
tvd.Close();
}
catch (System.Exception)
{
}
tvd = null;
}
if (tvf != null)
{
try
{
tvf.Close();
}
catch (System.Exception)
{
}
tvf = null;
}
// Reset fields writer
if (fieldsWriter != null)
{
try
{
fieldsWriter.Close();
}
catch (System.Exception)
{
}
fieldsWriter = null;
}
// Discard pending norms:
int numField = fieldInfos.Size();
for (int i = 0; i < numField; i++)
{
FieldInfo fi = fieldInfos.FieldInfo(i);
if (fi.isIndexed && !fi.omitNorms)
{
BufferedNorms n = norms[i];
if (n != null)
try
{
n.Reset();
}
catch (System.Exception)
{
}
}
}
// Reset all postings data
ResetPostingsData();
}
finally
{
ResumeAllThreads();
}
// If we have a root cause exception, re-throw it now:
if (ae != null)
{
System.Exception t = ae.InnerException;
if (t is System.IO.IOException)
throw (System.IO.IOException) t;
else if (t is System.SystemException)
throw (System.SystemException) t;
else if (t is System.ApplicationException)
throw (System.ApplicationException) t;
else
// Should not get here
System.Diagnostics.Debug.Assert(false, "unknown exception: " + t);
}
}
finally
{
if (ae != null)
abortCount--;
System.Threading.Monitor.PulseAll(this);
}
}
}
public StoredFieldsWriterPerThread(DocumentsWriter.DocState docState, StoredFieldsWriter storedFieldsWriter)
{
this.storedFieldsWriter = storedFieldsWriter;
this.docState = docState;
localFieldsWriter = new FieldsWriter((IndexOutput) null, (IndexOutput) null, storedFieldsWriter.fieldInfos);
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> CorruptIndexException if the index is corrupt </throws>
/// <throws> IOException if there is a low-level IO error </throws>
private int MergeFields()
{
if (!mergeDocStores)
{
// When we are not merging by doc stores, that means
// all segments were written as part of a single
// autoCommit=false IndexWriter session, so their field
// name -> number mapping are the same. So, we start
// with the fieldInfos of the last segment in this
// case, to keep that numbering.
SegmentReader sr = (SegmentReader)readers[readers.Count - 1];
fieldInfos = (FieldInfos)sr.fieldInfos.Clone();
}
else
{
fieldInfos = new FieldInfos(); // merge field names
}
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader)reader;
for (int j = 0; j < segmentReader.GetFieldInfos().Size(); j++)
{
FieldInfo fi = segmentReader.GetFieldInfos().FieldInfo(j);
fieldInfos.Add(fi.name, fi.isIndexed, fi.storeTermVector, fi.storePositionWithTermVector, fi.storeOffsetWithTermVector, !reader.HasNorms(fi.name), fi.storePayloads);
}
}
else
{
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.STORES_PAYLOADS), false, false, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
}
fieldInfos.Write(directory, segment + ".fnm");
int docCount = 0;
if (mergeDocStores)
{
// If the i'th reader is a SegmentReader and has
// identical fieldName -> number mapping, then this
// array will be non-null at position i:
SegmentReader[] matchingSegmentReaders = new SegmentReader[readers.Count];
// If this reader is a SegmentReader, and all of its
// field name -> number mappings match the "merged"
// FieldInfos, then we can do a bulk copy of the
// stored fields:
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
if (reader is SegmentReader)
{
SegmentReader segmentReader = (SegmentReader)reader;
bool same = true;
FieldInfos segmentFieldInfos = segmentReader.GetFieldInfos();
for (int j = 0; same && j < segmentFieldInfos.Size(); j++)
{
same = fieldInfos.FieldName(j).Equals(segmentFieldInfos.FieldName(j));
}
if (same)
{
matchingSegmentReaders[i] = segmentReader;
}
}
}
// Used for bulk-reading raw bytes for stored fields
int[] rawDocLengths = new int[MAX_RAW_MERGE_DOCS];
// for merging we don't want to compress/uncompress the data, so to tell the FieldsReader that we're
// in merge mode, we use this FieldSelector
FieldSelector fieldSelectorMerge = new AnonymousClassFieldSelector(this);
// merge field values
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader)readers[i];
SegmentReader matchingSegmentReader = matchingSegmentReaders[i];
FieldsReader matchingFieldsReader;
if (matchingSegmentReader != null)
{
matchingFieldsReader = matchingSegmentReader.GetFieldsReader();
}
else
{
matchingFieldsReader = null;
}
int maxDoc = reader.MaxDoc();
for (int j = 0; j < maxDoc;)
{
if (!reader.IsDeleted(j))
{
// skip deleted docs
if (matchingSegmentReader != null)
{
// We can optimize this case (doing a bulk
// byte copy) since the field numbers are
// identical
int start = j;
int numDocs = 0;
do
{
j++;
numDocs++;
}while (j < maxDoc && !matchingSegmentReader.IsDeleted(j) && numDocs < MAX_RAW_MERGE_DOCS);
IndexInput stream = matchingFieldsReader.RawDocs(rawDocLengths, start, numDocs);
fieldsWriter.AddRawDocuments(stream, rawDocLengths, numDocs);
docCount += numDocs;
if (checkAbort != null)
{
checkAbort.Work(300 * numDocs);
}
}
else
{
fieldsWriter.AddDocument(reader.Document(j, fieldSelectorMerge));
j++;
docCount++;
if (checkAbort != null)
{
checkAbort.Work(300);
}
}
}
else
{
j++;
}
}
}
}
finally
{
fieldsWriter.Close();
}
}
// If we are skipping the doc stores, that means there
// are no deletions in any of these segments, so we
// just sum numDocs() of each segment to get total docCount
else
{
for (int i = 0; i < readers.Count; i++)
{
docCount += ((IndexReader)readers[i]).NumDocs();
}
}
return(docCount);
}
public StoredFieldsWriterPerThread(DocFieldProcessorPerThread docFieldProcessorPerThread, StoredFieldsWriter storedFieldsWriter)
{
this.storedFieldsWriter = storedFieldsWriter;
this.docState = docFieldProcessorPerThread.docState;
localFieldsWriter = new FieldsWriter((IndexOutput)null, (IndexOutput)null, storedFieldsWriter.fieldInfos);
}
/// <summary>Clear the postings hash and return objects back to
/// shared pool
/// </summary>
public void ResetPostings()
{
fieldGen = 0;
maxPostingsVectors = 0;
doFlushAfter = false;
if (localFieldsWriter != null)
{
localFieldsWriter.Close();
localFieldsWriter = null;
}
postingsPool.Reset();
charPool.Reset();
Enclosing_Instance.RecyclePostings(postingsFreeList, postingsFreeCount);
postingsFreeCount = 0;
for (int i = 0; i < numAllFieldData; i++)
{
FieldData fp = allFieldDataArray[i];
fp.lastGen = - 1;
if (fp.numPostings > 0)
fp.ResetPostingArrays();
}
}
internal override void closeDocStore(DocumentsWriter.FlushState state)
{
lock (this)
{
int inc = state.numDocsInStore - lastDocID;
if (inc > 0)
{
initFieldsWriter();
fill(state.numDocsInStore - docWriter.GetDocStoreOffset());
}
if (fieldsWriter != null)
{
fieldsWriter.Close();
fieldsWriter = null;
lastDocID = 0;
System.Diagnostics.Debug.Assert(state.docStoreSegmentName != null);
string fdtFile = state.docStoreSegmentName + "." + IndexFileNames.FIELDS_EXTENSION;
string fdxFile = state.docStoreSegmentName + "." + IndexFileNames.FIELDS_INDEX_EXTENSION;
state.flushedFiles[fdtFile] = fdtFile;
state.flushedFiles[fdxFile] = fdxFile;
state.docWriter.RemoveOpenFile(fdtFile);
state.docWriter.RemoveOpenFile(fdxFile);
if (4 + state.numDocsInStore * 8 != state.directory.FileLength(fdxFile))
throw new System.SystemException("after flush: fdx size mismatch: " + state.numDocsInStore + " docs vs " + state.directory.FileLength(fdxFile) + " length in bytes of " + fdxFile);
}
}
}
/// <summary>Closes the current open doc stores an returns the doc
/// store segment name. This returns null if there are *
/// no buffered documents.
/// </summary>
internal System.String CloseDocStore()
{
System.Diagnostics.Debug.Assert(AllThreadsIdle());
System.Collections.IList flushedFiles = Files();
if (infoStream != null)
infoStream.WriteLine("\ncloseDocStore: " + flushedFiles.Count + " files to flush to segment " + docStoreSegment + " numDocs=" + numDocsInStore);
if (flushedFiles.Count > 0)
{
files = null;
if (tvx != null)
{
// At least one doc in this run had term vectors enabled
System.Diagnostics.Debug.Assert(docStoreSegment != null);
tvx.Close();
tvf.Close();
tvd.Close();
tvx = null;
System.Diagnostics.Debug.Assert(4 + numDocsInStore * 8 == directory.FileLength(docStoreSegment + "." + IndexFileNames.VECTORS_INDEX_EXTENSION),
"after flush: tvx size mismatch: " + numDocsInStore + " docs vs " + directory.FileLength(docStoreSegment + "." + IndexFileNames.VECTORS_INDEX_EXTENSION) +
" length in bytes of " + docStoreSegment + "." + IndexFileNames.VECTORS_INDEX_EXTENSION);
}
if (fieldsWriter != null)
{
System.Diagnostics.Debug.Assert(docStoreSegment != null);
fieldsWriter.Close();
fieldsWriter = null;
System.Diagnostics.Debug.Assert(numDocsInStore * 8 == directory.FileLength(docStoreSegment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION),
"after flush: fdx size mismatch: " + numDocsInStore + " docs vs " + directory.FileLength(docStoreSegment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION) +
" length in bytes of " + docStoreSegment + "." + IndexFileNames.FIELDS_INDEX_EXTENSION);
}
System.String s = docStoreSegment;
docStoreSegment = null;
docStoreOffset = 0;
numDocsInStore = 0;
return s;
}
else
{
return null;
}
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> IOException </throws>
private int MergeFields()
{
fieldInfos = new FieldInfos(); // merge field names
int docCount = 0;
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
fieldInfos.Write(directory, segment + ".fnm");
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
int maxDoc = reader.MaxDoc();
for (int j = 0; j < maxDoc; j++)
if (!reader.IsDeleted(j))
{
// skip deleted docs
fieldsWriter.AddDocument(reader.Document(j));
docCount++;
}
}
}
finally
{
fieldsWriter.Close();
}
return docCount;
}
public StoredFieldsWriterPerThread(DocFieldProcessorPerThread docFieldProcessorPerThread, StoredFieldsWriter storedFieldsWriter)
{
this.storedFieldsWriter = storedFieldsWriter;
this.docState = docFieldProcessorPerThread.docState;
localFieldsWriter = new FieldsWriter((IndexOutput)null, (IndexOutput)null, storedFieldsWriter.fieldInfos);
}
public StoredFieldsWriterPerThread(DocumentsWriter.DocState docState, StoredFieldsWriter storedFieldsWriter)
{
this.storedFieldsWriter = storedFieldsWriter;
this.docState = docState;
localFieldsWriter = new FieldsWriter((IndexOutput)null, (IndexOutput)null, storedFieldsWriter.fieldInfos);
}
/// <summary> </summary>
/// <returns> The number of documents in all of the readers
/// </returns>
/// <throws> IOException </throws>
private int MergeFields()
{
fieldInfos = new FieldInfos(); // merge field names
int docCount = 0;
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION_OFFSET), true, true, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_POSITION), true, true, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR_WITH_OFFSET), true, false, true);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.TERMVECTOR), true, false, false);
AddIndexed(reader, fieldInfos, reader.GetFieldNames(IndexReader.FieldOption.INDEXED), false, false, false);
fieldInfos.Add(reader.GetFieldNames(IndexReader.FieldOption.UNINDEXED), false);
}
fieldInfos.Write(directory, segment + ".fnm");
FieldsWriter fieldsWriter = new FieldsWriter(directory, segment, fieldInfos);
// for merging we don't want to compress/uncompress the data, so to tell the FieldsReader that we're
// in merge mode, we use this FieldSelector
FieldSelector fieldSelectorMerge = new AnonymousClassFieldSelector(this);
try
{
for (int i = 0; i < readers.Count; i++)
{
IndexReader reader = (IndexReader) readers[i];
int maxDoc = reader.MaxDoc();
for (int j = 0; j < maxDoc; j++)
if (!reader.IsDeleted(j))
{
// skip deleted docs
fieldsWriter.AddDocument(reader.Document(j, fieldSelectorMerge));
docCount++;
}
}
}
finally
{
fieldsWriter.Close();
}
return docCount;
}
