public void AddWordAndDocList(string word, DocumentPositionList first, int docsCount, IEnumerable <Entity.DocumentPositionList> docList)
{
FileLengthType length;
FilePositionType position = _IndexWriter.AddWordAndDocList(word, first, docsCount, docList, out length);
_WordFilePositionList.Add(new WordFilePosition(word, _MaxSerial, position, length));
}
/// <summary>
/// Get step doc index.
/// </summary>
/// <param name="position">first byte of the index (include step doc index)</param>
/// <param name="length">index buffer length</param>
/// <param name="count">count of the document records in this index</param>
/// <param name="indexPostion">first byte of the index (exclude step doc index)<</param>
/// <returns>step doc index. If no step doc index, return null</returns>
public List <DocumentPosition> GetStepDocIndex(long position, long length,
out int count, out long indexPostion)
{
Query.PerformanceReport performanceReport = new Hubble.Core.Query.PerformanceReport();
List <DocumentPosition> result;
//indexPostion = position;
_IndexFile.Seek(position, System.IO.SeekOrigin.Begin);
result = DocumentPositionList.DeserializeSkipDocIndex(_IndexFile, false, out count);
if (result != null)
{
count = DocumentPositionList.GetDocumentsCount(_IndexFile);
}
indexPostion = _IndexFile.Position; //The index position exclude count;
performanceReport.Stop(string.Format("Read index file: len={0}, {1} results. ", _IndexFile.Position - position,
count));
return(result);
}
/// <summary>
/// Get next document position list
/// </summary>
/// <returns>if end of the index, docid of result is -1</returns>
public DocumentPositionList GetNext()
{
if (_End)
{
return(new DocumentPositionList(-1));
}
if (_CurrentDocId < 0)
{
if (!NextFileIndex())
{
//End of the index
return(new DocumentPositionList(-1));
}
}
if (_CurrentIndexBuf.Position < _CurrentIndexBuf.Start + _CurrentIndexBuf.Length)
{
return(DocumentPositionList.GetNextDocumentPositionList(ref _CurrentDocId,
_CurrentIndexBuf, _WordStepDocIndex.SimpleMode));
}
else
{
if (!NextFileIndexBuf())
{
return(new DocumentPositionList(-1));
}
return(DocumentPositionList.GetNextDocumentPositionList(ref _CurrentDocId,
_CurrentIndexBuf, _WordStepDocIndex.SimpleMode));
}
}
internal DocIndexInfo(DocumentPositionList dpl)
{
this.DocumentId = dpl.DocumentId;
this.FirstPosition = dpl.FirstPosition;
this.Count = dpl.Count;
this.TotalWordsInThisDocumentIndex = dpl.TotalWordsInThisDocumentIndex;
//this.PayloadData = dbProvider.GetPayloadData(this.DocumentId);
}
internal DocumentPositionList GetDocPositionList()
{
DocumentPositionList dpl = new DocumentPositionList();
dpl.DocumentId = this.DocumentId;
dpl.FirstPosition = this.FirstPosition;
dpl.Count = this.Count;
dpl.SetTotalWordsInThisDocumentIndex(this.TotalWordsInThisDocumentIndex);
//this.PayloadData = dbProvider.GetPayloadData(this.DocumentId);
return(dpl);
}
public void Reset()
{
_CurrentFileIndex = -1;
_CurrentIndexBuf = null;
_CurrentDocId = -1;
_LastDocIdInCurrentStep = -1;
_CurrentStepDocIndex = -1;
_End = false;
_NextDocPositionList = new DocumentPositionList(-1);
_NextODPL.DocumentId = -1;
}
/// <summary>
/// Add document list of one word into inverted index file.
/// </summary>
/// <param name="word">word</param>
/// <param name="simple">is it simple index?</param>
/// <param name="first">first document position information</param>
/// <param name="docsCount">document count that want to added</param>
/// <param name="docList">doucment position list</param>
/// <param name="length">output the length of index content</param>
/// <returns>position of the first byte of this word's index in .idx file</returns>
public long AddWordAndDocList(string word, bool simple,
DocumentPositionList first, int docsCount,
IEnumerable <Entity.DocumentPositionList> docList, out int length)
{
long position = _IndexFile.Position;
Entity.DocumentPositionList.Serialize(first, docsCount, docList, _IndexFile, simple);
length = (int)(_IndexFile.Position - position);
return(position);
}
public DocumentPositionList Get(int docId)
{
if (_End)
{
return(new DocumentPositionList(-1));
}
if (docId < 0)
{
throw new StoreException(string.Format("Invalid docid = {0}", docId));
}
if (_CurrentFileIndex < 0)
{
_NextDocPositionList = GetNext();
}
if (docId < _NextDocPositionList.DocumentId)
{
return(new DocumentPositionList(-1));
}
if (docId == _NextDocPositionList.DocumentId)
{
DocumentPositionList result = _NextDocPositionList;
_NextDocPositionList = GetNext();
return(result);
}
Match(docId); //Get the matched step doc index
_NextDocPositionList = GetNext();
while (_NextDocPositionList.DocumentId < docId && _NextDocPositionList.DocumentId >= 0)
{
_NextDocPositionList = GetNext();
}
if (docId == _NextDocPositionList.DocumentId)
{
DocumentPositionList result = _NextDocPositionList;
_NextDocPositionList = GetNext();
return(result);
}
else
{
return(new DocumentPositionList(-1));
}
}
internal int Alloc()
{
_Current++;
if (_Current >= DocPositionPool.Length)
{
DocumentPositionList[] tempPool = new DocumentPositionList[DocPositionPool.Length * 2];
Array.Copy(DocPositionPool, tempPool, DocPositionPool.Length);
DocPositionPool = tempPool;
}
return(_Current);
}
public long AddWordAndDocList(string word, DocumentPositionList first, int docsCount, IEnumerable <Entity.DocumentPositionList> docList, out int length)
{
long position = _IDXFile.AddWordAndDocList(word, _IndexMode == Hubble.Core.Data.Field.IndexMode.Simple,
first, docsCount, docList, out length);
//long position = _IndexFile.Position;
//bool simple = _IndexMode == Hubble.Core.Data.Field.IndexMode.Simple;
//Entity.DocumentPositionList.Serialize(first, docsCount, docList, _IndexFile, simple);
//length = (int)(_IndexFile.Position - position);
//WriteHeadFile(_HeadFile, word, position, length);
_DDXFile.Add(word, position, length);
return(position);
}
/// <summary>
/// Index one document into the _ListForWriter
/// </summary>
internal void Index()
{
try
{
if (_IndexMode == Hubble.Core.Data.Field.IndexMode.Simple)
{
Add(new DocumentPositionList(TempDocId, TempWordCountInThisDoc,
DocumentPositionList.GetTotalWordsInDocIndex(TempTotalWordsInDoc)));
//_ListForWriter.Add(new DocumentPositionList(DocId, WordCountInThisDoc,
// DocumentPositionList.GetTotalWordsInDocIndex(TempTotalWordsInDoc)));
}
else
{
if (TempWordCountInThisDoc > 0)
{
Add(new DocumentPositionList(TempDocId,
TempWordCountInThisDoc,
DocumentPositionList.GetTotalWordsInDocIndex(TempTotalWordsInDoc),
TempFirstPosition));
//_ListForWriter.Add(new DocumentPositionList(DocId,
// WordCountInThisDoc,
// DocumentPositionList.GetTotalWordsInDocIndex(TempTotalWordsInDoc),
// FirstPosition));
}
else
{
Add(new DocumentPositionList(TempDocId, TempWordCountInThisDoc,
DocumentPositionList.GetTotalWordsInDocIndex(TempTotalWordsInDoc)));
//_ListForWriter.Add(new DocumentPositionList(DocId, WordCountInThisDoc,
// DocumentPositionList.GetTotalWordsInDocIndex(TempTotalWordsInDoc)));
}
}
_Count++;
}
finally
{
}
}
private void Add(DocumentPositionList docPositionList)
{
int docPositionId = _DocPositionAlloc.Alloc();
if (_First < 0)
{
_First = docPositionId;
}
_DocPositionAlloc.DocPositionPool[docPositionId] = docPositionList;
if (_Cur < 0)
{
_Cur = _First;
}
else
{
_DocPositionAlloc.DocPositionPool[_Cur].Next = docPositionId;
_Cur = docPositionId;
}
}
/// <summary>
/// Get next original document position list
/// </summary>
/// <returns>if end of the index, docid of result is -1</returns>
public bool GetNextOriginal(ref OriginalDocumentPositionList odpl)
{
if (_End)
{
odpl.DocumentId = -1;
return(false);
//return new OriginalDocumentPositionList(-1);
}
if (_CurrentDocId < 0)
{
if (!NextFileIndex())
{
//End of the index
odpl.DocumentId = -1;
return(false);
//return new OriginalDocumentPositionList(-1);
}
}
if (_CurrentIndexBuf.Position < _CurrentIndexBuf.Start + _CurrentIndexBuf.Length)
{
DocumentPositionList.GetNextOriginalDocumentPositionList(ref odpl, ref _CurrentDocId,
_CurrentIndexBuf, _WordStepDocIndex.SimpleMode);
return(true);
}
else
{
if (!NextFileIndexBuf())
{
odpl.DocumentId = -1;
return(false);
//return new OriginalDocumentPositionList(-1);
}
DocumentPositionList.GetNextOriginalDocumentPositionList(ref odpl, ref _CurrentDocId,
_CurrentIndexBuf, _WordStepDocIndex.SimpleMode);
return(true);
}
}
public Store.WordDocumentsList GetDocumentPositionList(Store.WordDocumentsList docList)
{
lock (this)
{
Store.WordDocumentsList result;
bool needDel = false;
for (int i = 0; i < docList.Count; i++)
{
if (_DeleteTbl.ContainsKey(docList[i].DocumentId))
{
needDel = true;
docList[i] = new DocumentPositionList(-1);
}
}
if (needDel)
{
result = new Store.WordDocumentsList();
for (int i = 0; i < docList.Count; i++)
{
if (docList[i].DocumentId < 0)
{
result.Add(docList[i]);
}
}
}
else
{
result = docList;
}
return(result);
}
}
