public override void SetNextReader(AtomicReaderContext context)
{
AtomicReader reader = context.AtomicReader;
if (!(reader is BoboSegmentReader))
{
throw new ArgumentException("reader must be a BoboIndexReader");
}
m_currentReader = (BoboSegmentReader)reader;
int docBase = context.DocBase;
m_currentComparer = m_compSource.GetComparer(m_currentReader, docBase);
m_currentQueue = new DocIDPriorityQueue(m_currentComparer, m_numHits, docBase);
if (m_groupBy != null)
{
if (m_facetCountCollectorMulti != null) // _facetCountCollectorMulti.Length >= 1
{
for (int i = 0; i < m_facetCountCollectorMulti.Length; ++i)
{
m_facetCountCollectorMulti[i] = m_groupByMulti[i].GetFacetCountCollectorSource(null, null, true).GetFacetCountCollector(m_currentReader, docBase);
}
//if (_facetCountCollector != null)
// collectTotalGroups();
m_facetCountCollector = m_facetCountCollectorMulti[0];
if (m_facetAccessibleLists != null)
{
for (int i = 0; i < m_facetCountCollectorMulti.Length; ++i)
{
m_facetAccessibleLists[i].Add(m_facetCountCollectorMulti[i]);
}
}
}
if (m_currentValueDocMaps != null)
{
m_currentValueDocMaps.Clear();
}
// NightOwl888: The _collectDocIdCache setting seems to put arrays into
// memory, but then do nothing with the arrays. Seems wasteful and unnecessary.
//if (contextList != null)
//{
// _currentContext = new CollectorContext(_currentReader, docBase, _currentComparer);
// contextList.Add(_currentContext);
//}
}
MyScoreDoc myScoreDoc = (MyScoreDoc)m_tmpScoreDoc;
myScoreDoc.m_queue = m_currentQueue;
myScoreDoc.m_reader = m_currentReader;
myScoreDoc.m_sortValue = null;
m_pqList.Add(m_currentQueue);
m_queueFull = false;
}
protected static BrowseHit[] BuildHits(MyScoreDoc[] scoreDocs, SortField[] sortFields,
IDictionary <string, IFacetHandler> facetHandlerMap, bool fetchStoredFields,
IEnumerable <string> termVectorsToFetch, IFacetHandler groupBy, CombinedFacetAccessible[] groupAccessibles)
{
BrowseHit[] hits = new BrowseHit[scoreDocs.Length];
IEnumerable <IFacetHandler> facetHandlers = facetHandlerMap.Values;
for (int i = scoreDocs.Length - 1; i >= 0; i--)
{
MyScoreDoc fdoc = scoreDocs[i];
BoboIndexReader reader = fdoc.reader;
BrowseHit hit = new BrowseHit();
if (fetchStoredFields)
{
hit.StoredFields = reader.Document(fdoc.Doc);
}
if (termVectorsToFetch != null && termVectorsToFetch.Count() > 0)
{
var tvMap = new Dictionary <string, BrowseHit.TermFrequencyVector>();
hit.TermFreqMap = tvMap;
foreach (string field in termVectorsToFetch)
{
ITermFreqVector tv = reader.GetTermFreqVector(fdoc.Doc, field);
if (tv != null)
{
int[] freqs = tv.GetTermFrequencies();
string[] terms = tv.GetTerms();
tvMap[field] = new BrowseHit.TermFrequencyVector(terms, freqs);
}
}
}
var map = new Dictionary <string, string[]>();
var rawMap = new Dictionary <string, object[]>();
foreach (var facetHandler in facetHandlers)
{
map[facetHandler.Name] = facetHandler.GetFieldValues(reader, fdoc.Doc);
rawMap[facetHandler.Name] = facetHandler.GetRawFieldValues(reader, fdoc.Doc);
}
hit.FieldValues = map;
hit.RawFieldValues = rawMap;
hit.DocId = fdoc.Doc + fdoc.queue.@base;
hit.Score = fdoc.Score;
hit.Comparable = fdoc.Value;
if (groupBy != null)
{
hit.GroupField = groupBy.Name;
hit.GroupValue = hit.GetField(groupBy.Name);
hit.RawGroupValue = hit.GetRawField(groupBy.Name);
if (groupAccessibles != null &&
hit.GroupValue != null &&
groupAccessibles != null &&
groupAccessibles.Length > 0)
{
BrowseFacet facet = groupAccessibles[0].GetFacet(hit.GroupValue);
hit.GroupHitsCount = facet.FacetValueHitCount;
}
}
hits[i] = hit;
}
return(hits);
}
public override void Collect(int doc)
{
++_totalHits;
if (groupBy != null)
{
if (_facetCountCollectorMulti != null)
{
for (int i = 0; i < _facetCountCollectorMulti.Length; ++i)
{
if (_facetCountCollectorMulti[i] != null)
{
_facetCountCollectorMulti[i].Collect(doc);
}
}
if (_count > 0)
{
float score = (_doScoring ? _scorer.Score() : 0.0f);
// NightOwl888: The _collectDocIdCache setting seems to put arrays into
// memory, but then do nothing with the arrays. Seems wasteful and unnecessary.
//if (_collectDocIdCache)
//{
// if (_totalHits > _docIdCacheCapacity)
// {
// _currentDocIdArray = intarraymgr.Get(BLOCK_SIZE);
// docidarraylist.Add(_currentDocIdArray);
// if (_doScoring)
// {
// _currentScoreArray = floatarraymgr.Get(BLOCK_SIZE);
// scorearraylist.Add(_currentScoreArray);
// }
// _docIdCacheCapacity += BLOCK_SIZE;
// _docIdArrayCursor = 0;
// }
// _currentDocIdArray[_docIdArrayCursor] = doc;
// if (_doScoring)
// _currentScoreArray[_docIdArrayCursor] = score;
// ++_docIdArrayCursor;
// ++_currentContext.length;
//}
}
return;
}
else
{
//if (_facetCountCollector != null)
//_facetCountCollector.collect(doc);
if (_count > 0)
{
float score = (_doScoring ? _scorer.Score() : 0.0f);
// NightOwl888: The _collectDocIdCache setting seems to put arrays into
// memory, but then do nothing with the arrays. Seems wasteful and unnecessary.
//if (_collectDocIdCache)
//{
// if (_totalHits > _docIdCacheCapacity)
// {
// _currentDocIdArray = intarraymgr.Get(BLOCK_SIZE);
// docidarraylist.Add(_currentDocIdArray);
// if (_doScoring)
// {
// _currentScoreArray = floatarraymgr.Get(BLOCK_SIZE);
// scorearraylist.Add(_currentScoreArray);
// }
// _docIdCacheCapacity += BLOCK_SIZE;
// _docIdArrayCursor = 0;
// }
// _currentDocIdArray[_docIdArrayCursor] = doc;
// if (_doScoring)
// _currentScoreArray[_docIdArrayCursor] = score;
// ++_docIdArrayCursor;
// ++_currentContext.length;
//}
_tmpScoreDoc.Doc = doc;
_tmpScoreDoc.Score = score;
if (!_queueFull || _currentComparator.Compare(_bottom, _tmpScoreDoc) > 0)
{
int order = groupBy.GetFacetData <FacetDataCache>(_currentReader).OrderArray.Get(doc);
ScoreDoc pre = _currentValueDocMaps.Get(order);
if (pre != null)
{
if (_currentComparator.Compare(pre, _tmpScoreDoc) > 0)
{
ScoreDoc tmp = pre;
_bottom = _currentQueue.Replace(_tmpScoreDoc, pre);
_currentValueDocMaps.Put(order, _tmpScoreDoc);
_tmpScoreDoc = tmp;
}
}
else
{
if (_queueFull)
{
MyScoreDoc tmp = (MyScoreDoc)_bottom;
_currentValueDocMaps.Remove(groupBy.GetFacetData <FacetDataCache>(tmp.reader).OrderArray.Get(tmp.Doc));
_bottom = _currentQueue.Replace(_tmpScoreDoc);
_currentValueDocMaps.Put(order, _tmpScoreDoc);
_tmpScoreDoc = tmp;
}
else
{
ScoreDoc tmp = new MyScoreDoc(doc, score, _currentQueue, _currentReader);
_bottom = _currentQueue.Add(tmp);
_currentValueDocMaps.Put(order, tmp);
_queueFull = (_currentQueue.size >= _numHits);
}
}
}
}
}
}
else
{
if (_count > 0)
{
float score = (_doScoring ? _scorer.Score() : 0.0f);
if (_queueFull)
{
_tmpScoreDoc.Doc = doc;
_tmpScoreDoc.Score = score;
if (_currentComparator.Compare(_bottom, _tmpScoreDoc) > 0)
{
ScoreDoc tmp = _bottom;
_bottom = _currentQueue.Replace(_tmpScoreDoc);
_tmpScoreDoc = tmp;
}
}
else
{
_bottom = _currentQueue.Add(new MyScoreDoc(doc, score, _currentQueue, _currentReader));
_queueFull = (_currentQueue.size >= _numHits);
}
}
}
if (this.Collector != null)
{
this.Collector.Collect(doc);
}
}
protected static BrowseHit[] BuildHits(MyScoreDoc[] scoreDocs, SortField[] sortFields,
IDictionary <string, IFacetHandler> facetHandlerMap, bool fetchStoredFields,
ICollection <string> termVectorsToFetch, IFacetHandler groupBy, CombinedFacetAccessible[] groupAccessibles)
{
BrowseHit[] hits = new BrowseHit[scoreDocs.Length];
IEnumerable <IFacetHandler> facetHandlers = facetHandlerMap.Values;
for (int i = scoreDocs.Length - 1; i >= 0; i--)
{
MyScoreDoc fdoc = scoreDocs[i];
BoboSegmentReader reader = fdoc.m_reader;
BrowseHit hit = new BrowseHit();
if (fetchStoredFields)
{
hit.SetStoredFields(reader.Document(fdoc.Doc));
}
if (termVectorsToFetch != null && termVectorsToFetch.Count > 0)
{
var tvMap = new Dictionary <string, IList <BrowseHit.BoboTerm> >();
hit.TermVectorMap = tvMap;
Fields fds = reader.GetTermVectors(fdoc.Doc);
foreach (string field in termVectorsToFetch)
{
Terms terms = fds.GetTerms(field);
if (terms == null)
{
continue;
}
TermsEnum termsEnum = terms.GetIterator(null);
BytesRef text;
DocsAndPositionsEnum docsAndPositions = null;
List <BrowseHit.BoboTerm> boboTermList = new List <BrowseHit.BoboTerm>();
while ((text = termsEnum.Next()) != null)
{
BrowseHit.BoboTerm boboTerm = new BrowseHit.BoboTerm();
boboTerm.Term = text.Utf8ToString();
boboTerm.Freq = (int)termsEnum.TotalTermFreq;
docsAndPositions = termsEnum.DocsAndPositions(null, docsAndPositions);
if (docsAndPositions != null)
{
docsAndPositions.NextDoc();
boboTerm.Positions = new List <int>();
boboTerm.StartOffsets = new List <int>();
boboTerm.EndOffsets = new List <int>();
for (int t = 0; t < boboTerm.Freq; ++t)
{
boboTerm.Positions.Add(docsAndPositions.NextPosition());
boboTerm.StartOffsets.Add(docsAndPositions.StartOffset);
boboTerm.EndOffsets.Add(docsAndPositions.EndOffset);
}
}
boboTermList.Add(boboTerm);
}
tvMap.Put(field, boboTermList);
}
}
var map = new Dictionary <string, string[]>();
var rawMap = new Dictionary <string, object[]>();
foreach (var facetHandler in facetHandlers)
{
map[facetHandler.Name] = facetHandler.GetFieldValues(reader, fdoc.Doc);
rawMap[facetHandler.Name] = facetHandler.GetRawFieldValues(reader, fdoc.Doc);
}
hit.FieldValues = map;
hit.RawFieldValues = rawMap;
hit.DocId = fdoc.Doc + fdoc.m_queue.m_base;
hit.Score = fdoc.Score;
hit.Comparable = fdoc.Value;
if (groupBy != null)
{
hit.GroupField = groupBy.Name;
hit.GroupValue = hit.GetField(groupBy.Name);
hit.RawGroupValue = hit.GetRawField(groupBy.Name);
if (groupAccessibles != null &&
hit.GroupValue != null &&
groupAccessibles != null &&
groupAccessibles.Length > 0)
{
BrowseFacet facet = groupAccessibles[0].GetFacet(hit.GroupValue);
hit.GroupHitsCount = facet.FacetValueHitCount;
}
}
hits[i] = hit;
}
return(hits);
}
