public override FacetDataCache Load(BoboIndexReader reader, BoboIndexReader.WorkArea workArea)
{
MultiValueFacetDataCache dataCache = new MultiValueFacetDataCache();
dataCache.MaxItems = maxItems;
if (sizePayloadTerm == null)
{
dataCache.Load(_indexFieldName, reader, _termListFactory, workArea);
}
else
{
dataCache.Load(_indexFieldName, reader, _termListFactory, sizePayloadTerm);
}
return(dataCache);
}
/// <summary>
/// loads multi-value facet data. This method uses a workarea to prepare loading.
/// </summary>
/// <param name="fieldName"></param>
/// <param name="reader"></param>
/// <param name="listFactory"></param>
/// <param name="workArea"></param>
public virtual void Load(string fieldName, IndexReader reader, TermListFactory listFactory, BoboIndexReader.WorkArea workArea)
{
long t0 = Environment.TickCount;
int maxdoc = reader.MaxDoc;
BigNestedIntArray.BufferedLoader loader = GetBufferedLoader(maxdoc, workArea);
TermEnum tenum = null;
TermDocs tdoc = null;
ITermValueList list = (listFactory == null ? (ITermValueList) new TermStringList() : listFactory.CreateTermList());
List <int> minIDList = new List <int>();
List <int> maxIDList = new List <int>();
List <int> freqList = new List <int>();
OpenBitSet bitset = new OpenBitSet();
int negativeValueCount = GetNegativeValueCount(reader, string.Intern(fieldName));
int t = 0; // current term number
list.Add(null);
minIDList.Add(-1);
maxIDList.Add(-1);
freqList.Add(0);
t++;
_overflow = false;
try
{
tdoc = reader.TermDocs();
tenum = reader.Terms(new Term(fieldName, ""));
if (tenum != null)
{
do
{
Term term = tenum.Term;
if (term == null || !fieldName.Equals(term.Field))
{
break;
}
string val = term.Text;
if (val != null)
{
list.Add(val);
tdoc.Seek(tenum);
//freqList.add(tenum.docFreq()); // removed because the df doesn't take into account the num of deletedDocs
int df = 0;
int minID = -1;
int maxID = -1;
int valId = (t - 1 < negativeValueCount) ? (negativeValueCount - t + 1) : t;
if (tdoc.Next())
{
df++;
int docid = tdoc.Doc;
if (!loader.Add(docid, valId))
{
LogOverflow(fieldName);
}
minID = docid;
bitset.Set(docid);
while (tdoc.Next())
{
df++;
docid = tdoc.Doc;
if (!loader.Add(docid, valId))
{
LogOverflow(fieldName);
}
bitset.Set(docid);
}
maxID = docid;
}
freqList.Add(df);
minIDList.Add(minID);
maxIDList.Add(maxID);
}
t++;
}while (tenum.Next());
}
}
finally
{
try
{
if (tdoc != null)
{
tdoc.Dispose();
}
}
finally
{
if (tenum != null)
{
tenum.Dispose();
}
}
}
list.Seal();
try
{
_nestedArray.Load(maxdoc + 1, loader);
}
catch (System.IO.IOException e)
{
throw e;
}
catch (Exception e)
{
throw new RuntimeException("failed to load due to " + e.ToString(), e);
}
this.valArray = list;
this.freqs = freqList.ToArray();
this.minIDs = minIDList.ToArray();
this.maxIDs = maxIDList.ToArray();
int doc = 0;
while (doc <= maxdoc && !_nestedArray.Contains(doc, 0, true))
{
++doc;
}
if (doc <= maxdoc)
{
this.minIDs[0] = doc;
doc = maxdoc;
while (doc > 0 && !_nestedArray.Contains(doc, 0, true))
{
--doc;
}
if (doc > 0)
{
this.maxIDs[0] = doc;
}
}
this.freqs[0] = maxdoc + 1 - (int)bitset.Cardinality();
}
protected virtual BigNestedIntArray.BufferedLoader GetBufferedLoader(int maxdoc, BoboIndexReader.WorkArea workArea)
{
if (workArea == null)
{
return(new BigNestedIntArray.BufferedLoader(maxdoc, _maxItems, new BigIntBuffer()));
}
else
{
BigIntBuffer buffer = workArea.Get <BigIntBuffer>();
if (buffer == null)
{
buffer = new BigIntBuffer();
workArea.Put(buffer);
}
else
{
buffer.Reset();
}
BigNestedIntArray.BufferedLoader loader = workArea.Get <BigNestedIntArray.BufferedLoader>();
if (loader == null || loader.Capacity < maxdoc)
{
loader = new BigNestedIntArray.BufferedLoader(maxdoc, _maxItems, buffer);
workArea.Put(loader);
}
else
{
loader.Reset(maxdoc, _maxItems, buffer);
}
return(loader);
}
}
public override void Load(string fieldName, IndexReader reader, TermListFactory listFactory, BoboIndexReader.WorkArea workArea)
{
long t0 = System.Environment.TickCount;
int maxdoc = reader.MaxDoc;
BigNestedIntArray.BufferedLoader loader = GetBufferedLoader(maxdoc, workArea);
BigNestedIntArray.BufferedLoader weightLoader = GetBufferedLoader(maxdoc, null);
TermEnum tenum = null;
TermDocs tdoc = null;
var list = (listFactory == null ? new TermStringList() : listFactory.CreateTermList());
List <int> minIDList = new List <int>();
List <int> maxIDList = new List <int>();
List <int> freqList = new List <int>();
OpenBitSet bitset = new OpenBitSet(maxdoc + 1);
int negativeValueCount = GetNegativeValueCount(reader, string.Intern(fieldName));
int t = 0; // current term number
list.Add(null);
minIDList.Add(-1);
maxIDList.Add(-1);
freqList.Add(0);
t++;
_overflow = false;
string pre = null;
int df = 0;
int minID = -1;
int maxID = -1;
int valId = 0;
try
{
tdoc = reader.TermDocs();
tenum = reader.Terms(new Term(fieldName, ""));
if (tenum != null)
{
do
{
Term term = tenum.Term;
if (term == null || !fieldName.Equals(term.Field))
{
break;
}
string val = term.Text;
if (val != null)
{
int weight = 0;
string[] split = val.Split(new char[] { '\0' }, StringSplitOptions.RemoveEmptyEntries);
if (split.Length > 1)
{
val = split[0];
weight = int.Parse(split[split.Length - 1]);
}
if (pre == null || !val.Equals(pre))
{
if (pre != null)
{
freqList.Add(df);
minIDList.Add(minID);
maxIDList.Add(maxID);
}
list.Add(val);
df = 0;
minID = -1;
maxID = -1;
valId = (t - 1 < negativeValueCount) ? (negativeValueCount - t + 1) : t;
t++;
}
tdoc.Seek(tenum);
if (tdoc.Next())
{
df++;
int docid = tdoc.Doc;
if (!loader.Add(docid, valId))
{
LogOverflow(fieldName);
}
else
{
weightLoader.Add(docid, weight);
}
if (docid < minID)
{
minID = docid;
}
bitset.FastSet(docid);
while (tdoc.Next())
{
df++;
docid = tdoc.Doc;
if (!loader.Add(docid, valId))
{
LogOverflow(fieldName);
}
else
{
weightLoader.Add(docid, weight);
}
bitset.FastSet(docid);
}
if (docid > maxID)
{
maxID = docid;
}
}
pre = val;
}
}while (tenum.Next());
if (pre != null)
{
freqList.Add(df);
minIDList.Add(minID);
maxIDList.Add(maxID);
}
}
}
finally
{
try
{
if (tdoc != null)
{
tdoc.Dispose();
}
}
finally
{
if (tenum != null)
{
tenum.Dispose();
}
}
}
list.Seal();
try
{
_nestedArray.Load(maxdoc + 1, loader);
_weightArray.Load(maxdoc + 1, weightLoader);
}
catch (System.IO.IOException e)
{
throw e;
}
catch (Exception e)
{
throw new RuntimeException("failed to load due to " + e.ToString(), e);
}
this.valArray = list;
this.freqs = freqList.ToArray();
this.minIDs = minIDList.ToArray();
this.maxIDs = maxIDList.ToArray();
int doc = 0;
while (doc <= maxdoc && !_nestedArray.Contains(doc, 0, true))
{
++doc;
}
if (doc <= maxdoc)
{
this.minIDs[0] = doc;
doc = maxdoc;
while (doc > 0 && !_nestedArray.Contains(doc, 0, true))
{
--doc;
}
if (doc > 0)
{
this.maxIDs[0] = doc;
}
}
this.freqs[0] = maxdoc + 1 - (int)bitset.Cardinality();
}
public virtual void LoadFacetData(BoboIndexReader reader, BoboIndexReader.WorkArea workArea)
{
reader.PutFacetData(_name, Load(reader, workArea));
}
public virtual D Load(BoboIndexReader reader, BoboIndexReader.WorkArea workArea)
{
return(Load(reader));
}
public void TestIndexReload()
{
try
{
RAMDirectory idxDir = new RAMDirectory();
Document[] docs = BoboTestCase.BuildData();
BoboIndexReader.WorkArea workArea = new BoboIndexReader.WorkArea();
BrowseRequest req;
BrowseSelection sel;
BoboBrowser browser;
BrowseResult result;
IndexWriter writer = new IndexWriter(idxDir, new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_CURRENT), true, IndexWriter.MaxFieldLength.UNLIMITED);
writer.Close();
int dup = 0;
for (int j = 0; j < 50; j++)
{
IndexReader idxReader = IndexReader.Open(idxDir, true);
BoboIndexReader reader = BoboIndexReader.GetInstance(idxReader, _fconf, workArea);
req = new BrowseRequest();
req.Offset = 0;
req.Count = 10;
sel = new BrowseSelection("color");
sel.AddValue("red");
req.AddSelection(sel);
browser = new BoboBrowser(reader);
result = browser.Browse(req);
Assert.AreEqual(3 * dup, result.NumHits);
req = new BrowseRequest();
req.Offset = 0;
req.Count = 10;
sel = new BrowseSelection("tag");
sel.AddValue("dog");
req.AddSelection(sel);
browser = new BoboBrowser(reader);
result = browser.Browse(req);
Assert.AreEqual(2 * dup, result.NumHits);
req = new BrowseRequest();
req.Offset = 0;
req.Count = 10;
sel = new BrowseSelection("tag");
sel.AddValue("funny");
req.AddSelection(sel);
browser = new BoboBrowser(reader);
result = browser.Browse(req);
Assert.AreEqual(3 * dup, result.NumHits);
writer = new IndexWriter(idxDir, new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_CURRENT), false, IndexWriter.MaxFieldLength.UNLIMITED);
for (int k = 0; k <= j; k++)
{
for (int i = 0; i < docs.Length; i++)
{
writer.AddDocument(docs[i]);
}
dup++;
}
writer.Close();
}
idxDir.Close();
}
catch (Exception e)
{
Assert.Fail(e.Message);
}
}
public IEnumerable <IFacetHandler> LoadFacetHandlers(string springConfigFile, BoboIndexReader.WorkArea workArea)
{
if (File.Exists(springConfigFile))
{
XmlApplicationContext appCtx = new XmlApplicationContext(springConfigFile);
return(appCtx.GetObjectsOfType(typeof(IFacetHandler)).Values.OfType <IFacetHandler>().ToList());
}
else
{
return(new List <IFacetHandler>());
}
}
