public virtual void TestPositionReader()
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
TermPositionVector vector;
System.String[] terms;
vector = (TermPositionVector)reader.Get(0, testFields[0]);
Assert.IsTrue(vector != null);
terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
//System.out.println("Term: " + term);
Assert.IsTrue(term.Equals(testTerms[i]));
int[] positions = vector.GetTermPositions(i);
Assert.IsTrue(positions != null);
Assert.IsTrue(positions.Length == this.positions[i].Length);
for (int j = 0; j < positions.Length; j++)
{
int position = positions[j];
Assert.IsTrue(position == this.positions[i][j]);
}
TermVectorOffsetInfo[] offset = vector.GetOffsets(i);
Assert.IsTrue(offset != null);
Assert.IsTrue(offset.Length == this.offsets[i].Length);
for (int j = 0; j < offset.Length; j++)
{
TermVectorOffsetInfo termVectorOffsetInfo = offset[j];
Assert.IsTrue(termVectorOffsetInfo.Equals(offsets[i][j]));
}
}
ITermFreqVector freqVector = reader.Get(0, testFields[1]); //no pos, no offset
Assert.IsTrue(freqVector != null);
Assert.IsTrue(freqVector is TermPositionVector == false);
terms = freqVector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
//System.out.println("Term: " + term);
Assert.IsTrue(term.Equals(testTerms[i]));
}
}
public virtual void TestDocsEnum()
{
TermVectorsReader reader = Codec.Default.TermVectorsFormat().VectorsReader(Dir, Seg.Info, FieldInfos, NewIOContext(Random()));
for (int j = 0; j < 5; j++)
{
Terms vector = reader.Get(j).Terms(TestFields[0]);
Assert.IsNotNull(vector);
Assert.AreEqual(TestTerms.Length, vector.Size());
TermsEnum termsEnum = vector.Iterator(null);
DocsEnum docsEnum = null;
for (int i = 0; i < TestTerms.Length; i++)
{
BytesRef text = termsEnum.Next();
Assert.IsNotNull(text);
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(TestTerms[i], term);
docsEnum = TestUtil.Docs(Random(), termsEnum, null, docsEnum, DocsEnum.FLAG_NONE);
Assert.IsNotNull(docsEnum);
int doc = docsEnum.DocID();
Assert.AreEqual(-1, doc);
Assert.IsTrue(docsEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, docsEnum.NextDoc());
}
Assert.IsNull(termsEnum.Next());
}
reader.Dispose();
}
public virtual void TestDocsEnum()
{
TermVectorsReader reader = Codec.Default.TermVectorsFormat.VectorsReader(dir, seg.Info, fieldInfos, NewIOContext(Random));
for (int j = 0; j < 5; j++)
{
Terms vector = reader.Get(j).GetTerms(testFields[0]);
Assert.IsNotNull(vector);
Assert.AreEqual(testTerms.Length, vector.Count);
TermsEnum termsEnum = vector.GetEnumerator();
DocsEnum docsEnum = null;
for (int i = 0; i < testTerms.Length; i++)
{
Assert.IsTrue(termsEnum.MoveNext());
BytesRef text = termsEnum.Term;
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(testTerms[i], term);
docsEnum = TestUtil.Docs(Random, termsEnum, null, docsEnum, DocsFlags.NONE);
Assert.IsNotNull(docsEnum);
int doc = docsEnum.DocID;
Assert.AreEqual(-1, doc);
Assert.IsTrue(docsEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, docsEnum.NextDoc());
}
Assert.IsFalse(termsEnum.MoveNext());
}
reader.Dispose();
}
public virtual void TestBadParams()
{
try
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
//Bad document number, good Field number
TermFreqVector vector = reader.Get(50, testFields[0]);
Assert.IsTrue(vector == null);
}
catch (System.Exception e)
{
Assert.IsTrue(false);
}
try
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
//good document number, bad Field number
TermFreqVector vector = reader.Get(0, "f50");
Assert.IsTrue(vector == null);
}
catch (System.Exception e)
{
Assert.IsTrue(false);
}
}
public override Fields GetTermVectors(int docID)
{
TermVectorsReader termVectorsReader = TermVectorsReader;
if (termVectorsReader == null)
{
return null;
}
CheckBounds(docID);
return termVectorsReader.Get(docID);
}
/// <summary>Return an array of term frequency vectors for the specified document.
/// The array contains a vector for each vectorized field in the document.
/// Each vector vector contains term numbers and frequencies for all terms
/// in a given vectorized field.
/// If no such fields existed, the method returns null.
/// </summary>
/// <throws> IOException </throws>
public override TermFreqVector[] GetTermFreqVectors(int docNumber)
{
if (termVectorsReaderOrig == null)
return null;
TermVectorsReader termVectorsReader = GetTermVectorsReader();
if (termVectorsReader == null)
return null;
return termVectorsReader.Get(docNumber);
}
/// <summary>Return a term frequency vector for the specified document and field. The
/// vector returned contains term numbers and frequencies for all terms in
/// the specified field of this document, if the field had storeTermVector
/// flag set. If the flag was not set, the method returns null.
/// </summary>
/// <throws> IOException </throws>
public override TermFreqVector GetTermFreqVector(int docNumber, System.String field)
{
// Check if this field is invalid or has no stored term vector
FieldInfo fi = fieldInfos.FieldInfo(field);
if (fi == null || !fi.storeTermVector || termVectorsReaderOrig == null)
return null;
TermVectorsReader termVectorsReader = GetTermVectorsReader();
if (termVectorsReader == null)
return null;
return termVectorsReader.Get(docNumber, field);
}
private void CheckTermVector(TermVectorsReader reader, int docNum, System.String field)
{
TermFreqVector vector = reader.Get(docNum, field);
Assert.IsTrue(vector != null);
System.String[] terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
Assert.IsTrue(term.Equals(testTerms[i]));
}
}
/// <summary>Return an array of term frequency vectors for the specified document.
/// The array contains a vector for each vectorized field in the document.
/// Each vector vector contains term numbers and frequencies for all terms
/// in a given vectorized field.
/// If no such fields existed, the method returns null.
/// </summary>
/// <throws> IOException </throws>
public override TermFreqVector[] GetTermFreqVectors(int docNumber)
{
if (termVectorsReaderOrig == null)
{
return(null);
}
TermVectorsReader termVectorsReader = GetTermVectorsReader();
if (termVectorsReader == null)
{
return(null);
}
return(termVectorsReader.Get(docNumber));
}
public override void GetTermFreqVector(int docNumber, TermVectorMapper mapper)
{
EnsureOpen();
if (termVectorsReaderOrig == null)
{
return;
}
TermVectorsReader termVectorsReader = GetTermVectorsReader();
if (termVectorsReader == null)
{
return;
}
termVectorsReader.Get(docNumber, mapper);
}
public virtual void TestOffsetReader()
{
TermVectorsReader reader = Codec.Default.TermVectorsFormat().VectorsReader(Dir, Seg.Info, FieldInfos, NewIOContext(Random()));
Terms vector = reader.Get(0).Terms(TestFields[0]);
Assert.IsNotNull(vector);
TermsEnum termsEnum = vector.Iterator(null);
Assert.IsNotNull(termsEnum);
Assert.AreEqual(TestTerms.Length, vector.Size());
DocsAndPositionsEnum dpEnum = null;
for (int i = 0; i < TestTerms.Length; i++)
{
BytesRef text = termsEnum.Next();
Assert.IsNotNull(text);
string term = text.Utf8ToString();
Assert.AreEqual(TestTerms[i], term);
dpEnum = termsEnum.DocsAndPositions(null, dpEnum);
Assert.IsNotNull(dpEnum);
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.AreEqual(dpEnum.Freq(), Positions[i].Length);
for (int j = 0; j < Positions[i].Length; j++)
{
Assert.AreEqual(Positions[i][j], dpEnum.NextPosition());
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, dpEnum.NextDoc());
dpEnum = termsEnum.DocsAndPositions(null, dpEnum);
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.IsNotNull(dpEnum);
Assert.AreEqual(dpEnum.Freq(), Positions[i].Length);
for (int j = 0; j < Positions[i].Length; j++)
{
Assert.AreEqual(Positions[i][j], dpEnum.NextPosition());
Assert.AreEqual(j * 10, dpEnum.StartOffset());
Assert.AreEqual(j * 10 + TestTerms[i].Length, dpEnum.EndOffset());
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, dpEnum.NextDoc());
}
reader.Dispose();
}
public override void GetTermFreqVector(int docNumber, System.String field, TermVectorMapper mapper)
{
EnsureOpen();
FieldInfo fi = fieldInfos.FieldInfo(field);
if (fi == null || !fi.storeTermVector || termVectorsReaderOrig == null)
{
return;
}
TermVectorsReader termVectorsReader = GetTermVectorsReader();
if (termVectorsReader == null)
{
return;
}
termVectorsReader.Get(docNumber, field, mapper);
}
public virtual void TestReader()
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
for (int j = 0; j < 5; j++)
{
ITermFreqVector vector = reader.Get(j, testFields[0]);
Assert.IsTrue(vector != null);
System.String[] terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
//System.out.println("Term: " + term);
Assert.IsTrue(term.Equals(testTerms[i]));
}
}
}
public virtual void TestReader()
{
TermVectorsReader reader = Codec.Default.TermVectorsFormat().VectorsReader(Dir, Seg.Info, FieldInfos, NewIOContext(Random()));
for (int j = 0; j < 5; j++)
{
Terms vector = reader.Get(j).Terms(TestFields[0]);
Assert.IsNotNull(vector);
Assert.AreEqual(TestTerms.Length, vector.Size());
TermsEnum termsEnum = vector.Iterator(null);
for (int i = 0; i < TestTerms.Length; i++)
{
BytesRef text = termsEnum.Next();
Assert.IsNotNull(text);
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(TestTerms[i], term);
}
Assert.IsNull(termsEnum.Next());
}
reader.Dispose();
}
public virtual void TestReader()
{
TermVectorsReader reader = Codec.Default.TermVectorsFormat.VectorsReader(dir, seg.Info, fieldInfos, NewIOContext(Random));
for (int j = 0; j < 5; j++)
{
Terms vector = reader.Get(j).GetTerms(testFields[0]);
Assert.IsNotNull(vector);
Assert.AreEqual(testTerms.Length, vector.Count);
TermsEnum termsEnum = vector.GetEnumerator();
for (int i = 0; i < testTerms.Length; i++)
{
Assert.IsTrue(termsEnum.MoveNext());
BytesRef text = termsEnum.Term;
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(testTerms[i], term);
}
Assert.IsFalse(termsEnum.MoveNext());
}
reader.Dispose();
}
public virtual void TestBadParams()
{
var reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
//Bad document number, good field number
Assert.Throws <System.IO.IOException>(() => reader.Get(50, testFields[0]));
reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
//Bad document number, no field
Assert.Throws <System.IO.IOException>(() => reader.Get(50));
reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
Assert.DoesNotThrow(() =>
{
//good document number, bad field number
ITermFreqVector vector = reader.Get(0, "f50");
Assert.IsTrue(vector == null);
});
}
public virtual void TestReader()
{
try
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
TermFreqVector vector = reader.Get(0, testFields[0]);
Assert.IsTrue(vector != null);
System.String[] terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
//System.out.println("Term: " + term);
Assert.IsTrue(term.Equals(testTerms[i]));
}
}
catch (System.IO.IOException e)
{
System.Console.Error.WriteLine(e.StackTrace);
Assert.IsTrue(false);
}
}
public virtual void TestOffsetReader()
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
TermPositionVector vector = (TermPositionVector) reader.Get(0, testFields[0]);
Assert.IsTrue(vector != null);
System.String[] terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
//System.out.println("Term: " + term);
Assert.IsTrue(term.Equals(testTerms[i]));
int[] positions = vector.GetTermPositions(i);
Assert.IsTrue(positions != null);
Assert.IsTrue(positions.Length == this.positions[i].Length);
for (int j = 0; j < positions.Length; j++)
{
int position = positions[j];
Assert.IsTrue(position == this.positions[i][j]);
}
TermVectorOffsetInfo[] offset = vector.GetOffsets(i);
Assert.IsTrue(offset != null);
Assert.IsTrue(offset.Length == this.offsets[i].Length);
for (int j = 0; j < offset.Length; j++)
{
TermVectorOffsetInfo termVectorOffsetInfo = offset[j];
Assert.IsTrue(termVectorOffsetInfo.Equals(offsets[i][j]));
}
}
}
public virtual void TestReader()
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
for (int j = 0; j < 5; j++)
{
TermFreqVector vector = reader.Get(j, testFields[0]);
Assert.IsTrue(vector != null);
System.String[] terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
//System.out.println("Term: " + term);
Assert.IsTrue(term.Equals(testTerms[i]));
}
}
}
public virtual void TestPositionReader()
{
TermVectorsReader reader = Codec.Default.TermVectorsFormat.VectorsReader(dir, seg.Info, fieldInfos, NewIOContext(Random));
//BytesRef[] terms; // LUCENENET NOTE: Not used in Lucene
Terms vector = reader.Get(0).GetTerms(testFields[0]);
Assert.IsNotNull(vector);
Assert.AreEqual(testTerms.Length, vector.Count);
TermsEnum termsEnum = vector.GetEnumerator();
DocsAndPositionsEnum dpEnum = null;
for (int i = 0; i < testTerms.Length; i++)
{
Assert.IsTrue(termsEnum.MoveNext());
BytesRef text = termsEnum.Term;
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(testTerms[i], term);
dpEnum = termsEnum.DocsAndPositions(null, dpEnum);
Assert.IsNotNull(dpEnum);
int doc = dpEnum.DocID;
Assert.AreEqual(-1, doc);
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.AreEqual(dpEnum.Freq, positions[i].Length);
for (int j = 0; j < positions[i].Length; j++)
{
Assert.AreEqual(positions[i][j], dpEnum.NextPosition());
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, dpEnum.NextDoc());
dpEnum = termsEnum.DocsAndPositions(null, dpEnum);
doc = dpEnum.DocID;
Assert.AreEqual(-1, doc);
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.IsNotNull(dpEnum);
Assert.AreEqual(dpEnum.Freq, positions[i].Length);
for (int j = 0; j < positions[i].Length; j++)
{
Assert.AreEqual(positions[i][j], dpEnum.NextPosition());
Assert.AreEqual(j * 10, dpEnum.StartOffset);
Assert.AreEqual(j * 10 + testTerms[i].Length, dpEnum.EndOffset);
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, dpEnum.NextDoc());
}
Terms freqVector = reader.Get(0).GetTerms(testFields[1]); //no pos, no offset
Assert.IsNotNull(freqVector);
Assert.AreEqual(testTerms.Length, freqVector.Count);
termsEnum = freqVector.GetEnumerator();
Assert.IsNotNull(termsEnum);
for (int i = 0; i < testTerms.Length; i++)
{
Assert.IsTrue(termsEnum.MoveNext());
BytesRef text = termsEnum.Term;
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(testTerms[i], term);
Assert.IsNotNull(termsEnum.Docs(null, null));
Assert.IsNull(termsEnum.DocsAndPositions(null, null)); // no pos
}
reader.Dispose();
}
public virtual void TestMapper()
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
SortedTermVectorMapper mapper = new SortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
reader.Get(0, mapper);
SortedSet<TermVectorEntry> set_Renamed = mapper.GetTermVectorEntrySet();
Assert.IsTrue(set_Renamed != null, "set is null and it shouldn't be");
//three fields, 4 terms, all terms are the same
Assert.IsTrue(set_Renamed.Count == 4, "set Size: " + set_Renamed.Count + " is not: " + 4);
//Check offsets and positions
foreach(TermVectorEntry tve in set_Renamed)
{
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
Assert.IsTrue(tve.GetOffsets() != null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() != null, "tve.getPositions() is null and it shouldn't be");
}
mapper = new SortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
reader.Get(1, mapper);
set_Renamed = mapper.GetTermVectorEntrySet();
Assert.IsTrue(set_Renamed != null, "set is null and it shouldn't be");
//three fields, 4 terms, all terms are the same
Assert.IsTrue(set_Renamed.Count == 4, "set Size: " + set_Renamed.Count + " is not: " + 4);
//Should have offsets and positions b/c we are munging all the fields together
foreach(TermVectorEntry tve in set_Renamed)
{
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
Assert.IsTrue(tve.GetOffsets() != null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() != null, "tve.getPositions() is null and it shouldn't be");
}
FieldSortedTermVectorMapper fsMapper = new FieldSortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
reader.Get(0, fsMapper);
IDictionary<string, SortedSet<TermVectorEntry>> map = fsMapper.GetFieldToTerms();
Assert.IsTrue(map.Count == testFields.Length, "map Size: " + map.Count + " is not: " + testFields.Length);
foreach(KeyValuePair<string,SortedSet<TermVectorEntry>> entry in new Dictionary<string, SortedSet<TermVectorEntry>>(map))
{
SortedSet<TermVectorEntry> sortedSet = entry.Value;
Assert.IsTrue(sortedSet.Count == 4, "sortedSet Size: " + sortedSet.Count + " is not: " + 4);
foreach(TermVectorEntry tve in sortedSet)
{
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
//Check offsets and positions.
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
System.String field = tve.GetField();
if (field.Equals(testFields[0]))
{
//should have offsets
Assert.IsTrue(tve.GetOffsets() != null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() != null, "tve.getPositions() is null and it shouldn't be");
}
else if (field.Equals(testFields[1]))
{
//should not have offsets
Assert.IsTrue(tve.GetOffsets() == null, "tve.getOffsets() is not null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() == null, "tve.getPositions() is not null and it shouldn't be");
}
}
}
//Try mapper that ignores offs and positions
fsMapper = new FieldSortedTermVectorMapper(true, true, new TermVectorEntryFreqSortedComparator());
reader.Get(0, fsMapper);
map = fsMapper.GetFieldToTerms();
Assert.IsTrue(map.Count == testFields.Length, "map Size: " + map.Count + " is not: " + testFields.Length);
foreach(KeyValuePair<string,SortedSet<TermVectorEntry>> entry in new Dictionary<string,SortedSet<TermVectorEntry>>(map))
{
SortedSet<TermVectorEntry> sortedSet = entry.Value;
Assert.IsTrue(sortedSet.Count == 4, "sortedSet Size: " + sortedSet.Count + " is not: " + 4);
foreach(TermVectorEntry tve in sortedSet)
{
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
//Check offsets and positions.
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
System.String field = tve.GetField();
if (field.Equals(testFields[0]))
{
//should have offsets
Assert.IsTrue(tve.GetOffsets() == null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() == null, "tve.getPositions() is null and it shouldn't be");
}
else if (field.Equals(testFields[1]))
{
//should not have offsets
Assert.IsTrue(tve.GetOffsets() == null, "tve.getOffsets() is not null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() == null, "tve.getPositions() is not null and it shouldn't be");
}
}
}
// test setDocumentNumber()
IndexReader ir = IndexReader.Open(dir);
DocNumAwareMapper docNumAwareMapper = new DocNumAwareMapper();
Assert.AreEqual(- 1, docNumAwareMapper.GetDocumentNumber());
ir.GetTermFreqVector(0, docNumAwareMapper);
Assert.AreEqual(0, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(- 1);
ir.GetTermFreqVector(1, docNumAwareMapper);
Assert.AreEqual(1, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(- 1);
ir.GetTermFreqVector(0, "f1", docNumAwareMapper);
Assert.AreEqual(0, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(- 1);
ir.GetTermFreqVector(1, "f2", docNumAwareMapper);
Assert.AreEqual(1, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(- 1);
ir.GetTermFreqVector(0, "f1", docNumAwareMapper);
Assert.AreEqual(0, docNumAwareMapper.GetDocumentNumber());
ir.Close();
}
private void CheckTermVector(TermVectorsReader reader, int docNum, System.String field)
{
TermFreqVector vector = reader.Get(docNum, field);
Assert.IsTrue(vector != null);
System.String[] terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
Assert.IsTrue(term.Equals(testTerms[i]));
}
}
public virtual void TestMapper()
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
SortedTermVectorMapper mapper = new SortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
reader.Get(0, mapper);
var set_Renamed = mapper.TermVectorEntrySet;
Assert.IsTrue(set_Renamed != null, "set is null and it shouldn't be");
//three fields, 4 terms, all terms are the same
Assert.IsTrue(set_Renamed.Count == 4, "set Size: " + set_Renamed.Count + " is not: " + 4);
//Check offsets and positions
for (System.Collections.IEnumerator iterator = set_Renamed.GetEnumerator(); iterator.MoveNext();)
{
TermVectorEntry tve = (TermVectorEntry)iterator.Current;
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
Assert.IsTrue(tve.GetOffsets() != null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() != null, "tve.getPositions() is null and it shouldn't be");
}
mapper = new SortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
reader.Get(1, mapper);
set_Renamed = mapper.TermVectorEntrySet;
Assert.IsTrue(set_Renamed != null, "set is null and it shouldn't be");
//three fields, 4 terms, all terms are the same
Assert.IsTrue(set_Renamed.Count == 4, "set Size: " + set_Renamed.Count + " is not: " + 4);
//Should have offsets and positions b/c we are munging all the fields together
for (System.Collections.IEnumerator iterator = set_Renamed.GetEnumerator(); iterator.MoveNext();)
{
TermVectorEntry tve = (TermVectorEntry)iterator.Current;
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
Assert.IsTrue(tve.GetOffsets() != null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() != null, "tve.getPositions() is null and it shouldn't be");
}
FieldSortedTermVectorMapper fsMapper = new FieldSortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
reader.Get(0, fsMapper);
var map = fsMapper.FieldToTerms;
Assert.IsTrue(map.Count == testFields.Length, "map Size: " + map.Count + " is not: " + testFields.Length);
for (var iterator = map.GetEnumerator(); iterator.MoveNext();)
{
var entry = iterator.Current;
var sortedSet = entry.Value;
Assert.IsTrue(sortedSet.Count == 4, "sortedSet Size: " + sortedSet.Count + " is not: " + 4);
for (var inner = sortedSet.GetEnumerator(); inner.MoveNext();)
{
TermVectorEntry tve = inner.Current;
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
//Check offsets and positions.
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
System.String field = tve.Field;
if (field.Equals(testFields[0]))
{
//should have offsets
Assert.IsTrue(tve.GetOffsets() != null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() != null, "tve.getPositions() is null and it shouldn't be");
}
else if (field.Equals(testFields[1]))
{
//should not have offsets
Assert.IsTrue(tve.GetOffsets() == null, "tve.getOffsets() is not null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() == null, "tve.getPositions() is not null and it shouldn't be");
}
}
}
//Try mapper that ignores offs and positions
fsMapper = new FieldSortedTermVectorMapper(true, true, new TermVectorEntryFreqSortedComparator());
reader.Get(0, fsMapper);
map = fsMapper.FieldToTerms;
Assert.IsTrue(map.Count == testFields.Length, "map Size: " + map.Count + " is not: " + testFields.Length);
for (var iterator = map.GetEnumerator(); iterator.MoveNext();)
{
var entry = iterator.Current;
var sortedSet = entry.Value;
Assert.IsTrue(sortedSet.Count == 4, "sortedSet Size: " + sortedSet.Count + " is not: " + 4);
for (var inner = sortedSet.GetEnumerator(); inner.MoveNext();)
{
TermVectorEntry tve = inner.Current;
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
//Check offsets and positions.
Assert.IsTrue(tve != null, "tve is null and it shouldn't be");
System.String field = tve.Field;
if (field.Equals(testFields[0]))
{
//should have offsets
Assert.IsTrue(tve.GetOffsets() == null, "tve.getOffsets() is null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() == null, "tve.getPositions() is null and it shouldn't be");
}
else if (field.Equals(testFields[1]))
{
//should not have offsets
Assert.IsTrue(tve.GetOffsets() == null, "tve.getOffsets() is not null and it shouldn't be");
Assert.IsTrue(tve.GetPositions() == null, "tve.getPositions() is not null and it shouldn't be");
}
}
}
// test setDocumentNumber()
IndexReader ir = IndexReader.Open(dir, true);
DocNumAwareMapper docNumAwareMapper = new DocNumAwareMapper();
Assert.AreEqual(-1, docNumAwareMapper.GetDocumentNumber());
ir.GetTermFreqVector(0, docNumAwareMapper);
Assert.AreEqual(0, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(-1);
ir.GetTermFreqVector(1, docNumAwareMapper);
Assert.AreEqual(1, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(-1);
ir.GetTermFreqVector(0, "f1", docNumAwareMapper);
Assert.AreEqual(0, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(-1);
ir.GetTermFreqVector(1, "f2", docNumAwareMapper);
Assert.AreEqual(1, docNumAwareMapper.GetDocumentNumber());
docNumAwareMapper.SetDocumentNumber(-1);
ir.GetTermFreqVector(0, "f1", docNumAwareMapper);
Assert.AreEqual(0, docNumAwareMapper.GetDocumentNumber());
ir.Close();
}
public virtual void TestBadParams()
{
try
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
//Bad document number, good field number
reader.Get(50, testFields[0]);
Assert.Fail();
}
catch (System.IO.IOException e)
{
// expected exception
}
try
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
//Bad document number, no field
reader.Get(50);
Assert.Fail();
}
catch (System.IO.IOException e)
{
// expected exception
}
try
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
//good document number, bad field number
TermFreqVector vector = reader.Get(0, "f50");
Assert.IsTrue(vector == null);
}
catch (System.IO.IOException e)
{
Assert.Fail();
}
}
public virtual void TestReader()
{
try
{
TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
Assert.IsTrue(reader != null);
TermFreqVector vector = reader.Get(0, testFields[0]);
Assert.IsTrue(vector != null);
System.String[] terms = vector.GetTerms();
Assert.IsTrue(terms != null);
Assert.IsTrue(terms.Length == testTerms.Length);
for (int i = 0; i < terms.Length; i++)
{
System.String term = terms[i];
//System.out.println("Term: " + term);
Assert.IsTrue(term.Equals(testTerms[i]));
}
}
catch (System.IO.IOException e)
{
System.Console.Error.WriteLine(e.StackTrace);
Assert.IsTrue(false);
}
}