internal virtual void SeekEnum(SegmentTermEnum enumerator, int indexOffset)
{
PagedBytesDataInput input = (PagedBytesDataInput)dataInput.Clone();
input.SetPosition(indexToDataOffset.Get(indexOffset));
// read the term
int fieldId = input.ReadVInt32();
Term field = fields[fieldId];
Term term = new Term(field.Field, input.ReadString());
// read the terminfo
var termInfo = new TermInfo();
termInfo.DocFreq = input.ReadVInt32();
if (termInfo.DocFreq >= skipInterval)
{
termInfo.SkipOffset = input.ReadVInt32();
}
else
{
termInfo.SkipOffset = 0;
}
termInfo.FreqPointer = input.ReadVInt64();
termInfo.ProxPointer = input.ReadVInt64();
long pointer = input.ReadVInt64();
// perform the seek
enumerator.Seek(pointer, ((long)indexOffset * totalIndexInterval) - 1, term, termInfo);
}
/// <summary>
/// Binary search for the given term.
/// </summary>
/// <param name="term">
/// The term to locate. </param>
/// <exception cref="IOException"> If there is a low-level I/O error. </exception>
internal virtual int GetIndexOffset(Term term)
{
int lo = 0;
int hi = indexSize - 1;
PagedBytesDataInput input = (PagedBytesDataInput)dataInput.Clone();
BytesRef scratch = new BytesRef();
while (hi >= lo)
{
int mid = (int)((uint)(lo + hi) >> 1);
int delta = CompareTo(term, mid, input, scratch);
if (delta < 0)
{
hi = mid - 1;
}
else if (delta > 0)
{
lo = mid + 1;
}
else
{
return(mid);
}
}
return(hi);
}
public override object Clone()
{
PagedBytesDataInput clone = outerInstance.GetDataInput();
clone.SetPosition(GetPosition());
return(clone);
}
public override object Clone()
{
PagedBytesDataInput clone = OuterInstance.DataInput;
clone.Position = Position;
return(clone);
}
/// <summary>
/// Loads the segment information at segment load time.
/// </summary>
/// <param name="indexEnum">
/// The term enum. </param>
/// <param name="indexDivisor">
/// The index divisor. </param>
/// <param name="tiiFileLength">
/// The size of the tii file, used to approximate the size of the
/// buffer. </param>
/// <param name="totalIndexInterval">
/// The total index interval. </param>
public TermInfosReaderIndex(SegmentTermEnum indexEnum, int indexDivisor, long tiiFileLength, int totalIndexInterval)
{
this.totalIndexInterval = totalIndexInterval;
indexSize = 1 + ((int)indexEnum.size - 1) / indexDivisor;
skipInterval = indexEnum.skipInterval;
// this is only an inital size, it will be GCed once the build is complete
long initialSize = (long)(tiiFileLength * 1.5) / indexDivisor;
PagedBytes dataPagedBytes = new PagedBytes(EstimatePageBits(initialSize));
PagedBytesDataOutput dataOutput = dataPagedBytes.GetDataOutput();
int bitEstimate = 1 + MathUtil.Log(tiiFileLength, 2);
GrowableWriter indexToTerms = new GrowableWriter(bitEstimate, indexSize, PackedInt32s.DEFAULT);
string currentField = null;
IList <string> fieldStrs = new List <string>();
int fieldCounter = -1;
for (int i = 0; indexEnum.Next(); i++)
{
Term term = indexEnum.Term();
if (currentField == null || !currentField.Equals(term.Field, StringComparison.Ordinal))
{
currentField = term.Field;
fieldStrs.Add(currentField);
fieldCounter++;
}
TermInfo termInfo = indexEnum.TermInfo();
indexToTerms.Set(i, dataOutput.GetPosition());
dataOutput.WriteVInt32(fieldCounter);
dataOutput.WriteString(term.Text());
dataOutput.WriteVInt32(termInfo.DocFreq);
if (termInfo.DocFreq >= skipInterval)
{
dataOutput.WriteVInt32(termInfo.SkipOffset);
}
dataOutput.WriteVInt64(termInfo.FreqPointer);
dataOutput.WriteVInt64(termInfo.ProxPointer);
dataOutput.WriteVInt64(indexEnum.indexPointer);
for (int j = 1; j < indexDivisor; j++)
{
if (!indexEnum.Next())
{
break;
}
}
}
fields = new Term[fieldStrs.Count];
for (int i = 0; i < fields.Length; i++)
{
fields[i] = new Term(fieldStrs[i]);
}
dataPagedBytes.Freeze(true);
dataInput = dataPagedBytes.GetDataInput();
indexToDataOffset = indexToTerms.Mutable;
ramBytesUsed = fields.Length * (RamUsageEstimator.NUM_BYTES_OBJECT_REF + RamUsageEstimator.ShallowSizeOfInstance(typeof(Term))) + dataPagedBytes.RamBytesUsed() + indexToDataOffset.RamBytesUsed();
}
/// <summary>
/// Gets the term at the given position. For testing.
/// </summary>
/// <param name="termIndex">
/// The position to read the term from the index. </param>
/// <returns> The term. </returns>
/// <exception cref="IOException"> If there is a low-level I/O error. </exception>
internal virtual Term GetTerm(int termIndex)
{
PagedBytesDataInput input = (PagedBytesDataInput)dataInput.Clone();
input.SetPosition(indexToDataOffset.Get(termIndex));
// read the term
int fieldId = input.ReadVInt32();
Term field = fields[fieldId];
return(new Term(field.Field, input.ReadString()));
}
/// <summary>
/// Gets the term at the given position. For testing.
/// </summary>
/// <param name="termIndex">
/// the position to read the term from the index. </param>
/// <returns> the term. </returns>
/// <exception cref="IOException"> If there is a low-level I/O error. </exception>
public virtual Term GetTerm(int termIndex)
{
PagedBytesDataInput input = (PagedBytesDataInput)DataInput.Clone();
input.Position = IndexToDataOffset.Get(termIndex);
// read the term
int fieldId = input.ReadVInt();
Term field = Fields[fieldId];
return(new Term(field.Field, input.ReadString()));
}
/// <summary>
/// Compare the fields of the terms first, and if not equals return from
/// compare. If equal compare terms.
/// </summary>
/// <param name="term">
/// The term to compare. </param>
/// <param name="termIndex">
/// The position of the term in the input to compare </param>
/// <param name="input">
/// The input buffer. </param>
/// <returns> int. </returns>
/// <exception cref="IOException"> If there is a low-level I/O error. </exception>
private int CompareTo(Term term, int termIndex, PagedBytesDataInput input, BytesRef reuse)
{
// if term field does not equal mid's field index, then compare fields
// else if they are equal, compare term's string values...
int c = CompareField(term, termIndex, input);
if (c == 0)
{
reuse.Length = input.ReadVInt32();
reuse.Grow(reuse.Length);
input.ReadBytes(reuse.Bytes, 0, reuse.Length);
return(comparer.Compare(term.Bytes, reuse));
}
return(c);
}
/// <summary>
/// Compares the fields before checking the text of the terms.
/// </summary>
/// <param name="term">
/// The given term. </param>
/// <param name="termIndex">
/// The term that exists in the data block. </param>
/// <param name="input">
/// The data block. </param>
/// <returns> int. </returns>
/// <exception cref="IOException"> If there is a low-level I/O error. </exception>
private int CompareField(Term term, int termIndex, PagedBytesDataInput input)
{
input.SetPosition(indexToDataOffset.Get(termIndex));
return(term.Field.CompareToOrdinal(fields[input.ReadVInt32()].Field));
}
/// <summary>
/// Compares the fields before checking the text of the terms.
/// </summary>
/// <param name="term">
/// the given term. </param>
/// <param name="termIndex">
/// the term that exists in the data block. </param>
/// <param name="input">
/// the data block. </param>
/// <returns> int. </returns>
/// <exception cref="IOException"> If there is a low-level I/O error. </exception>
private int CompareField(Term term, int termIndex, PagedBytesDataInput input)
{
input.Position = IndexToDataOffset.Get(termIndex);
return(System.String.Compare(term.Field, Fields[input.ReadVInt()].Field, System.StringComparison.Ordinal));
}
/// <summary>
/// Compares the fields before checking the text of the terms.
/// </summary>
/// <param name="term">
/// the given term. </param>
/// <param name="termIndex">
/// the term that exists in the data block. </param>
/// <param name="input">
/// the data block. </param>
/// <returns> int. </returns>
/// <exception cref="IOException"> If there is a low-level I/O error. </exception>
private int CompareField(Term term, int termIndex, PagedBytesDataInput input)
{
input.Position = IndexToDataOffset.Get(termIndex);
return(term.Field().CompareTo(Fields[input.ReadVInt()].Field()));
}
