public FstFieldWriter(FieldInfo fieldInfo, long termsFilePointer, VariableGapTermsIndexWriter vgtiw)
{
_vgtiw = vgtiw;
FieldInfo = fieldInfo;
PositiveIntOutputs fstOutputs = PositiveIntOutputs.Singleton;
_fstBuilder = new Builder <long>(FST.INPUT_TYPE.BYTE1, fstOutputs);
IndexStart = _vgtiw.Output.FilePointer;
// Always put empty string in
_fstBuilder.Add(new IntsRef(), termsFilePointer);
_startTermsFilePointer = termsFilePointer;
}
internal TermsWriter(FSTOrdTermsWriter outerInstance, FieldInfo fieldInfo)
{
this.outerInstance = outerInstance;
this.numTerms = 0;
this.fieldInfo = fieldInfo;
this.longsSize = outerInstance.postingsWriter.setField(fieldInfo);
this.outputs = PositiveIntOutputs.Singleton;
this.builder = new Builder <>(FST.INPUT_TYPE.BYTE1, outputs);
this.lastBlockStatsFP = 0;
this.lastBlockMetaLongsFP = 0;
this.lastBlockMetaBytesFP = 0;
this.lastBlockLongs = new long[longsSize];
this.lastLongs = new long[longsSize];
this.lastMetaBytesFP = 0;
}
internal TermsWriter(FSTOrdTermsWriter outerInstance, FieldInfo fieldInfo)
{
_outerInstance = outerInstance;
_numTerms = 0;
_fieldInfo = fieldInfo;
_longsSize = outerInstance.postingsWriter.SetField(fieldInfo);
_outputs = PositiveIntOutputs.Singleton;
_builder = new Builder <long>(FST.INPUT_TYPE.BYTE1, _outputs);
_lastBlockStatsFp = 0;
_lastBlockMetaLongsFp = 0;
_lastBlockMetaBytesFp = 0;
_lastBlockLongs = new long[_longsSize];
_lastLongs = new long[_longsSize];
_lastMetaBytesFp = 0;
}
private void WriteFST(FieldInfo field, IEnumerable <BytesRef> values)
{
Meta.WriteVInt(field.Number);
Meta.WriteByte(Lucene42DocValuesProducer.FST);
Meta.WriteLong(Data.FilePointer);
PositiveIntOutputs outputs = PositiveIntOutputs.Singleton;
Builder <long> builder = new Builder <long>(INPUT_TYPE.BYTE1, outputs);
IntsRef scratch = new IntsRef();
long ord = 0;
foreach (BytesRef v in values)
{
builder.Add(Util.ToIntsRef(v, scratch), ord);
ord++;
}
Lucene.Net.Util.Fst.FST <long> fst = builder.Finish();
if (fst != null)
{
fst.Save(Data);
}
Meta.WriteVLong(ord);
}
public override void Build(InputIterator iterator)
{
if (iterator.HasPayloads)
{
throw new ArgumentException("this suggester doesn't support payloads");
}
if (iterator.HasContexts)
{
throw new ArgumentException("this suggester doesn't support contexts");
}
count = 0;
BytesRef scratch = new BytesRef();
InputIterator iter = new WFSTInputIterator(this, iterator);
IntsRef scratchInts = new IntsRef();
BytesRef previous = null;
PositiveIntOutputs outputs = PositiveIntOutputs.Singleton;
Builder <long?> builder = new Builder <long?>(FST.INPUT_TYPE.BYTE1, outputs);
while ((scratch = iter.Next()) != null)
{
long cost = iter.Weight;
if (previous == null)
{
previous = new BytesRef();
}
else if (scratch.Equals(previous))
{
continue; // for duplicate suggestions, the best weight is actually
// added
}
Util.ToIntsRef(scratch, scratchInts);
builder.Add(scratchInts, cost);
previous.CopyBytes(scratch);
count++;
}
fst = builder.Finish();
}
private void WriteFST(FieldInfo field, IEnumerable <BytesRef> values)
{
meta.WriteVInt(field.Number);
meta.WriteByte(FST);
meta.WriteLong(data.FilePointer);
PositiveIntOutputs outputs = PositiveIntOutputs.Singleton;
var builder = new Builder <long?>(INPUT_TYPE.BYTE1, outputs);
var scratch = new IntsRef();
long ord = 0;
foreach (BytesRef v in values)
{
builder.Add(Util.ToIntsRef(v, scratch), ord);
ord++;
}
FST <long?> fst = builder.Finish();
if (fst != null)
{
fst.Save(data);
}
meta.WriteVLong(ord);
}
internal TermsWriter(FSTOrdTermsWriter outerInstance, FieldInfo fieldInfo)
{
_outerInstance = outerInstance;
_numTerms = 0;
_fieldInfo = fieldInfo;
_longsSize = outerInstance.postingsWriter.SetField(fieldInfo);
_outputs = PositiveIntOutputs.Singleton;
_builder = new Builder<long>(FST.INPUT_TYPE.BYTE1, _outputs);
_lastBlockStatsFp = 0;
_lastBlockMetaLongsFp = 0;
_lastBlockMetaBytesFp = 0;
_lastBlockLongs = new long[_longsSize];
_lastLongs = new long[_longsSize];
_lastMetaBytesFp = 0;
}
public FSTFieldWriter(FieldInfo fieldInfo, long termsFilePointer)
{
this.fieldInfo = fieldInfo;
fstOutputs = PositiveIntOutputs.Singleton;
fstBuilder = new Builder<>(FST.INPUT_TYPE.BYTE1, fstOutputs);
indexStart = output.FilePointer;
////System.out.println("VGW: field=" + fieldInfo.name);
// Always put empty string in
fstBuilder.Add(new IntsRef(), termsFilePointer);
startTermsFilePointer = termsFilePointer;
}
internal virtual void LoadTerms()
{
PositiveIntOutputs posIntOutputs = PositiveIntOutputs.Singleton;
Builder <PairOutputs.Pair <long?, PairOutputs.Pair <long?, long?> > > b;
PairOutputs <long?, long?> outputsInner = new PairOutputs <long?, long?>(posIntOutputs, posIntOutputs);
PairOutputs <long?, PairOutputs.Pair <long?, long?> > outputs =
new PairOutputs <long?, PairOutputs.Pair <long?, long?> >(posIntOutputs, outputsInner);
b = new Builder <>(FST.INPUT_TYPE.BYTE1, outputs);
IndexInput @in = (IndexInput)outerInstance._input.Clone();
@in.Seek(termsStart);
BytesRef lastTerm = new BytesRef(10);
long lastDocsStart = -1;
int docFreq = 0;
long totalTermFreq = 0;
FixedBitSet visitedDocs = new FixedBitSet(maxDoc);
IntsRef scratchIntsRef = new IntsRef();
while (true)
{
SimpleTextUtil.ReadLine(@in, scratch);
if (scratch.Equals(END) || StringHelper.StartsWith(scratch, FIELD))
{
if (lastDocsStart != -1)
{
b.Add(Util.ToIntsRef(lastTerm, scratchIntsRef),
outputs.NewPair(lastDocsStart, outputsInner.NewPair((long)docFreq, totalTermFreq)));
sumTotalTermFreq += totalTermFreq;
}
break;
}
else if (StringHelper.StartsWith(scratch, DOC))
{
docFreq++;
sumDocFreq++;
UnicodeUtil.UTF8toUTF16(scratch.Bytes, scratch.Offset + DOC.length, scratch.Length - DOC.length,
scratchUTF16);
int docID = ArrayUtil.ParseInt(scratchUTF16.Chars, 0, scratchUTF16.length);
visitedDocs.Set(docID);
}
else if (StringHelper.StartsWith(scratch, FREQ))
{
UnicodeUtil.UTF8toUTF16(scratch.Bytes, scratch.Offset + FREQ.length,
scratch.Length - FREQ.length, scratchUTF16);
totalTermFreq += ArrayUtil.ParseInt(scratchUTF16.Chars, 0, scratchUTF16.length);
}
else if (StringHelper.StartsWith(scratch, TERM))
{
if (lastDocsStart != -1)
{
b.Add(Util.ToIntsRef(lastTerm, scratchIntsRef),
outputs.NewPair(lastDocsStart, outputsInner.NewPair((long)docFreq, totalTermFreq)));
}
lastDocsStart = @in.FilePointer;
int len = scratch.Length - TERM.length;
if (len > lastTerm.Length)
{
lastTerm.Grow(len);
}
Array.Copy(scratch.Bytes, TERM.length, lastTerm.Bytes, 0, len);
lastTerm.Length = len;
docFreq = 0;
sumTotalTermFreq += totalTermFreq;
totalTermFreq = 0;
termCount++;
}
}
docCount = visitedDocs.Cardinality();
fst = b.Finish();
}
