private void LoadTerms()
{
var posIntOutputs = PositiveIntOutputs.Singleton;
var outputsInner = new PairOutputs <long, long>(posIntOutputs, posIntOutputs);
//var outputs = new PairOutputs<long, PairOutputs.Pair<long, long>>(posIntOutputs, outputsInner);
var outputs = new PairOutputs <long, PairOutputs.Pair>(posIntOutputs, outputsInner);
// honestly, wtf kind of generic mess is this.
var b = new Builder <PairOutputs <long, PairOutputs.Pair> .Pair>(FST.INPUT_TYPE.BYTE1, outputs);
var input = (IndexInput)_outerInstance._input.Clone();
input.Seek(_termsStart);
var lastTerm = new BytesRef(10);
long lastDocsStart = -1;
int docFreq = 0;
long totalTermFreq = 0;
var visitedDocs = new FixedBitSet(_maxDoc);
var scratchIntsRef = new IntsRef();
while (true)
{
SimpleTextUtil.ReadLine(input, _scratch);
if (_scratch.Equals(SimpleTextFieldsWriter.END) || StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FIELD))
{
if (lastDocsStart != -1)
{
b.Add(Util.ToIntsRef(lastTerm, scratchIntsRef),
outputs.NewPair(lastDocsStart, outputsInner.NewPair(docFreq, totalTermFreq)));
_sumTotalTermFreq += totalTermFreq;
}
break;
}
if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.DOC))
{
docFreq++;
_sumDocFreq++;
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.DOC.Length, _scratch.Length - SimpleTextFieldsWriter.DOC.Length,
_scratchUtf16);
int docId = ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.length);
visitedDocs.Set(docId);
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FREQ))
{
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.FREQ.Length,
_scratch.Length - SimpleTextFieldsWriter.FREQ.Length, _scratchUtf16);
totalTermFreq += ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.length);
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.TERM))
{
if (lastDocsStart != -1)
{
b.Add(Util.ToIntsRef(lastTerm, scratchIntsRef),
outputs.NewPair(lastDocsStart, outputsInner.NewPair(docFreq, totalTermFreq)));
}
lastDocsStart = input.FilePointer;
int len = _scratch.Length - SimpleTextFieldsWriter.TERM.Length;
if (len > lastTerm.Length)
{
lastTerm.Grow(len);
}
Array.Copy(_scratch.Bytes, SimpleTextFieldsWriter.TERM.Length, lastTerm.Bytes, 0, len);
lastTerm.Length = len;
docFreq = 0;
_sumTotalTermFreq += totalTermFreq;
totalTermFreq = 0;
_termCount++;
}
}
_docCount = visitedDocs.Cardinality();
_fst = b.Finish();
}
public override int NextDoc()
{
if (_docId == NO_MORE_DOCS)
{
return(_docId);
}
bool first = true;
int termFreq = 0;
while (true)
{
long lineStart = _in.FilePointer;
SimpleTextUtil.ReadLine(_in, _scratch);
if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.DOC))
{
if (!first && (_liveDocs == null || _liveDocs.Get(_docId)))
{
_in.Seek(lineStart);
if (!_omitTf)
{
_tf = termFreq;
}
return(_docId);
}
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.DOC.Length, _scratch.Length - SimpleTextFieldsWriter.DOC.Length,
_scratchUtf16);
_docId = ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.length);
termFreq = 0;
first = false;
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FREQ))
{
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.FREQ.Length,
_scratch.Length - SimpleTextFieldsWriter.FREQ.Length, _scratchUtf16);
termFreq = ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.length);
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.POS))
{
// skip termFreq++;
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.START_OFFSET))
{
// skip
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.END_OFFSET))
{
// skip
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.PAYLOAD))
{
// skip
}
else
{
Debug.Assert(
StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.TERM) || StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FIELD) ||
StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.END), "scratch=" + _scratch.Utf8ToString());
if (!first && (_liveDocs == null || _liveDocs.Get(_docId)))
{
_in.Seek(lineStart);
if (!_omitTf)
{
_tf = termFreq;
}
return(_docId);
}
return(_docId = NO_MORE_DOCS);
}
}
}
public override int NextDoc()
{
bool first = true;
_in.Seek(_nextDocStart);
long posStart = 0;
while (true)
{
long lineStart = _in.FilePointer;
SimpleTextUtil.ReadLine(_in, _scratch);
//System.out.println("NEXT DOC: " + scratch.utf8ToString());
if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.DOC))
{
if (!first && (_liveDocs == null || _liveDocs.Get(_docId)))
{
_nextDocStart = lineStart;
_in.Seek(posStart);
return(_docId);
}
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.DOC.Length, _scratch.Length - SimpleTextFieldsWriter.DOC.Length,
_scratchUtf16);
_docId = ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.length);
_tf = 0;
first = false;
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FREQ))
{
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.FREQ.Length,
_scratch.Length - SimpleTextFieldsWriter.FREQ.Length, _scratchUtf16);
_tf = ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.length);
posStart = _in.FilePointer;
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.POS))
{
// skip
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.START_OFFSET))
{
// skip
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.END_OFFSET))
{
// skip
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.PAYLOAD))
{
// skip
}
else
{
Debug.Assert(StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.TERM) || StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FIELD) ||
StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.END));
if (!first && (_liveDocs == null || _liveDocs.Get(_docId)))
{
_nextDocStart = lineStart;
_in.Seek(posStart);
return(_docId);
}
return(_docId = NO_MORE_DOCS);
}
}
}
private static int ParseIntAt(BytesRef bytes, int offset, CharsRef scratch)
{
UnicodeUtil.UTF8toUTF16(bytes.Bytes, bytes.Offset + offset, bytes.Length - offset, scratch);
return(ArrayUtil.ParseInt(scratch.Chars, 0, scratch.Length));
}