Beispiel #1
0
            public override void FinishTerm(BytesRef text, TermStats stats)
            {
                // write term meta data into fst

                var state = _outerInstance._postingsWriter.NewTermState();

                var meta = new FSTTermOutputs.TermData
                {
                    LONGS           = new long[_longsSize],
                    BYTES           = null,
                    DOC_FREQ        = state.DocFreq = stats.DocFreq,
                    TOTAL_TERM_FREQ = state.TotalTermFreq = stats.TotalTermFreq
                };

                _outerInstance._postingsWriter.FinishTerm(state);
                _outerInstance._postingsWriter.EncodeTerm(meta.LONGS, _metaWriter, _fieldInfo, state, true);
                var bytesSize = (int)_metaWriter.FilePointer;

                if (bytesSize > 0)
                {
                    meta.BYTES = new byte[bytesSize];
                    _metaWriter.WriteTo(meta.BYTES, 0);
                    _metaWriter.Reset();
                }
                _builder.Add(Util.ToIntsRef(text, _scratchTerm), meta);
                _numTerms++;
            }
 /// <summary>
 /// Load frame for target arc(node) on fst, so that
 ///  arc.label >= label and !fsa.reject(arc.label)
 /// </summary>
 internal Frame LoadCeilFrame(int label, Frame top, Frame frame)
 {
     FST.Arc <FSTTermOutputs.TermData> arc = frame.fstArc;
     arc = Util.readCeilArc(label, fst, top.fstArc, arc, fstReader);
     if (arc == null)
     {
         return(null);
     }
     frame.fsaState = fsa.Step(top.fsaState, arc.Label);
     //if (TEST) System.out.println(" loadCeil frame="+frame);
     if (frame.fsaState == -1)
     {
         return(LoadNextFrame(top, frame));
     }
     return(frame);
 }
            private void LoadTerms()
            {
                var posIntOutputs = PositiveIntOutputs.Singleton;
                var outputsInner  = new PairOutputs <long?, long?>(posIntOutputs, posIntOutputs);
                var outputs       = new PairOutputs <long?, PairOutputs <long?, long?> .Pair>(posIntOutputs, outputsInner);

                // honestly, wtf kind of generic mess is this.
                var b     = new Builder <PairOutputs <long?, PairOutputs <long?, long?> .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();
            }