public DirectField(SegmentReadState state, string field, Terms termsIn, int minSkipCount, int lowFreqCutoff)
            {
                FieldInfo fieldInfo = state.FieldInfos.FieldInfo(field);

                sumTotalTermFreq = termsIn.SumTotalTermFreq;
                sumDocFreq = termsIn.SumDocFreq;
                docCount = termsIn.DocCount;

                int numTerms = (int) termsIn.Size();
                if (numTerms == -1)
                {
                    throw new System.ArgumentException("codec does not provide Terms.size()");
                }
                terms = new TermAndSkip[numTerms];
                termOffsets = new int[1 + numTerms];

                byte[] termBytes = new byte[1024];

                this.minSkipCount = minSkipCount;

                hasFreq = fieldInfo.FieldIndexOptions.Value.CompareTo(IndexOptions.DOCS_ONLY) > 0;
                hasPos = fieldInfo.FieldIndexOptions.Value.CompareTo(IndexOptions.DOCS_AND_FREQS) > 0;
                hasOffsets_Renamed = fieldInfo.FieldIndexOptions.Value.CompareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS) > 0;
                hasPayloads_Renamed = fieldInfo.HasPayloads();

                BytesRef term;
                DocsEnum docsEnum = null;
                DocsAndPositionsEnum docsAndPositionsEnum = null;
                TermsEnum termsEnum = termsIn.Iterator(null);
                int termOffset = 0;

                IntArrayWriter scratch = new IntArrayWriter();

                // Used for payloads, if any:
                RAMOutputStream ros = new RAMOutputStream();

                // if (DEBUG) {
                //   System.out.println("\nLOAD terms seg=" + state.segmentInfo.name + " field=" + field + " hasOffsets=" + hasOffsets + " hasFreq=" + hasFreq + " hasPos=" + hasPos + " hasPayloads=" + hasPayloads);
                // }

                while ((term = termsEnum.Next()) != null)
                {
                    int docFreq = termsEnum.DocFreq();
                    long totalTermFreq = termsEnum.TotalTermFreq();

                    // if (DEBUG) {
                    //   System.out.println("  term=" + term.utf8ToString());
                    // }

                    termOffsets[count] = termOffset;

                    if (termBytes.Length < (termOffset + term.Length))
                    {
                        termBytes = ArrayUtil.Grow(termBytes, termOffset + term.Length);
                    }
                    Array.Copy(term.Bytes, term.Offset, termBytes, termOffset, term.Length);
                    termOffset += term.Length;
                    termOffsets[count + 1] = termOffset;

                    if (hasPos)
                    {
                        docsAndPositionsEnum = termsEnum.DocsAndPositions(null, docsAndPositionsEnum);
                    }
                    else
                    {
                        docsEnum = termsEnum.Docs(null, docsEnum);
                    }

                    TermAndSkip ent;

                    DocsEnum docsEnum2;
                    docsEnum2 = hasPos ? docsAndPositionsEnum : docsEnum;

                    int docID;

                    if (docFreq <= lowFreqCutoff)
                    {

                        ros.Reset();

                        // Pack postings for low-freq terms into a single int[]:
                        while ((docID = docsEnum2.NextDoc()) != DocsEnum.NO_MORE_DOCS)
                        {
                            scratch.Add(docID);
                            if (hasFreq)
                            {
                                int freq = docsEnum2.Freq();
                                scratch.Add(freq);
                                if (hasPos)
                                {
                                    for (int pos = 0; pos < freq; pos++)
                                    {
                                        scratch.Add(docsAndPositionsEnum.NextPosition());
                                        if (hasOffsets_Renamed)
                                        {
                                            scratch.Add(docsAndPositionsEnum.StartOffset());
                                            scratch.Add(docsAndPositionsEnum.EndOffset());
                                        }
                                        if (hasPayloads_Renamed)
                                        {
                                            BytesRef payload = docsAndPositionsEnum.Payload;
                                            if (payload != null)
                                            {
                                                scratch.Add(payload.Length);
                                                ros.WriteBytes(payload.Bytes, payload.Offset, payload.Length);
                                            }
                                            else
                                            {
                                                scratch.Add(0);
                                            }
                                        }
                                    }
                                }
                            }
                        }


                        byte[] payloads;
                        if (hasPayloads_Renamed)
                        {
                            ros.Flush();
                            payloads = new byte[(int) ros.Length];
                            ros.WriteTo(payloads, 0);
                        }
                        else
                        {
                            payloads = null;
                        }

                        int[] postings = scratch.Get();

                        ent = new LowFreqTerm(postings, payloads, docFreq, (int) totalTermFreq);
                    }
                    else
                    {
                        var docs = new int[docFreq];
                        int[] freqs;
                        int[][] positions;
                        byte[][][] payloads;

                        if (hasFreq)
                        {
                            freqs = new int[docFreq];
                            if (hasPos)
                            {
                                positions = new int[docFreq][];
                                if (hasPayloads_Renamed)
                                {
                                    payloads = new byte[docFreq][][];
                                }
                                else
                                {
                                    payloads = null;
                                }
                            }
                            else
                            {
                                positions = null;
                                payloads = null;
                            }
                        }
                        else
                        {
                            freqs = null;
                            positions = null;
                            payloads = null;
                        }

                        // Use separate int[] for the postings for high-freq
                        // terms:
                        int upto = 0;
                        while ((docID = docsEnum2.NextDoc()) != DocsEnum.NO_MORE_DOCS)
                        {
                            docs[upto] = docID;
                            if (hasFreq)
                            {
                                int freq = docsEnum2.Freq();
                                freqs[upto] = freq;
                                if (hasPos)
                                {
                                    int mult;
                                    if (hasOffsets_Renamed)
                                    {
                                        mult = 3;
                                    }
                                    else
                                    {
                                        mult = 1;
                                    }
                                    if (hasPayloads_Renamed)
                                    {
                                        payloads[upto] = new byte[freq][];
                                    }
                                    positions[upto] = new int[mult*freq];
                                    int posUpto = 0;
                                    for (int pos = 0; pos < freq; pos++)
                                    {
                                        positions[upto][posUpto] = docsAndPositionsEnum.NextPosition();
                                        if (hasPayloads_Renamed)
                                        {
                                            BytesRef payload = docsAndPositionsEnum.Payload;
                                            if (payload != null)
                                            {
                                                var payloadBytes = new byte[payload.Length];
                                                Array.Copy(payload.Bytes, payload.Offset, payloadBytes, 0,
                                                    payload.Length);
                                                payloads[upto][pos] = payloadBytes;
                                            }
                                        }
                                        posUpto++;
                                        if (hasOffsets_Renamed)
                                        {
                                            positions[upto][posUpto++] = docsAndPositionsEnum.StartOffset();
                                            positions[upto][posUpto++] = docsAndPositionsEnum.EndOffset();
                                        }
                                    }
                                }
                            }

                            upto++;
                        }
                        Debug.Assert(upto == docFreq);
                        ent = new HighFreqTerm(docs, freqs, positions, payloads, totalTermFreq);
                    }

                    terms[count] = ent;
                    SetSkips(count, termBytes);
                    count++;
                }

                // End sentinel:
                termOffsets[count] = termOffset;

                FinishSkips();

                //System.out.println(skipCount + " skips: " + field);

                this.termBytes = new byte[termOffset];
                Array.Copy(termBytes, 0, this.termBytes, 0, termOffset);

                // Pack skips:
                this.skips = new int[skipCount];
                this.skipOffsets = new int[1 + numTerms];

                int skipOffset = 0;
                for (int i = 0; i < numTerms; i++)
                {
                    int[] termSkips = terms[i].skips;
                    skipOffsets[i] = skipOffset;
                    if (termSkips != null)
                    {
                        Array.Copy(termSkips, 0, skips, skipOffset, termSkips.Length);
                        skipOffset += termSkips.Length;
                        terms[i].skips = null;
                    }
                }
                this.skipOffsets[numTerms] = skipOffset;
                Debug.Assert(skipOffset == skipCount);
            }
Exemplo n.º 2
0
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public DirectField(index.SegmentReadState state, String field, index.Terms termsIn, int minSkipCount, int lowFreqCutoff) throws java.io.IOException
            public DirectField(SegmentReadState state, string field, Terms termsIn, int minSkipCount, int lowFreqCutoff)
            {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final index.FieldInfo fieldInfo = state.fieldInfos.fieldInfo(field);
                FieldInfo fieldInfo = state.fieldInfos.fieldInfo(field);

                sumTotalTermFreq = termsIn.SumTotalTermFreq;
                sumDocFreq = termsIn.SumDocFreq;
                docCount = termsIn.DocCount;

//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int numTerms = (int) termsIn.size();
                int numTerms = (int) termsIn.size();
                if (numTerms == -1)
                {
                    throw new System.ArgumentException("codec does not provide Terms.size()");
                }
                terms = new TermAndSkip[numTerms];
                termOffsets = new int[1 + numTerms];

                sbyte[] termBytes = new sbyte[1024];

                this.minSkipCount = minSkipCount;

                hasFreq = fieldInfo.IndexOptions.compareTo(IndexOptions.DOCS_ONLY) > 0;
                hasPos = fieldInfo.IndexOptions.compareTo(IndexOptions.DOCS_AND_FREQS) > 0;
                hasOffsets_Renamed = fieldInfo.IndexOptions.compareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS) > 0;
                hasPayloads_Renamed = fieldInfo.hasPayloads();

                BytesRef term;
                DocsEnum docsEnum = null;
                DocsAndPositionsEnum docsAndPositionsEnum = null;
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final index.TermsEnum termsEnum = termsIn.iterator(null);
                TermsEnum termsEnum = termsIn.iterator(null);
                int termOffset = 0;

//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final IntArrayWriter scratch = new IntArrayWriter();
                IntArrayWriter scratch = new IntArrayWriter();

                // Used for payloads, if any:
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final store.RAMOutputStream ros = new store.RAMOutputStream();
                RAMOutputStream ros = new RAMOutputStream();

                // if (DEBUG) {
                //   System.out.println("\nLOAD terms seg=" + state.segmentInfo.name + " field=" + field + " hasOffsets=" + hasOffsets + " hasFreq=" + hasFreq + " hasPos=" + hasPos + " hasPayloads=" + hasPayloads);
                // }

                while ((term = termsEnum.next()) != null)
                {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int docFreq = termsEnum.docFreq();
                    int docFreq = termsEnum.docFreq();
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final long totalTermFreq = termsEnum.totalTermFreq();
                    long totalTermFreq = termsEnum.totalTermFreq();

                    // if (DEBUG) {
                    //   System.out.println("  term=" + term.utf8ToString());
                    // }

                    termOffsets[count] = termOffset;

                    if (termBytes.Length < (termOffset + term.length))
                    {
                        termBytes = ArrayUtil.grow(termBytes, termOffset + term.length);
                    }
                    Array.Copy(term.bytes, term.offset, termBytes, termOffset, term.length);
                    termOffset += term.length;
                    termOffsets[count + 1] = termOffset;

                    if (hasPos)
                    {
                        docsAndPositionsEnum = termsEnum.docsAndPositions(null, docsAndPositionsEnum);
                    }
                    else
                    {
                        docsEnum = termsEnum.docs(null, docsEnum);
                    }

//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final TermAndSkip ent;
                    TermAndSkip ent;

//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final index.DocsEnum docsEnum2;
                    DocsEnum docsEnum2;
                    if (hasPos)
                    {
                        docsEnum2 = docsAndPositionsEnum;
                    }
                    else
                    {
                        docsEnum2 = docsEnum;
                    }

                    int docID;

                    if (docFreq <= lowFreqCutoff)
                    {

                        ros.reset();

                        // Pack postings for low-freq terms into a single int[]:
                        while ((docID = docsEnum2.nextDoc()) != DocsEnum.NO_MORE_DOCS)
                        {
                            scratch.add(docID);
                            if (hasFreq)
                            {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int freq = docsEnum2.freq();
                                int freq = docsEnum2.freq();
                                scratch.add(freq);
                                if (hasPos)
                                {
                                    for (int pos = 0; pos < freq; pos++)
                                    {
                                        scratch.add(docsAndPositionsEnum.nextPosition());
                                        if (hasOffsets_Renamed)
                                        {
                                            scratch.add(docsAndPositionsEnum.startOffset());
                                            scratch.add(docsAndPositionsEnum.endOffset());
                                        }
                                        if (hasPayloads_Renamed)
                                        {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final util.BytesRef payload = docsAndPositionsEnum.getPayload();
                                            BytesRef payload = docsAndPositionsEnum.Payload;
                                            if (payload != null)
                                            {
                                                scratch.add(payload.length);
                                                ros.writeBytes(payload.bytes, payload.offset, payload.length);
                                            }
                                            else
                                            {
                                                scratch.add(0);
                                            }
                                        }
                                    }
                                }
                            }
                        }

//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final byte[] payloads;
                        sbyte[] payloads;
                        if (hasPayloads_Renamed)
                        {
                            ros.flush();
                            payloads = new sbyte[(int) ros.length()];
                            ros.writeTo(payloads, 0);
                        }
                        else
                        {
                            payloads = null;
                        }

//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int[] postings = scratch.get();
                        int[] postings = scratch.get();

                        ent = new LowFreqTerm(postings, payloads, docFreq, (int) totalTermFreq);
                    }
                    else
                    {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int[] docs = new int[docFreq];
                        int[] docs = new int[docFreq];
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int[] freqs;
                        int[] freqs;
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int[][] positions;
                        int[][] positions;
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final byte[][][] payloads;
                        sbyte[][][] payloads;
                        if (hasFreq)
                        {
                            freqs = new int[docFreq];
                            if (hasPos)
                            {
                                positions = new int[docFreq][];
                                if (hasPayloads_Renamed)
                                {
                                    payloads = new sbyte[docFreq][][];
                                }
                                else
                                {
                                    payloads = null;
                                }
                            }
                            else
                            {
                                positions = null;
                                payloads = null;
                            }
                        }
                        else
                        {
                            freqs = null;
                            positions = null;
                            payloads = null;
                        }

                        // Use separate int[] for the postings for high-freq
                        // terms:
                        int upto = 0;
                        while ((docID = docsEnum2.nextDoc()) != DocsEnum.NO_MORE_DOCS)
                        {
                            docs[upto] = docID;
                            if (hasFreq)
                            {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int freq = docsEnum2.freq();
                                int freq = docsEnum2.freq();
                                freqs[upto] = freq;
                                if (hasPos)
                                {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int mult;
                                    int mult;
                                    if (hasOffsets_Renamed)
                                    {
                                        mult = 3;
                                    }
                                    else
                                    {
                                        mult = 1;
                                    }
                                    if (hasPayloads_Renamed)
                                    {
                                        payloads[upto] = new sbyte[freq][];
                                    }
                                    positions[upto] = new int[mult*freq];
                                    int posUpto = 0;
                                    for (int pos = 0; pos < freq; pos++)
                                    {
                                        positions[upto][posUpto] = docsAndPositionsEnum.nextPosition();
                                        if (hasPayloads_Renamed)
                                        {
                                            BytesRef payload = docsAndPositionsEnum.Payload;
                                            if (payload != null)
                                            {
                                                sbyte[] payloadBytes = new sbyte[payload.length];
                                                Array.Copy(payload.bytes, payload.offset, payloadBytes, 0,
                                                    payload.length);
                                                payloads[upto][pos] = payloadBytes;
                                            }
                                        }
                                        posUpto++;
                                        if (hasOffsets_Renamed)
                                        {
                                            positions[upto][posUpto++] = docsAndPositionsEnum.startOffset();
                                            positions[upto][posUpto++] = docsAndPositionsEnum.endOffset();
                                        }
                                    }
                                }
                            }

                            upto++;
                        }
                        Debug.Assert(upto == docFreq);
                        ent = new HighFreqTerm(docs, freqs, positions, payloads, totalTermFreq);
                    }

                    terms[count] = ent;
                    setSkips(count, termBytes);
                    count++;
                }

                // End sentinel:
                termOffsets[count] = termOffset;

                finishSkips();

                //System.out.println(skipCount + " skips: " + field);

                this.termBytes = new sbyte[termOffset];
                Array.Copy(termBytes, 0, this.termBytes, 0, termOffset);

                // Pack skips:
                this.skips = new int[skipCount];
                this.skipOffsets = new int[1 + numTerms];

                int skipOffset = 0;
                for (int i = 0; i < numTerms; i++)
                {
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int[] termSkips = terms[i].skips;
                    int[] termSkips = terms[i].skips;
                    skipOffsets[i] = skipOffset;
                    if (termSkips != null)
                    {
                        Array.Copy(termSkips, 0, skips, skipOffset, termSkips.Length);
                        skipOffset += termSkips.Length;
                        terms[i].skips = null;
                    }
                }
                this.skipOffsets[numTerms] = skipOffset;
                Debug.Assert(skipOffset == skipCount);
            }