Ejemplo n.º 1
0
func newLucene49NormsConsumer(state *SegmentWriteState,
	dataCodec, dataExtension, metaCodec, metaExtension string) (nc *NormsConsumer, err error) {

	assert(packed.PackedFormat(packed.PACKED_SINGLE_BLOCK).IsSupported(1))
	assert(packed.PackedFormat(packed.PACKED_SINGLE_BLOCK).IsSupported(2))
	assert(packed.PackedFormat(packed.PACKED_SINGLE_BLOCK).IsSupported(4))

	nc = &NormsConsumer{maxDoc: state.SegmentInfo.DocCount()}
	var success = false
	defer func() {
		if !success {
			util.CloseWhileSuppressingError(nc)
		}
	}()

	dataName := util.SegmentFileName(state.SegmentInfo.Name, state.SegmentSuffix, dataExtension)
	if nc.data, err = state.Directory.CreateOutput(dataName, state.Context); err != nil {
		return nil, err
	}
	if err = codec.WriteHeader(nc.data, dataCodec, VERSION_CURRENT); err != nil {
		return nil, err
	}
	metaName := util.SegmentFileName(state.SegmentInfo.Name, state.SegmentSuffix, metaExtension)
	if nc.meta, err = state.Directory.CreateOutput(metaName, state.Context); err != nil {
		return nil, err
	}
	if err = codec.WriteHeader(nc.meta, metaCodec, VERSION_CURRENT); err != nil {
		return nil, err
	}
	success = true
	return nc, nil
}
Ejemplo n.º 2
0
func (w *CompoundFileWriter) writeEntryTable(entries map[string]*FileEntry,
	entryOut IndexOutput) (err error) {
	if err = codec.WriteHeader(entryOut, CFD_ENTRY_CODEC, CFD_VERSION_CURRENT); err == nil {
		if err = entryOut.WriteVInt(int32(len(entries))); err == nil {
			var names []string
			for name, _ := range entries {
				names = append(names, name)
			}
			sort.Strings(names)
			for _, name := range names {
				// for _, fe := range entries {
				fe := entries[name]
				if err = Stream(entryOut).
					WriteString(util.StripSegmentName(fe.file)).
					WriteLong(fe.offset).
					WriteLong(fe.length).
					Close(); err != nil {
					break
				}
			}
		}
	}
	if err == nil {
		err = codec.WriteFooter(entryOut)
	}
	return err
}
Ejemplo n.º 3
0
/*
Writes this vector to the file name in Directory d, in a format that
can be read by the constructor BitVector(Directory, String, IOContext)
*/
func (bv *BitVector) Write(d store.Directory, name string, ctx store.IOContext) (err error) {
	assert(reflect.TypeOf(d).Name() != "CompoundFileDirectory")
	var output store.IndexOutput
	if output, err = d.CreateOutput(name, ctx); err != nil {
		return err
	}
	defer func() {
		err = mergeError(err, output.Close())
	}()

	if err = output.WriteInt(-2); err != nil {
		return err
	}
	if err = codec.WriteHeader(output, CODEC, BV_VERSION_CURRENT); err != nil {
		return err
	}
	if bv.isSparse() {
		// sparse bit-set more efficiently saved as d-gaps.
		err = bv.writeClearedDgaps(output)
	} else {
		err = bv.writeBits(output)
	}
	if err != nil {
		return err
	}
	if err = codec.WriteFooter(output); err != nil {
		return err
	}
	bv.assertCount()
	return nil
}
Ejemplo n.º 4
0
func (w *Lucene41PostingsWriter) Init(termsOut store.IndexOutput) error {
	err := codec.WriteHeader(termsOut, LUCENE41_TERMS_CODEC, LUCENE41_VERSION_CURRENT)
	if err == nil {
		err = termsOut.WriteVInt(LUCENE41_BLOCK_SIZE)
	}
	return err
}
Ejemplo n.º 5
0
func (w *WriterImpl) writeHeader() error {
	assert(w.valueCount != -1)
	err := codec.WriteHeader(w.out, PACKED_CODEC_NAME, VERSION_CURRENT)
	if err == nil {
		err = w.out.WriteVInt(int32(w.bitsPerValue))
		if err == nil {
			err = w.out.WriteVInt(int32(w.valueCount))
			if err == nil {
				err = w.out.WriteVInt(int32(PackedFormat(w.format).Id()))
			}
		}
	}
	return err
}
Ejemplo n.º 6
0
func (w *Lucene40SegmentInfoWriter) Write(dir store.Directory,
	si *SegmentInfo, fis FieldInfos, ctx store.IOContext) (err error) {

	filename := util.SegmentFileName(si.Name, "", LUCENE40_SI_EXTENSION)
	si.AddFile(filename)

	var output store.IndexOutput
	output, err = dir.CreateOutput(filename, ctx)
	if err != nil {
		return err
	}

	var success = false
	defer func() {
		if !success {
			util.CloseWhileSuppressingError(output)
			si.Dir.DeleteFile(filename) // ignore error
		} else {
			err = mergeError(err, output.Close())
		}
	}()

	err = codec.WriteHeader(output, LUCENE40_CODEC_NAME, LUCENE40_VERSION_CURRENT)
	if err != nil {
		return err
	}
	// Write the Lucene version that created this segment, since 3.1
	err = store.Stream(output).WriteString(si.Version().String()).
		WriteInt(int32(si.DocCount())).
		WriteByte(func() byte {
			if si.IsCompoundFile() {
				return SEGMENT_INFO_YES
			}
			return byte((SEGMENT_INFO_NO + 256) % 256) // Go byte is non-negative, unlike Java
		}()).WriteStringStringMap(si.Diagnostics()).
		WriteStringStringMap(map[string]string{}).
		WriteStringSet(si.Files()).Close()
	if err != nil {
		return err
	}

	success = true
	return nil
}
Ejemplo n.º 7
0
func (w *CompoundFileWriter) output(ctx IOContext) (IndexOutput, error) {
	w.Lock()
	defer w.Unlock()
	if w.dataOut == nil {
		var success = false
		defer func() {
			if !success {
				util.CloseWhileSuppressingError(w.dataOut)
			}
		}()

		var err error
		w.dataOut, err = w.directory.CreateOutput(w.dataFileName, ctx)
		if err != nil {
			return nil, err
		}
		err = codec.WriteHeader(w.dataOut, CFD_DATA_CODEC, CFD_VERSION_CURRENT)
		if err != nil {
			return nil, err
		}
		success = true
	}
	return w.dataOut, nil
}
Ejemplo n.º 8
0
func (sis *SegmentInfos) write(directory store.Directory) (err error) {
	segmentsFilename := sis.nextSegmentFilename()

	// Always advance the generation on write:
	if sis.generation == -1 {
		sis.generation = 1
	} else {
		sis.generation++
	}

	var segnOutput store.IndexOutput
	var success = false
	// var upgradedSIFiles = make(map[string]bool)

	defer func() {
		if !success {
			// We hit an error above; try to close the file but suppress
			// any errors
			util.CloseWhileSuppressingError(segnOutput)

			// for filename, _ := range upgradedSIFiles {
			// 	directory.DeleteFile(filename) // ignore error
			// }

			// Try not to leave a truncated segments_N fle in the index:
			directory.DeleteFile(segmentsFilename) // ignore error
		}
	}()

	if segnOutput, err = directory.CreateOutput(segmentsFilename, store.IO_CONTEXT_DEFAULT); err != nil {
		return
	}
	if err = codec.WriteHeader(segnOutput, "segments", VERSION_49); err != nil {
		return
	}
	if err = segnOutput.WriteLong(sis.version); err == nil {
		if err = segnOutput.WriteInt(int32(sis.counter)); err == nil {
			err = segnOutput.WriteInt(int32(len(sis.Segments)))
		}
	}
	if err != nil {
		return
	}
	for _, siPerCommit := range sis.Segments {
		si := siPerCommit.Info
		if err = segnOutput.WriteString(si.Name); err == nil {
			if err = segnOutput.WriteString(si.Codec().(Codec).Name()); err == nil {
				if err = segnOutput.WriteLong(siPerCommit.DelGen()); err == nil {
					assert2(siPerCommit.DelCount() >= 0 && siPerCommit.DelCount() <= si.DocCount(),
						"cannot write segment: invalid docCount segment=%v docCount=%v delCount=%v",
						si.Name, si.DocCount(), siPerCommit.DelCount())
					if err = segnOutput.WriteInt(int32(siPerCommit.DelCount())); err == nil {
						if err = segnOutput.WriteLong(siPerCommit.FieldInfosGen()); err == nil {
							if err = segnOutput.WriteLong(siPerCommit.DocValuesGen()); err == nil {
								if err = segnOutput.WriteStringSet(siPerCommit.FieldInfosFiles()); err == nil {
									dvUpdatesFiles := siPerCommit.DocValuesUpdatesFiles()
									if err = segnOutput.WriteInt(int32(len(dvUpdatesFiles))); err == nil {
										for k, v := range dvUpdatesFiles {
											if err = segnOutput.WriteInt(int32(k)); err != nil {
												break
											}
											if err = segnOutput.WriteStringSet(v); err != nil {
												break
											}
										}
									}
								}
							}
						}
					}
				}
			}
		}
		if err != nil {
			return
		}
		assert(si.Dir == directory)

		// If this segment is pre-4.x, perform a one-time "upgrade" to
		// write the .si file for it:
		if version := si.Version(); len(version) == 0 || !version.OnOrAfter(util.VERSION_4_0) {
			panic("not implemented yet")
		}
	}
	if err = segnOutput.WriteStringStringMap(sis.userData); err != nil {
		return
	}
	sis.pendingSegnOutput = segnOutput
	success = true
	return nil
}
Ejemplo n.º 9
0
func (w *BlockTreeTermsWriter) WriteIndexHeader(out store.IndexOutput) error {
	return codec.WriteHeader(out, TERMS_INDEX_CODEC_NAME, TERMS_VERSION_CURRENT)
}
Ejemplo n.º 10
0
func NewCompressingStoredFieldsWriter(dir store.Directory, si *model.SegmentInfo,
	segmentSuffix string, ctx store.IOContext, formatName string,
	compressionMode CompressionMode, chunkSize int) (*CompressingStoredFieldsWriter, error) {

	assert(dir != nil)
	ans := &CompressingStoredFieldsWriter{
		directory:       dir,
		segment:         si.Name,
		segmentSuffix:   segmentSuffix,
		compressionMode: compressionMode,
		compressor:      compressionMode.NewCompressor(),
		chunkSize:       chunkSize,
		docBase:         0,
		bufferedDocs:    newGrowableByteArrayDataOutput(chunkSize),
		numStoredFields: make([]int, 16),
		endOffsets:      make([]int, 16),
		numBufferedDocs: 0,
	}

	var success = false
	indexStream, err := dir.CreateOutput(util.SegmentFileName(si.Name, segmentSuffix,
		lucene40.FIELDS_INDEX_EXTENSION), ctx)
	if err != nil {
		return nil, err
	}
	assert(indexStream != nil)
	defer func() {
		if !success {
			util.CloseWhileSuppressingError(indexStream)
			ans.Abort()
		}
	}()

	ans.fieldsStream, err = dir.CreateOutput(util.SegmentFileName(si.Name, segmentSuffix,
		lucene40.FIELDS_EXTENSION), ctx)
	if err != nil {
		return nil, err
	}

	codecNameIdx := formatName + CODEC_SFX_IDX
	codecNameDat := formatName + CODEC_SFX_DAT
	err = codec.WriteHeader(indexStream, codecNameIdx, VERSION_CURRENT)
	if err != nil {
		return nil, err
	}
	err = codec.WriteHeader(ans.fieldsStream, codecNameDat, VERSION_CURRENT)
	if err != nil {
		return nil, err
	}
	assert(int64(codec.HeaderLength(codecNameIdx)) == indexStream.FilePointer())
	assert(int64(codec.HeaderLength(codecNameDat)) == ans.fieldsStream.FilePointer())

	ans.indexWriter, err = NewStoredFieldsIndexWriter(indexStream)
	if err != nil {
		return nil, err
	}
	assert(ans.indexWriter != nil)
	indexStream = nil

	err = ans.fieldsStream.WriteVInt(int32(chunkSize))
	if err != nil {
		return nil, err
	}
	err = ans.fieldsStream.WriteVInt(packed.VERSION_CURRENT)
	if err != nil {
		return nil, err
	}

	success = true
	return ans, nil
}
Ejemplo n.º 11
0
/* Creates a postings writer with the specified PackedInts overhead ratio */
func newLucene41PostingsWriter(state *SegmentWriteState,
	accetableOverheadRatio float32) (*Lucene41PostingsWriter, error) {
	docOut, err := state.Directory.CreateOutput(
		util.SegmentFileName(state.SegmentInfo.Name,
			state.SegmentSuffix,
			LUCENE41_DOC_EXTENSION),
		state.Context)
	if err != nil {
		return nil, err
	}

	ans := new(Lucene41PostingsWriter)
	if err = func() error {
		var posOut store.IndexOutput
		var payOut store.IndexOutput
		var success = false
		defer func() {
			if !success {
				util.CloseWhileSuppressingError(docOut, posOut, payOut)
			}
		}()

		err := codec.WriteHeader(docOut, LUCENE41_DOC_CODEC, LUCENE41_VERSION_CURRENT)
		if err != nil {
			return err
		}
		ans.forUtil, err = NewForUtilInto(accetableOverheadRatio, docOut)
		if err != nil {
			return err
		}
		if state.FieldInfos.HasProx {
			ans.posDeltaBuffer = make([]int, MAX_DATA_SIZE)
			posOut, err = state.Directory.CreateOutput(util.SegmentFileName(
				state.SegmentInfo.Name, state.SegmentSuffix, LUCENE41_POS_EXTENSION),
				state.Context)
			if err != nil {
				return err
			}

			err = codec.WriteHeader(posOut, LUCENE41_POS_CODEC, LUCENE41_VERSION_CURRENT)
			if err != nil {
				return err
			}

			if state.FieldInfos.HasPayloads {
				ans.payloadBytes = make([]byte, 128)
				ans.payloadLengthBuffer = make([]int, MAX_DATA_SIZE)
			}

			if state.FieldInfos.HasOffsets {
				ans.offsetStartDeltaBuffer = make([]int, MAX_DATA_SIZE)
				ans.offsetLengthBuffer = make([]int, MAX_DATA_SIZE)
			}

			if state.FieldInfos.HasPayloads || state.FieldInfos.HasOffsets {
				payOut, err = state.Directory.CreateOutput(util.SegmentFileName(
					state.SegmentInfo.Name, state.SegmentSuffix, LUCENE41_PAY_EXTENSION),
					state.Context)
				if err != nil {
					return err
				}
				err = codec.WriteHeader(payOut, LUCENE41_PAY_CODEC, LUCENE41_VERSION_CURRENT)
			}
		}
		ans.payOut, ans.posOut = payOut, posOut
		ans.docOut = docOut
		success = true
		return nil
	}(); err != nil {
		return nil, err
	}

	ans.docDeltaBuffer = make([]int, MAX_DATA_SIZE)
	ans.freqBuffer = make([]int, MAX_DATA_SIZE)
	ans.encoded = make([]byte, MAX_ENCODED_SIZE)

	// TODO: should we try skipping every 2/4 blocks...?
	ans.skipWriter = NewSkipWriter(
		maxSkipLevels,
		LUCENE41_BLOCK_SIZE,
		state.SegmentInfo.DocCount(),
		ans.docOut,
		ans.posOut,
		ans.payOut)

	return ans, nil
}
Ejemplo n.º 12
0
	filename := util.SegmentFileName(segName, suffix, FI_EXTENSION)
	var output store.IndexOutput
	if output, err = dir.CreateOutput(filename, ctx); err != nil {
		return
	}

	var success = false
	defer func() {
		if success {
			err = output.Close()
		} else {
			util.CloseWhileSuppressingError(output)
		}
	}()

	if err = codec.WriteHeader(output, FI_CODEC_NAME, FI_FORMAT_CURRENT); err != nil {
		return
	}
	if err = output.WriteVInt(int32(infos.Size())); err != nil {
		return
	}
	for _, fi := range infos.Values {
		indexOptions := fi.IndexOptions()
		bits := byte(0)
		if fi.HasVectors() {
			bits |= FI_STORE_TERMVECTOR
		}
		if fi.OmitsNorms() {
			bits |= FI_OMIT_NORMS
		}
		if fi.HasPayloads() {
Ejemplo n.º 13
0
func (sis *SegmentInfos) write(directory store.Directory) error {
	segmentsFilename := sis.nextSegmentFilename()

	// Always advance the generation on write:
	if sis.generation == -1 {
		sis.generation = 1
	} else {
		sis.generation++
	}

	var segnOutput *store.ChecksumIndexOutput
	var success = false
	var upgradedSIFiles = make(map[string]bool)

	defer func() {
		if !success {
			// We hit an error above; try to close the file but suppress
			// any errors
			util.CloseWhileSuppressingError(segnOutput)

			for filename, _ := range upgradedSIFiles {
				directory.DeleteFile(filename) // ignore error
			}

			// Try not to leave a truncated segments_N fle in the index:
			directory.DeleteFile(segmentsFilename) // ignore error
		}
	}()

	out, err := directory.CreateOutput(segmentsFilename, store.IO_CONTEXT_DEFAULT)
	if err != nil {
		return err
	}
	segnOutput = store.NewChecksumIndexOutput(out)
	err = codec.WriteHeader(segnOutput, "segments", VERSION_40)
	if err != nil {
		return err
	}
	err = segnOutput.WriteLong(sis.version)
	if err == nil {
		err = segnOutput.WriteInt(int32(sis.counter))
		if err == nil {
			err = segnOutput.WriteInt(int32(len(sis.Segments)))
		}
	}
	if err != nil {
		return err
	}
	for _, siPerCommit := range sis.Segments {
		si := siPerCommit.info
		err = segnOutput.WriteString(si.Name)
		if err == nil {
			err = segnOutput.WriteString(si.Codec().(Codec).Name())
			if err == nil {
				err = segnOutput.WriteLong(siPerCommit.delGen)
				if err == nil {
					err = segnOutput.WriteInt(int32(siPerCommit.delCount))
				}
			}
		}
		if err != nil {
			return err
		}
		assert(si.Dir == directory)

		assert(siPerCommit.delCount <= si.DocCount())

		// If this segment is pre-4.x, perform a one-time "upgrade" to
		// write the .si file for it:
		if version := si.Version(); version == "" || versionLess(version, "4.0") {
			panic("not implemented yet")
		}
	}
	err = segnOutput.WriteStringStringMap(sis.userData)
	if err != nil {
		return err
	}
	sis.pendingSegnOutput = segnOutput
	success = true
	return nil
}