func TestClone(t *testing.T) {
fmt.Println("Testing Loading FST...")
path := "../search/testdata/belfrysample"
d, err := OpenFSDirectory(path)
if err != nil {
t.Error(err)
}
ctx := NewIOContextBool(false)
in, err := d.OpenInput("_0_Lucene41_0.tip", ctx)
if err != nil {
t.Error(err)
}
version, err := codec.CheckHeader(in, "BLOCK_TREE_TERMS_INDEX", 0, 1)
var indexDirOffset int64 = 0
if version < 1 {
indexDirOffset, err = in.ReadLong()
if err != nil {
t.Error(err)
}
} else { // >= 1
in.Seek(in.Length() - 8)
indexDirOffset, err = in.ReadLong()
if err != nil {
t.Error(err)
}
}
fmt.Println("indexDirOffset:", indexDirOffset)
in.Seek(indexDirOffset)
indexStartFP, err := in.ReadVLong()
if err != nil {
t.Error(err)
}
fmt.Println("indexStartFP:", indexStartFP)
fmt.Println("Before clone", in)
clone := in.Clone()
fmt.Println("After clone", clone)
if _, ok := clone.(*SimpleFSIndexInput); !ok {
t.Error("Clone() should return *SimpleFSIndexInput.")
}
clone.Seek(indexStartFP)
fmt.Println("After clone.Seek()", clone)
_, err = codec.CheckHeader(clone, "FST", 3, 4)
if err != nil {
t.Error(err)
}
// panic("bingo")
// _, err = util.LoadFST(in, util.ByteSequenceOutputsSingleton())
// if err != nil {
// t.Error(err)
// }
}
func NewLucene41PostingsReader(dir store.Directory, fis FieldInfos, si SegmentInfo,
ctx store.IOContext, segmentSuffix string) (r PostingsReaderBase, err error) {
log.Print("Initializing Lucene41PostingsReader...")
success := false
var docIn, posIn, payIn store.IndexInput = nil, nil, nil
defer func() {
if !success {
log.Print("Failed to initialize Lucene41PostingsReader.")
if err != nil {
log.Print("DEBUG ", err)
}
util.CloseWhileSuppressingError(docIn, posIn, payIn)
}
}()
docIn, err = dir.OpenInput(util.SegmentFileName(si.name, segmentSuffix, LUCENE41_DOC_EXTENSION), ctx)
if err != nil {
return r, err
}
_, err = codec.CheckHeader(docIn, LUCENE41_DOC_CODEC, LUCENE41_VERSION_CURRENT, LUCENE41_VERSION_CURRENT)
if err != nil {
return r, err
}
forUtil, err := NewForUtil(docIn)
if err != nil {
return r, err
}
if fis.hasProx {
posIn, err = dir.OpenInput(util.SegmentFileName(si.name, segmentSuffix, LUCENE41_POS_EXTENSION), ctx)
if err != nil {
return r, err
}
_, err = codec.CheckHeader(posIn, LUCENE41_POS_CODEC, LUCENE41_VERSION_CURRENT, LUCENE41_VERSION_CURRENT)
if err != nil {
return r, err
}
if fis.hasPayloads || fis.hasOffsets {
payIn, err = dir.OpenInput(util.SegmentFileName(si.name, segmentSuffix, LUCENE41_PAY_EXTENSION), ctx)
if err != nil {
return r, err
}
_, err = codec.CheckHeader(payIn, LUCENE41_PAY_CODEC, LUCENE41_VERSION_CURRENT, LUCENE41_VERSION_CURRENT)
if err != nil {
return r, err
}
}
}
success = true
return &Lucene41PostingsReader{docIn, posIn, payIn, forUtil}, nil
}
func newLucene42DocValuesProducer(state SegmentReadState,
dataCodec, dataExtension, metaCodec, metaExtension string) (dvp *Lucene42DocValuesProducer, err error) {
dvp = &Lucene42DocValuesProducer{}
dvp.maxDoc = int(state.segmentInfo.docCount)
metaName := util.SegmentFileName(state.segmentInfo.name, state.segmentSuffix, metaExtension)
// read in the entries from the metadata file.
in, err := state.dir.OpenInput(metaName, state.context)
if err != nil {
return dvp, err
}
success := false
defer func() {
if success {
err = util.Close(in)
} else {
util.CloseWhileSuppressingError(in)
}
}()
version, err := codec.CheckHeader(in, metaCodec, LUCENE42_DV_VERSION_START, LUCENE42_DV_VERSION_CURRENT)
if err != nil {
return dvp, err
}
dvp.numerics = make(map[int]NumericEntry)
dvp.binaries = make(map[int]BinaryEntry)
dvp.fsts = make(map[int]FSTEntry)
err = dvp.readFields(in, state.fieldInfos)
if err != nil {
return dvp, err
}
success = true
success = false
dataName := util.SegmentFileName(state.segmentInfo.name, state.segmentSuffix, dataExtension)
dvp.data, err = state.dir.OpenInput(dataName, state.context)
if err != nil {
return dvp, err
}
version2, err := codec.CheckHeader(dvp.data, dataCodec, LUCENE42_DV_VERSION_START, LUCENE42_DV_VERSION_CURRENT)
if err != nil {
return dvp, err
}
if version != version2 {
return dvp, errors.New("Format versions mismatch")
}
return dvp, nil
}
func (r *BlockTreeTermsReader) readIndexHeader(input store.IndexInput) (version int, err error) {
version, err = asInt(codec.CheckHeader(input, BTT_INDEX_CODEC_NAME, BTT_INDEX_VERSION_START, BTT_INDEX_VERSION_CURRENT))
if err != nil {
return version, err
}
if version < BTT_INDEX_VERSION_APPEND_ONLY {
r.indexDirOffset, err = input.ReadLong()
if err != nil {
return version, err
}
}
return version, nil
}
func newPackedReader(in DataInput) (r PackedIntsReader, err error) {
if version, err := codec.CheckHeader(in, PACKED_CODEC_NAME, PACKED_VERSION_START, PACKED_VERSION_CURRENT); err == nil {
if bitsPerValue, err := asUint32(in.ReadVInt()); err == nil {
// assert bitsPerValue > 0 && bitsPerValue <= 64
if valueCount, err := in.ReadVInt(); err == nil {
if id, err := in.ReadVInt(); err == nil {
format := PackedFormat(id)
return NewPackedReaderNoHeader(in, format, version, valueCount, bitsPerValue)
}
}
}
}
return
}
func (r *Lucene41PostingsReader) Init(termsIn store.IndexInput) error {
log.Printf("Initializing from: %v", termsIn)
// Make sure we are talking to the matching postings writer
_, err := codec.CheckHeader(termsIn, LUCENE41_TERMS_CODEC, LUCENE41_VERSION_START, LUCENE41_VERSION_CURRENT)
if err != nil {
return err
}
indexBlockSize, err := termsIn.ReadVInt()
if err != nil {
return err
}
log.Printf("Index block size: %v", indexBlockSize)
if indexBlockSize != LUCENE41_BLOCK_SIZE {
panic(fmt.Sprintf("index-time BLOCK_SIZE (%v) != read-time BLOCK_SIZE (%v)", indexBlockSize, LUCENE41_BLOCK_SIZE))
}
return nil
}
func TestReadingFromSlicedIndexInputOSX(t *testing.T) {
path := "../search/testdata/osx/belfrysample"
d, err := OpenFSDirectory(path)
if err != nil {
t.Error(err)
}
ctx := NewIOContextBool(false)
cd, err := NewCompoundFileDirectory(d, "_0.cfs", ctx, false)
name := util.SegmentFileName("_0", "Lucene41_0", "pos")
posIn, err := cd.OpenInput(name, ctx)
if err != nil {
t.Error(err)
}
t.Log(posIn)
codec.CheckHeader(posIn, "Lucene41PostingsWriterPos", 0, 0)
// codec header mismatch: actual header=0 vs expected header=1071082519 (resource: SlicedIndexInput(SlicedIndexInput(_0_Lucene41_0.pos in SimpleFSIndexInput(path='/private/tmp/kc/index/belfrysample/_0.cfs')) in SimpleFSIndexInput(path='/private/tmp/kc/index/belfrysample/_0.cfs') slice=1461:3426))
}
func TestCheckHeaderWin8(t *testing.T) {
path := "../search/testdata/win8/belfrysample"
d, err := OpenFSDirectory(path)
if err != nil {
t.Error(err)
}
ctx := NewIOContextBool(false)
cd, err := NewCompoundFileDirectory(d, "_0.cfs", ctx, false)
if err != nil {
t.Error(err)
}
r, err := cd.OpenInput("_0_Lucene41_0.pos", ctx)
_, err = codec.CheckHeader(r, "Lucene41PostingsWriterPos", 0, 0)
if err != nil {
t.Error(err)
}
}
if err != nil {
return fi, err
}
log.Printf("Reading %v", input)
success := false
defer func() {
if success {
input.Close()
} else {
util.CloseWhileHandlingError(err, input)
}
}()
_, err = codec.CheckHeader(input,
LUCENE42_FI_CODEC_NAME,
LUCENE42_FI_FORMAT_START,
LUCENE42_FI_FORMAT_CURRENT)
if err != nil {
return fi, err
}
size, err := input.ReadVInt() //read in the size
if err != nil {
return fi, err
}
log.Printf("Found %v FieldInfos.", size)
infos := make([]FieldInfo, size)
for i, _ := range infos {
name, err := input.ReadString()
if err != nil {
// CompressingStoredFieldsReader.java L90
func newCompressingStoredFieldsReader(d store.Directory, si SegmentInfo, segmentSuffix string, fn FieldInfos,
ctx store.IOContext, formatName string, compressionMode codec.CompressionMode) (r *CompressingStoredFieldsReader, err error) {
r = &CompressingStoredFieldsReader{}
r.compressionMode = compressionMode
segment := si.name
r.fieldInfos = fn
r.numDocs = int(si.docCount)
var indexStream store.IndexInput
success := false
defer func() {
if !success {
log.Println("Failed to initialize CompressionStoredFieldsReader.")
if err != nil {
log.Print(err)
}
util.Close(r, indexStream)
}
}()
// Load the index into memory
indexStreamFN := util.SegmentFileName(segment, segmentSuffix, LUCENE40_SF_FIELDS_INDEX_EXTENSION)
indexStream, err = d.OpenInput(indexStreamFN, ctx)
if err != nil {
return nil, err
}
codecNameIdx := formatName + CODEC_SFX_IDX
codec.CheckHeader(indexStream, codecNameIdx, CODEC_SFX_VERSION_START, CODEC_SFX_VERSION_CURRENT)
if int64(codec.HeaderLength(codecNameIdx)) != indexStream.FilePointer() {
panic("assert fail")
}
r.indexReader, err = newCompressingStoredFieldsIndexReader(indexStream, si)
if err != nil {
return nil, err
}
err = indexStream.Close()
if err != nil {
return nil, err
}
indexStream = nil
// Open the data file and read metadata
fieldsStreamFN := util.SegmentFileName(segment, segmentSuffix, LUCENE40_SF_FIELDS_EXTENSION)
r.fieldsStream, err = d.OpenInput(fieldsStreamFN, ctx)
if err != nil {
return nil, err
}
codecNameDat := formatName + CODEC_SFX_DAT
codec.CheckHeader(r.fieldsStream, codecNameDat, CODEC_SFX_VERSION_START, CODEC_SFX_VERSION_CURRENT)
if int64(codec.HeaderLength(codecNameDat)) != r.fieldsStream.FilePointer() {
panic("assert fail")
}
n, err := r.fieldsStream.ReadVInt()
if err != nil {
return nil, err
}
r.packedIntsVersion = int(n)
r.decompressor = compressionMode.NewDecompressor()
r.bytes = make([]byte, 0)
success = true
return r, nil
}
/** Load a previously saved FST; maxBlockBits allows you to
* control the size of the byte[] pages used to hold the FST bytes. */
func loadFST3(in DataInput, outputs Outputs, maxBlockBits uint32) (fst *FST, err error) {
log.Printf("Loading FST from %v and output to %v...", in, outputs)
defer func() {
if err != nil {
log.Print("Failed to load FST.")
log.Printf("DEBUG ", err)
}
}()
fst = &FST{outputs: outputs, startNode: -1}
if maxBlockBits < 1 || maxBlockBits > 30 {
panic(fmt.Sprintf("maxBlockBits should 1..30; got %v", maxBlockBits))
}
// NOTE: only reads most recent format; we don't have
// back-compat promise for FSTs (they are experimental):
fst.version, err = codec.CheckHeader(in, FST_FILE_FORMAT_NAME, FST_VERSION_PACKED, FST_VERSION_VINT_TARGET)
if err != nil {
return fst, err
}
if b, err := in.ReadByte(); err == nil {
fst.packed = (b == 1)
} else {
return fst, err
}
if b, err := in.ReadByte(); err == nil {
if b == 1 {
// accepts empty string
// 1 KB blocks:
emptyBytes := newBytesStoreFromBits(10)
if numBytes, err := in.ReadVInt(); err == nil {
log.Printf("Number of bytes: %v", numBytes)
emptyBytes.CopyBytes(in, int64(numBytes))
// De-serialize empty-string output:
var reader BytesReader
if fst.packed {
log.Printf("Forward reader.")
reader = emptyBytes.forwardReader()
} else {
log.Printf("Reverse reader.")
reader = emptyBytes.reverseReader()
// NoOutputs uses 0 bytes when writing its output,
// so we have to check here else BytesStore gets
// angry:
if numBytes > 0 {
reader.setPosition(int64(numBytes - 1))
}
}
log.Printf("Reading final output from %v to %v...", reader, outputs)
fst.emptyOutput, err = outputs.ReadFinalOutput(reader)
}
} // else emptyOutput = nil
}
if err != nil {
return fst, err
}
if t, err := in.ReadByte(); err == nil {
switch t {
case 0:
fst.inputType = INPUT_TYPE_BYTE1
case 1:
fst.inputType = INPUT_TYPE_BYTE2
case 2:
fst.inputType = INPUT_TYPE_BYTE4
default:
panic(fmt.Sprintf("invalid input type %v", t))
}
}
if err != nil {
return fst, err
}
if fst.packed {
fst.nodeRefToAddress, err = newPackedReader(in)
if err != nil {
return fst, err
}
} // else nodeRefToAddress = nil
if fst.startNode, err = in.ReadVLong(); err == nil {
if fst.nodeCount, err = in.ReadVLong(); err == nil {
if fst.arcCount, err = in.ReadVLong(); err == nil {
if fst.arcWithOutputCount, err = in.ReadVLong(); err == nil {
if numBytes, err := in.ReadVLong(); err == nil {
if fst.bytes, err = newBytesStoreFromInput(in, numBytes, 1<<maxBlockBits); err == nil {
log.Println("DEBUG ", outputs.NoOutput())
fst.NO_OUTPUT = outputs.NoOutput()
fst.cacheRootArcs()
// NOTE: bogus because this is only used during
// building; we need to break out mutable FST from
// immutable
fst.allowArrayArcs = false
}
}
}
}
}
}
return fst, err
}
func readEntries(handle IndexInputSlicer, dir Directory, name string) (mapping map[string]FileEntry, err error) {
var stream, entriesStream IndexInput = nil, nil
defer func() {
err = util.CloseWhileHandlingError(err, stream, entriesStream)
}()
// read the first VInt. If it is negative, it's the version number
// otherwise it's the count (pre-3.1 indexes)
mapping = make(map[string]FileEntry)
stream = handle.openFullSlice()
log.Printf("Reading from stream: %v", stream)
firstInt, err := stream.ReadVInt()
if err != nil {
return mapping, err
}
// impossible for 3.0 to have 63 files in a .cfs, CFS writer was not visible
// and separate norms/etc are outside of cfs.
if firstInt == int32(CODEC_MAGIC_BYTE1) {
if secondByte, err := stream.ReadByte(); err == nil {
if thirdByte, err := stream.ReadByte(); err == nil {
if fourthByte, err := stream.ReadByte(); err == nil {
if secondByte != CODEC_MAGIC_BYTE2 ||
thirdByte != CODEC_MAGIC_BYTE3 ||
fourthByte != CODEC_MAGIC_BYTE4 {
return mapping, errors.New(fmt.Sprintf(
"Illegal/impossible header for CFS file: %v,%v,%v",
secondByte, thirdByte, fourthByte))
}
}
}
}
if err != nil {
return mapping, err
}
_, err = codec.CheckHeaderNoMagic(stream, CFD_DATA_CODEC, CFD_VERSION_START, CFD_VERSION_START)
if err != nil {
return mapping, err
}
entriesFileName := util.SegmentFileName(util.StripExtension(name), "", COMPOUND_FILE_ENTRIES_EXTENSION)
entriesStream, err = dir.OpenInput(entriesFileName, IO_CONTEXT_READONCE)
if err != nil {
return mapping, err
}
_, err = codec.CheckHeader(entriesStream, CFD_ENTRY_CODEC, CFD_VERSION_START, CFD_VERSION_START)
if err != nil {
return mapping, err
}
numEntries, err := entriesStream.ReadVInt()
if err != nil {
return mapping, err
}
log.Printf("Entries number: %v", numEntries)
for i := int32(0); i < numEntries; i++ {
id, err := entriesStream.ReadString()
if err != nil {
return mapping, err
}
if _, ok := mapping[id]; ok {
return mapping, errors.New(fmt.Sprintf(
"Duplicate cfs entry id=%v in CFS: %v", id, entriesStream))
}
log.Printf("Found entry: %v", id)
offset, err := entriesStream.ReadLong()
if err != nil {
return mapping, err
}
length, err := entriesStream.ReadLong()
if err != nil {
return mapping, err
}
mapping[id] = FileEntry{offset, length}
}
} else {
// TODO remove once 3.x is not supported anymore
panic("not supported yet; will also be obsolete soon")
}
return mapping, nil
}
fileName := util.SegmentFileName(segment, "", LUCENE40_SI_EXTENSION)
input, err := dir.OpenInput(fileName, context)
if err != nil {
return si, err
}
success := false
defer func() {
if !success {
util.CloseWhileSuppressingError(input)
} else {
input.Close()
}
}()
_, err = codec.CheckHeader(input, LUCENE40_CODEC_NAME, LUCENE40_VERSION_START, LUCENE40_VERSION_CURRENT)
if err != nil {
return si, err
}
version, err := input.ReadString()
if err != nil {
return si, err
}
docCount, err := input.ReadInt()
if err != nil {
return si, err
}
if docCount < 0 {
return si, errors.New(fmt.Sprintf("invalid docCount: %v (resource=%v)", docCount, input))
}
sicf, err := input.ReadByte()
