mirror of
https://github.com/dolthub/dolt.git
synced 2026-01-29 19:10:13 -06:00
110 lines
2.1 KiB
Go
110 lines
2.1 KiB
Go
package chunks
|
|
|
|
import (
|
|
"bytes"
|
|
"crypto/sha1"
|
|
"encoding/binary"
|
|
"io"
|
|
"sync"
|
|
|
|
"github.com/attic-labs/noms/d"
|
|
"github.com/attic-labs/noms/ref"
|
|
)
|
|
|
|
/*
|
|
Chunk Serialization:
|
|
Chunk 0
|
|
Chunk 1
|
|
..
|
|
Chunk N
|
|
|
|
Chunk:
|
|
Ref // 20-byte sha1 hash
|
|
Len // 4-byte int
|
|
Data // len(Data) == Len
|
|
*/
|
|
|
|
// Creates a serializer which is a ChunkSink. Put() chunks will be serialized to |writer|. Close() must be called when no more chunks will be serialized.
|
|
func NewSerializer(writer io.Writer) ChunkSink {
|
|
s := &serializer{
|
|
writer,
|
|
make(chan Chunk, 64),
|
|
make(chan struct{}),
|
|
}
|
|
|
|
go func() {
|
|
for chunk := range s.chs {
|
|
d.Chk.NotNil(chunk.Data)
|
|
|
|
digest := chunk.Ref().Digest()
|
|
n, err := io.Copy(s.writer, bytes.NewReader(digest[:]))
|
|
d.Chk.NoError(err)
|
|
d.Chk.Equal(int64(sha1.Size), n)
|
|
|
|
// Because of chunking at higher levels, no chunk should never be more than 4GB
|
|
chunkSize := uint32(len(chunk.Data()))
|
|
err = binary.Write(s.writer, binary.LittleEndian, chunkSize)
|
|
d.Chk.NoError(err)
|
|
|
|
n, err = io.Copy(s.writer, bytes.NewReader(chunk.Data()))
|
|
d.Chk.NoError(err)
|
|
d.Chk.Equal(uint32(n), chunkSize)
|
|
}
|
|
|
|
s.done <- struct{}{}
|
|
}()
|
|
|
|
return s
|
|
}
|
|
|
|
type serializer struct {
|
|
writer io.Writer
|
|
chs chan Chunk
|
|
done chan struct{}
|
|
}
|
|
|
|
func (sz *serializer) Put(c Chunk) {
|
|
sz.chs <- c
|
|
}
|
|
|
|
func (sz *serializer) Close() error {
|
|
close(sz.chs)
|
|
<-sz.done
|
|
return nil
|
|
}
|
|
|
|
// Deserialize reads off of |reader| until EOF, sending chunks to |cs|.
|
|
func Deserialize(reader io.Reader, cs ChunkSink) {
|
|
wg := sync.WaitGroup{}
|
|
|
|
for {
|
|
digest := ref.Sha1Digest{}
|
|
n, err := io.ReadFull(reader, digest[:])
|
|
if err == io.EOF {
|
|
break
|
|
}
|
|
d.Chk.NoError(err)
|
|
d.Chk.Equal(int(sha1.Size), n)
|
|
r := ref.New(digest)
|
|
|
|
chunkSize := uint32(0)
|
|
err = binary.Read(reader, binary.LittleEndian, &chunkSize)
|
|
d.Chk.NoError(err)
|
|
|
|
w := NewChunkWriter()
|
|
n2, err := io.CopyN(w, reader, int64(chunkSize))
|
|
d.Chk.NoError(err)
|
|
d.Chk.Equal(int64(chunkSize), n2)
|
|
c := w.Chunk()
|
|
d.Chk.Equal(r, c.Ref())
|
|
|
|
wg.Add(1)
|
|
go func() {
|
|
cs.Put(c)
|
|
wg.Done()
|
|
}()
|
|
}
|
|
|
|
wg.Wait()
|
|
}
|