seaweedfs/weed/storage/needle/needle_read_write.go

420 lines
12 KiB
Go
Raw Normal View History

2019-04-19 12:43:36 +08:00
package needle
2012-12-21 13:57:03 +08:00
import (
2012-12-21 14:32:21 +08:00
"errors"
2013-01-17 16:22:18 +08:00
"fmt"
"io"
2012-12-21 13:57:03 +08:00
"os"
2019-04-19 12:43:36 +08:00
"math"
"github.com/joeslay/seaweedfs/weed/glog"
"github.com/joeslay/seaweedfs/weed/storage/memory_map"
. "github.com/joeslay/seaweedfs/weed/storage/types"
"github.com/joeslay/seaweedfs/weed/util"
2012-12-21 13:57:03 +08:00
)
const (
FlagGzip = 0x01
FlagHasName = 0x02
FlagHasMime = 0x04
FlagHasLastModifiedDate = 0x08
FlagHasTtl = 0x10
FlagHasPairs = 0x20
2015-12-15 14:38:58 +08:00
FlagIsChunkManifest = 0x80
LastModifiedBytesLength = 5
TtlBytesLength = 2
)
2018-07-24 16:36:04 +08:00
func (n *Needle) DiskSize(version Version) int64 {
return GetActualSize(n.Size, version)
}
2017-01-07 02:22:20 +08:00
func (n *Needle) prepareWriteBuffer(version Version) ([]byte, uint32, int64, error) {
writeBytes := make([]byte, 0)
2013-01-17 16:22:18 +08:00
switch version {
case Version1:
2019-04-19 15:39:34 +08:00
header := make([]byte, NeedleHeaderSize)
CookieToBytes(header[0:CookieSize], n.Cookie)
NeedleIdToBytes(header[CookieSize:CookieSize+NeedleIdSize], n.Id)
2012-12-21 16:36:55 +08:00
n.Size = uint32(len(n.Data))
util.Uint32toBytes(header[CookieSize+NeedleIdSize:CookieSize+NeedleIdSize+SizeSize], n.Size)
size := n.Size
actualSize := NeedleHeaderSize + int64(n.Size)
writeBytes = append(writeBytes, header...)
writeBytes = append(writeBytes, n.Data...)
2018-07-24 16:36:04 +08:00
padding := PaddingLength(n.Size, version)
util.Uint32toBytes(header[0:NeedleChecksumSize], n.Checksum.Value())
writeBytes = append(writeBytes, header[0:NeedleChecksumSize+padding]...)
return writeBytes, size, actualSize, nil
2018-07-24 16:36:04 +08:00
case Version2, Version3:
2019-04-19 15:39:34 +08:00
header := make([]byte, NeedleHeaderSize+TimestampSize) // adding timestamp to reuse it and avoid extra allocation
CookieToBytes(header[0:CookieSize], n.Cookie)
NeedleIdToBytes(header[CookieSize:CookieSize+NeedleIdSize], n.Id)
if len(n.Name) >= math.MaxUint8 {
2018-12-11 12:57:51 +08:00
n.NameSize = math.MaxUint8
} else {
n.NameSize = uint8(len(n.Name))
}
n.DataSize, n.MimeSize = uint32(len(n.Data)), uint8(len(n.Mime))
if n.DataSize > 0 {
n.Size = 4 + n.DataSize + 1
if n.HasName() {
n.Size = n.Size + 1 + uint32(n.NameSize)
}
if n.HasMime() {
n.Size = n.Size + 1 + uint32(n.MimeSize)
}
if n.HasLastModifiedDate() {
n.Size = n.Size + LastModifiedBytesLength
}
if n.HasTtl() {
n.Size = n.Size + TtlBytesLength
}
if n.HasPairs() {
n.Size += 2 + uint32(n.PairsSize)
}
} else {
n.Size = 0
}
util.Uint32toBytes(header[CookieSize+NeedleIdSize:CookieSize+NeedleIdSize+SizeSize], n.Size)
writeBytes = append(writeBytes, header[0:NeedleHeaderSize]...)
if n.DataSize > 0 {
util.Uint32toBytes(header[0:4], n.DataSize)
writeBytes = append(writeBytes, header[0:4]...)
writeBytes = append(writeBytes, n.Data...)
util.Uint8toBytes(header[0:1], n.Flags)
writeBytes = append(writeBytes, header[0:1]...)
if n.HasName() {
util.Uint8toBytes(header[0:1], n.NameSize)
writeBytes = append(writeBytes, header[0:1]...)
writeBytes = append(writeBytes, n.Name[:n.NameSize]...)
}
if n.HasMime() {
util.Uint8toBytes(header[0:1], n.MimeSize)
writeBytes = append(writeBytes, header[0:1]...)
writeBytes = append(writeBytes, n.Mime...)
}
if n.HasLastModifiedDate() {
util.Uint64toBytes(header[0:8], n.LastModified)
writeBytes = append(writeBytes, header[8-LastModifiedBytesLength:8]...)
}
2015-07-09 14:21:53 +08:00
if n.HasTtl() && n.Ttl != nil {
n.Ttl.ToBytes(header[0:TtlBytesLength])
writeBytes = append(writeBytes, header[0:TtlBytesLength]...)
}
if n.HasPairs() {
util.Uint16toBytes(header[0:2], n.PairsSize)
writeBytes = append(writeBytes, header[0:2]...)
writeBytes = append(writeBytes, n.Pairs...)
}
}
2018-07-24 16:36:04 +08:00
padding := PaddingLength(n.Size, version)
util.Uint32toBytes(header[0:NeedleChecksumSize], n.Checksum.Value())
2018-07-24 16:36:04 +08:00
if version == Version2 {
writeBytes = append(writeBytes, header[0:NeedleChecksumSize+padding]...)
2018-07-24 16:36:04 +08:00
} else {
// version3
util.Uint64toBytes(header[NeedleChecksumSize:NeedleChecksumSize+TimestampSize], n.AppendAtNs)
writeBytes = append(writeBytes, header[0:NeedleChecksumSize+TimestampSize+padding]...)
2018-07-24 16:36:04 +08:00
}
2017-01-07 02:22:20 +08:00
return writeBytes, n.DataSize, GetActualSize(n.Size, version), nil
2012-12-21 16:36:55 +08:00
}
return writeBytes, 0, 0, fmt.Errorf("Unsupported Version! (%d)", version)
}
func (n *Needle) Append(w *os.File, version Version) (offset uint64, size uint32, actualSize int64, err error) {
mMap, exists := memory_map.FileMemoryMap[w.Name()]
if !exists {
if end, e := w.Seek(0, io.SeekEnd); e == nil {
defer func(w *os.File, off int64) {
if err != nil {
if te := w.Truncate(end); te != nil {
glog.V(0).Infof("Failed to truncate %s back to %d with error: %v", w.Name(), end, te)
}
}
}(w, end)
offset = uint64(end)
} else {
err = fmt.Errorf("Cannot Read Current Volume Position: %v", e)
return
}
} else {
offset = uint64(mMap.End_of_file + 1)
}
bytesToWrite, size, actualSize, err := n.prepareWriteBuffer(version)
if err == nil {
if exists {
mMap.WriteMemory(offset, uint64(len(bytesToWrite)), bytesToWrite)
} else {
_, err = w.Write(bytesToWrite)
}
}
return offset, size, actualSize, err
2012-12-21 13:57:03 +08:00
}
2018-07-24 16:36:04 +08:00
func ReadNeedleBlob(r *os.File, offset int64, size uint32, version Version) (dataSlice []byte, err error) {
dataSize := GetActualSize(size, version)
dataSlice = make([]byte, dataSize)
mMap, exists := memory_map.FileMemoryMap[r.Name()]
if exists {
mBuffer, err := mMap.ReadMemory(uint64(offset), uint64(dataSize))
copy(dataSlice, mBuffer.Buffer)
mBuffer.ReleaseMemory()
return dataSlice, err
} else {
_, err = r.ReadAt(dataSlice, offset)
return dataSlice, err
}
}
2019-05-28 02:59:03 +08:00
// ReadBytes hydrates the needle from the bytes buffer, with only n.Id is set.
func (n *Needle) ReadBytes(bytes []byte, offset int64, size uint32, version Version) (err error) {
n.ParseNeedleHeader(bytes)
if n.Size != size {
return fmt.Errorf("entry not found: offset %d found id %d size %d, expected size %d", offset, n.Id, n.Size, size)
}
2013-01-17 16:22:18 +08:00
switch version {
case Version1:
2019-04-19 15:39:34 +08:00
n.Data = bytes[NeedleHeaderSize : NeedleHeaderSize+size]
2018-07-24 16:36:04 +08:00
case Version2, Version3:
2019-04-19 15:39:34 +08:00
err = n.readNeedleDataVersion2(bytes[NeedleHeaderSize : NeedleHeaderSize+int(n.Size)])
2012-12-21 13:57:03 +08:00
}
2019-04-30 11:22:19 +08:00
if err != nil && err != io.EOF {
2019-03-19 20:34:43 +08:00
return err
}
if size > 0 {
checksum := util.BytesToUint32(bytes[NeedleHeaderSize+size : NeedleHeaderSize+size+NeedleChecksumSize])
newChecksum := NewCRC(n.Data)
if checksum != newChecksum.Value() {
return errors.New("CRC error! Data On Disk Corrupted")
}
n.Checksum = newChecksum
}
2018-07-24 16:36:04 +08:00
if version == Version3 {
2019-04-19 15:39:34 +08:00
tsOffset := NeedleHeaderSize + size + NeedleChecksumSize
2018-07-24 16:38:08 +08:00
n.AppendAtNs = util.BytesToUint64(bytes[tsOffset : tsOffset+TimestampSize])
2018-07-24 16:36:04 +08:00
}
return nil
2012-12-21 13:57:03 +08:00
}
2019-05-28 02:59:03 +08:00
// ReadData hydrates the needle from the file, with only n.Id is set.
func (n *Needle) ReadData(r *os.File, offset int64, size uint32, version Version) (err error) {
bytes, err := ReadNeedleBlob(r, offset, size, version)
if err != nil {
return err
}
return n.ReadBytes(bytes, offset, size, version)
}
func (n *Needle) ParseNeedleHeader(bytes []byte) {
n.Cookie = BytesToCookie(bytes[0:CookieSize])
2018-07-24 16:38:08 +08:00
n.Id = BytesToNeedleId(bytes[CookieSize : CookieSize+NeedleIdSize])
2019-04-19 15:39:34 +08:00
n.Size = util.BytesToUint32(bytes[CookieSize+NeedleIdSize : NeedleHeaderSize])
}
2019-03-19 20:34:43 +08:00
func (n *Needle) readNeedleDataVersion2(bytes []byte) (err error) {
2012-12-21 18:20:38 +08:00
index, lenBytes := 0, len(bytes)
if index < lenBytes {
2018-07-24 16:38:08 +08:00
n.DataSize = util.BytesToUint32(bytes[index : index+4])
2012-12-21 18:20:38 +08:00
index = index + 4
if int(n.DataSize)+index > lenBytes {
return fmt.Errorf("index out of range %d", 1)
}
2018-07-24 16:38:08 +08:00
n.Data = bytes[index : index+int(n.DataSize)]
2012-12-21 18:20:38 +08:00
index = index + int(n.DataSize)
n.Flags = bytes[index]
index = index + 1
}
if index < lenBytes && n.HasName() {
n.NameSize = uint8(bytes[index])
index = index + 1
if int(n.NameSize)+index > lenBytes {
return fmt.Errorf("index out of range %d", 2)
2019-03-19 20:34:43 +08:00
}
2018-07-24 16:38:08 +08:00
n.Name = bytes[index : index+int(n.NameSize)]
index = index + int(n.NameSize)
}
if index < lenBytes && n.HasMime() {
n.MimeSize = uint8(bytes[index])
index = index + 1
if int(n.MimeSize)+index > lenBytes {
return fmt.Errorf("index out of range %d", 3)
2019-03-19 20:34:43 +08:00
}
2018-07-24 16:38:08 +08:00
n.Mime = bytes[index : index+int(n.MimeSize)]
index = index + int(n.MimeSize)
}
if index < lenBytes && n.HasLastModifiedDate() {
if LastModifiedBytesLength+index > lenBytes {
return fmt.Errorf("index out of range %d", 4)
2019-03-19 20:34:43 +08:00
}
2018-07-24 16:38:08 +08:00
n.LastModified = util.BytesToUint64(bytes[index : index+LastModifiedBytesLength])
index = index + LastModifiedBytesLength
}
if index < lenBytes && n.HasTtl() {
if TtlBytesLength+index > lenBytes {
return fmt.Errorf("index out of range %d", 5)
2019-03-19 20:34:43 +08:00
}
2018-07-24 16:38:08 +08:00
n.Ttl = LoadTTLFromBytes(bytes[index : index+TtlBytesLength])
index = index + TtlBytesLength
}
if index < lenBytes && n.HasPairs() {
if 2+index > lenBytes {
return fmt.Errorf("index out of range %d", 6)
2019-03-19 20:34:43 +08:00
}
2018-07-24 16:38:08 +08:00
n.PairsSize = util.BytesToUint16(bytes[index : index+2])
index += 2
if int(n.PairsSize)+index > lenBytes {
return fmt.Errorf("index out of range %d", 7)
2019-03-19 20:34:43 +08:00
}
end := index + int(n.PairsSize)
n.Pairs = bytes[index:end]
index = end
}
2019-03-19 20:34:43 +08:00
return nil
}
func ReadNeedleHeader(r *os.File, version Version, offset int64) (n *Needle, bytes []byte, bodyLength int64, err error) {
2012-12-21 14:32:21 +08:00
n = new(Needle)
2018-07-24 16:36:04 +08:00
if version == Version1 || version == Version2 || version == Version3 {
2019-04-19 15:39:34 +08:00
bytes = make([]byte, NeedleHeaderSize)
mMap, exists := memory_map.FileMemoryMap[r.Name()]
if exists {
mem_buffer, err := mMap.ReadMemory(uint64(offset), NeedleHeaderSize)
copy(bytes, mem_buffer.Buffer)
mem_buffer.ReleaseMemory()
if err != nil {
return nil, bytes, 0, err
}
} else {
var count int
count, err = r.ReadAt(bytes, offset)
if count <= 0 || err != nil {
return nil, bytes, 0, err
}
2012-12-21 14:32:21 +08:00
}
n.ParseNeedleHeader(bytes)
2018-07-24 16:36:04 +08:00
bodyLength = NeedleBodyLength(n.Size, version)
2012-12-21 16:36:55 +08:00
}
2012-12-21 16:36:55 +08:00
return
}
2018-07-24 16:36:04 +08:00
func PaddingLength(needleSize uint32, version Version) uint32 {
if version == Version3 {
// this is same value as version2, but just listed here for clarity
2019-04-19 15:39:34 +08:00
return NeedlePaddingSize - ((NeedleHeaderSize + needleSize + NeedleChecksumSize + TimestampSize) % NeedlePaddingSize)
2018-07-24 16:36:04 +08:00
}
2019-04-19 15:39:34 +08:00
return NeedlePaddingSize - ((NeedleHeaderSize + needleSize + NeedleChecksumSize) % NeedlePaddingSize)
2018-07-24 16:36:04 +08:00
}
func NeedleBodyLength(needleSize uint32, version Version) int64 {
if version == Version3 {
return int64(needleSize) + NeedleChecksumSize + TimestampSize + int64(PaddingLength(needleSize, version))
}
return int64(needleSize) + NeedleChecksumSize + int64(PaddingLength(needleSize, version))
}
2012-12-21 16:36:55 +08:00
//n should be a needle already read the header
//the input stream will read until next file entry
func (n *Needle) ReadNeedleBody(r *os.File, version Version, offset int64, bodyLength int64) (bytes []byte, err error) {
2019-03-18 14:28:43 +08:00
if bodyLength <= 0 {
return nil, nil
}
2019-04-19 10:22:13 +08:00
bytes = make([]byte, bodyLength)
if _, err = r.ReadAt(bytes, offset); err != nil {
return
}
err = n.ReadNeedleBodyBytes(bytes, version)
return
}
func (n *Needle) ReadNeedleBodyBytes(needleBody []byte, version Version) (err error) {
if len(needleBody) <= 0 {
return nil
}
2013-01-17 16:22:18 +08:00
switch version {
case Version1:
2019-04-19 10:22:13 +08:00
n.Data = needleBody[:n.Size]
2012-12-21 16:36:55 +08:00
n.Checksum = NewCRC(n.Data)
2018-07-24 16:36:04 +08:00
case Version2, Version3:
2019-04-19 10:22:13 +08:00
err = n.readNeedleDataVersion2(needleBody[0:n.Size])
n.Checksum = NewCRC(n.Data)
2019-03-18 14:28:43 +08:00
if version == Version3 {
2019-03-19 20:34:43 +08:00
tsOffset := n.Size + NeedleChecksumSize
2019-04-19 10:22:13 +08:00
n.AppendAtNs = util.BytesToUint64(needleBody[tsOffset : tsOffset+TimestampSize])
2019-03-18 14:28:43 +08:00
}
default:
2019-04-19 10:22:13 +08:00
err = fmt.Errorf("unsupported version %d!", version)
2012-12-21 13:57:03 +08:00
}
2012-12-21 14:32:21 +08:00
return
2012-12-21 13:57:03 +08:00
}
func (n *Needle) IsGzipped() bool {
return n.Flags&FlagGzip > 0
}
func (n *Needle) SetGzipped() {
n.Flags = n.Flags | FlagGzip
}
func (n *Needle) HasName() bool {
return n.Flags&FlagHasName > 0
}
func (n *Needle) SetHasName() {
n.Flags = n.Flags | FlagHasName
}
func (n *Needle) HasMime() bool {
return n.Flags&FlagHasMime > 0
}
func (n *Needle) SetHasMime() {
n.Flags = n.Flags | FlagHasMime
}
func (n *Needle) HasLastModifiedDate() bool {
return n.Flags&FlagHasLastModifiedDate > 0
}
func (n *Needle) SetHasLastModifiedDate() {
n.Flags = n.Flags | FlagHasLastModifiedDate
}
func (n *Needle) HasTtl() bool {
return n.Flags&FlagHasTtl > 0
}
func (n *Needle) SetHasTtl() {
n.Flags = n.Flags | FlagHasTtl
}
func (n *Needle) IsChunkedManifest() bool {
2015-12-15 14:38:58 +08:00
return n.Flags&FlagIsChunkManifest > 0
}
2015-12-15 14:38:58 +08:00
func (n *Needle) SetIsChunkManifest() {
n.Flags = n.Flags | FlagIsChunkManifest
}
func (n *Needle) HasPairs() bool {
return n.Flags&FlagHasPairs != 0
}
func (n *Needle) SetHasPairs() {
n.Flags = n.Flags | FlagHasPairs
}
2019-04-19 12:43:36 +08:00
func GetActualSize(size uint32, version Version) int64 {
2019-04-19 15:39:34 +08:00
return NeedleHeaderSize + NeedleBodyLength(size, version)
2019-04-19 12:43:36 +08:00
}