Merge pull request #64 from jstarks/tar_ea

Add support for EAs to Windows tars

Author: jstarks

backup.go

@@ -68,10 +68,20 @@ func NewBackupStreamReader(r io.Reader) *BackupStreamReader {
 	return &BackupStreamReader{r, 0}
 }
 
-// Next returns the next backup stream and prepares for calls to Write(). It skips the remainder of the current stream if
+// Next returns the next backup stream and prepares for calls to Read(). It skips the remainder of the current stream if
 // it was not completely read.
 func (r *BackupStreamReader) Next() (*BackupHeader, error) {
 	if r.bytesLeft > 0 {
+		if s, ok := r.r.(io.Seeker); ok {
+			// Make sure Seek on io.SeekCurrent sometimes succeeds
+			// before trying the actual seek.
+			if _, err := s.Seek(0, io.SeekCurrent); err == nil {
+				if _, err = s.Seek(r.bytesLeft, io.SeekCurrent); err != nil {
+					return nil, err
+				}
+				r.bytesLeft = 0
+			}
+		}
 		if _, err := io.Copy(ioutil.Discard, r); err != nil {
 			return nil, err
 		}
@@ -220,7 +230,7 @@ type BackupFileWriter struct {
 	ctx             uintptr
 }
 
-// NewBackupFileWrtier returns a new BackupFileWriter from a file handle. If includeSecurity is true,
+// NewBackupFileWriter returns a new BackupFileWriter from a file handle. If includeSecurity is true,
 // Write() will attempt to restore the security descriptor from the stream.
 func NewBackupFileWriter(f *os.File, includeSecurity bool) *BackupFileWriter {
 	w := &BackupFileWriter{f, includeSecurity, 0}

backuptar/tar.go

@@ -36,6 +36,7 @@ const (
 	hdrSecurityDescriptor    = "sd"
 	hdrRawSecurityDescriptor = "rawsd"
 	hdrMountPoint            = "mountpoint"
+	hdrEaPrefix              = "xattr."
 )
 
 func writeZeroes(w io.Writer, count int64) error {
@@ -118,6 +119,21 @@ func BasicInfoHeader(name string, size int64, fileInfo *winio.FileBasicInfo) *ta
 func WriteTarFileFromBackupStream(t *tar.Writer, r io.Reader, name string, size int64, fileInfo *winio.FileBasicInfo) error {
 	name = filepath.ToSlash(name)
 	hdr := BasicInfoHeader(name, size, fileInfo)
+
+	// If r can be seeked, then this function is two-pass: pass 1 collects the
+	// tar header data, and pass 2 copies the data stream. If r cannot be
+	// seeked, then some header data (in particular EAs) will be silently lost.
+	var (
+		restartPos int64
+		err        error
+	)
+	sr, readTwice := r.(io.Seeker)
+	if readTwice {
+		if restartPos, err = sr.Seek(0, io.SeekCurrent); err != nil {
+			readTwice = false
+		}
+	}
+
 	br := winio.NewBackupStreamReader(r)
 	var dataHdr *winio.BackupHeader
 	for dataHdr == nil {
@@ -131,7 +147,9 @@ func WriteTarFileFromBackupStream(t *tar.Writer, r io.Reader, name string, size
 		switch bhdr.Id {
 		case winio.BackupData:
 			hdr.Mode |= c_ISREG
-			dataHdr = bhdr
+			if !readTwice {
+				dataHdr = bhdr
+			}
 		case winio.BackupSecurity:
 			sd, err := ioutil.ReadAll(br)
 			if err != nil {
@@ -151,18 +169,54 @@ func WriteTarFileFromBackupStream(t *tar.Writer, r io.Reader, name string, size
 				hdr.Winheaders[hdrMountPoint] = "1"
 			}
 			hdr.Linkname = rp.Target
-		case winio.BackupEaData, winio.BackupLink, winio.BackupPropertyData, winio.BackupObjectId, winio.BackupTxfsData:
+
+		case winio.BackupEaData:
+			eab, err := ioutil.ReadAll(br)
+			if err != nil {
+				return err
+			}
+			eas, err := winio.DecodeExtendedAttributes(eab)
+			if err != nil {
+				return err
+			}
+			for _, ea := range eas {
+				// Use base64 encoding for the binary value. Note that there
+				// is no way to encode the EA's flags, since their use doesn't
+				// make any sense for persisted EAs.
+				hdr.Winheaders[hdrEaPrefix+ea.Name] = base64.StdEncoding.EncodeToString(ea.Value)
+			}
+
+		case winio.BackupAlternateData, winio.BackupLink, winio.BackupPropertyData, winio.BackupObjectId, winio.BackupTxfsData:
 			// ignore these streams
 		default:
 			return fmt.Errorf("%s: unknown stream ID %d", name, bhdr.Id)
 		}
 	}
 
-	err := t.WriteHeader(hdr)
+	err = t.WriteHeader(hdr)
 	if err != nil {
 		return err
 	}
 
+	if readTwice {
+		// Get back to the data stream.
+		if _, err = sr.Seek(restartPos, io.SeekStart); err != nil {
+			return err
+		}
+		for dataHdr == nil {
+			bhdr, err := br.Next()
+			if err == io.EOF {
+				break
+			}
+			if err != nil {
+				return err
+			}
+			if bhdr.Id == winio.BackupData {
+				dataHdr = bhdr
+			}
+		}
+	}
+
 	if dataHdr != nil {
 		// A data stream was found. Copy the data.
 		if (dataHdr.Attributes & winio.StreamSparseAttributes) == 0 {
@@ -293,6 +347,38 @@ func WriteBackupStreamFromTarFile(w io.Writer, t *tar.Reader, hdr *tar.Header) (
 			return nil, err
 		}
 	}
+	var eas []winio.ExtendedAttribute
+	for k, v := range hdr.Winheaders {
+		if !strings.HasPrefix(k, hdrEaPrefix) {
+			continue
+		}
+		data, err := base64.StdEncoding.DecodeString(v)
+		if err != nil {
+			return nil, err
+		}
+		eas = append(eas, winio.ExtendedAttribute{
+			Name:  k[len(hdrEaPrefix):],
+			Value: data,
+		})
+	}
+	if len(eas) != 0 {
+		eadata, err := winio.EncodeExtendedAttributes(eas)
+		if err != nil {
+			return nil, err
+		}
+		bhdr := winio.BackupHeader{
+			Id:   winio.BackupEaData,
+			Size: int64(len(eadata)),
+		}
+		err = bw.WriteHeader(&bhdr)
+		if err != nil {
+			return nil, err
+		}
+		_, err = bw.Write(eadata)
+		if err != nil {
+			return nil, err
+		}
+	}
 	if hdr.Typeflag == tar.TypeSymlink {
 		_, isMountPoint := hdr.Winheaders[hdrMountPoint]
 		rp := winio.ReparsePoint{

backuptar/tar_test.go

@@ -80,5 +80,5 @@ func TestRoundTrip(t *testing.T) {
 		t.Errorf("got %#v, expected %#v", *bi, *bi2)
 	}
 
-	ensurePresent(t, hdr.Winheaders, "fileattr", "sd")
+	ensurePresent(t, hdr.Winheaders, "fileattr", "rawsd")
 }

ea.go

@@ -0,0 +1,137 @@
+package winio
+
+import (
+	"bytes"
+	"encoding/binary"
+	"errors"
+)
+
+type fileFullEaInformation struct {
+	NextEntryOffset uint32
+	Flags           uint8
+	NameLength      uint8
+	ValueLength     uint16
+}
+
+var (
+	fileFullEaInformationSize = binary.Size(&fileFullEaInformation{})
+
+	errInvalidEaBuffer = errors.New("invalid extended attribute buffer")
+	errEaNameTooLarge  = errors.New("extended attribute name too large")
+	errEaValueTooLarge = errors.New("extended attribute value too large")
+)
+
+// ExtendedAttribute represents a single Windows EA.
+type ExtendedAttribute struct {
+	Name  string
+	Value []byte
+	Flags uint8
+}
+
+func parseEa(b []byte) (ea ExtendedAttribute, nb []byte, err error) {
+	var info fileFullEaInformation
+	err = binary.Read(bytes.NewReader(b), binary.LittleEndian, &info)
+	if err != nil {
+		err = errInvalidEaBuffer
+		return
+	}
+
+	nameOffset := fileFullEaInformationSize
+	nameLen := int(info.NameLength)
+	valueOffset := nameOffset + int(info.NameLength) + 1
+	valueLen := int(info.ValueLength)
+	nextOffset := int(info.NextEntryOffset)
+	if valueLen+valueOffset > len(b) || nextOffset < 0 || nextOffset > len(b) {
+		err = errInvalidEaBuffer
+		return
+	}
+
+	ea.Name = string(b[nameOffset : nameOffset+nameLen])
+	ea.Value = b[valueOffset : valueOffset+valueLen]
+	ea.Flags = info.Flags
+	if info.NextEntryOffset != 0 {
+		nb = b[info.NextEntryOffset:]
+	}
+	return
+}
+
+// DecodeExtendedAttributes decodes a list of EAs from a FILE_FULL_EA_INFORMATION
+// buffer retrieved from BackupRead, ZwQueryEaFile, etc.
+func DecodeExtendedAttributes(b []byte) (eas []ExtendedAttribute, err error) {
+	for len(b) != 0 {
+		ea, nb, err := parseEa(b)
+		if err != nil {
+			return nil, err
+		}
+
+		eas = append(eas, ea)
+		b = nb
+	}
+	return
+}
+
+func writeEa(buf *bytes.Buffer, ea *ExtendedAttribute, last bool) error {
+	if int(uint8(len(ea.Name))) != len(ea.Name) {
+		return errEaNameTooLarge
+	}
+	if int(uint16(len(ea.Value))) != len(ea.Value) {
+		return errEaValueTooLarge
+	}
+	entrySize := uint32(fileFullEaInformationSize + len(ea.Name) + 1 + len(ea.Value))
+	withPadding := (entrySize + 3) &^ 3
+	nextOffset := uint32(0)
+	if !last {
+		nextOffset = withPadding
+	}
+	info := fileFullEaInformation{
+		NextEntryOffset: nextOffset,
+		Flags:           ea.Flags,
+		NameLength:      uint8(len(ea.Name)),
+		ValueLength:     uint16(len(ea.Value)),
+	}
+
+	err := binary.Write(buf, binary.LittleEndian, &info)
+	if err != nil {
+		return err
+	}
+
+	_, err = buf.Write([]byte(ea.Name))
+	if err != nil {
+		return err
+	}
+
+	err = buf.WriteByte(0)
+	if err != nil {
+		return err
+	}
+
+	_, err = buf.Write(ea.Value)
+	if err != nil {
+		return err
+	}
+
+	_, err = buf.Write([]byte{0, 0, 0}[0 : withPadding-entrySize])
+	if err != nil {
+		return err
+	}
+
+	return nil
+}
+
+// EncodeExtendedAttributes encodes a list of EAs into a FILE_FULL_EA_INFORMATION
+// buffer for use with BackupWrite, ZwSetEaFile, etc.
+func EncodeExtendedAttributes(eas []ExtendedAttribute) ([]byte, error) {
+	var buf bytes.Buffer
+	for i := range eas {
+		last := false
+		if i == len(eas)-1 {
+			last = true
+		}
+
+		err := writeEa(&buf, &eas[i], last)
+		if err != nil {
+			return nil, err
+		}
+	}
+	return buf.Bytes(), nil
+}