Include repeated files as hard-links in the tarstream

Signed-off-by: Paul "TBBle" Hampson <Paul.Hampson@Pobox.com>

internal/ociwclayer/export.go

@@ -50,6 +50,8 @@ func ExportLayer(w io.Writer, path string, parentLayerPaths []string) error {
 }
 
 func writeTarFromLayer(r hcsshim.LayerReader, w io.Writer) error {
+	linkRecords := make(map[[16]byte]string)
+
 	t := tar.NewWriter(w)
 	for {
 		name, size, fileInfo, err := r.Next()
@@ -69,6 +71,27 @@ func writeTarFromLayer(r hcsshim.LayerReader, w io.Writer) error {
 				return err
 			}
 		} else {
+			fileIDInfo, err := r.LinkInfo()
+			if err != nil {
+				return err
+			}
+			if linkName, ok := linkRecords[fileIDInfo.FileID]; ok {
+				// We've seen this file before, by another name, so put a hardlink in the tar stream.
+				hdr := backuptar.BasicInfoHeader(name, 0, fileInfo)
+				hdr.Mode = 0644
+				hdr.Typeflag = tar.TypeLink
+				hdr.Linkname = linkName
+				if err := t.WriteHeader(hdr); err != nil {
+					return err
+				}
+				continue
+			}
+
+			// All subsequent names for this file will be hard-linked to this name
+			// Optimisation opportunity: Only store records for files with multiple links.
+			// e.g., go-winio could expose FileStandardInfo to enable this.
+			linkRecords[fileIDInfo.FileID] = filepath.ToSlash(name)
+
 			err = backuptar.WriteTarFileFromBackupStream(t, r, name, size, fileInfo)
 			if err != nil {
 				return err

internal/wclayer/baselayerreader.go

@@ -186,6 +186,11 @@ func (r *baseLayerReader) Next() (path string, size int64, fileInfo *winio.FileB
 	return
 }
 
+func (r *baseLayerReader) LinkInfo() (fileIDInfo *winio.FileIDInfo, err error) {
+	fileIDInfo, err = winio.GetFileID(r.currentFile)
+	return
+}
+
 func (r *baseLayerReader) Read(b []byte) (int, error) {
 	if r.backupReader == nil {
 		if r.currentFile == nil {

internal/wclayer/exportlayer.go

@@ -44,6 +44,8 @@ func ExportLayer(ctx context.Context, path string, exportFolderPath string, pare
 type LayerReader interface {
 	// Next advances to the next file and returns the name, size, and file info
 	Next() (string, int64, *winio.FileBasicInfo, error)
+	// LinkInfo returns the file identifier for the current file.
+	LinkInfo() (*winio.FileIDInfo, error)
 	// Read reads data from the current file, in the format of a Win32 backup stream, and
 	// returns the number of bytes read.
 	Read(b []byte) (int, error)

internal/wclayer/legacy.go

@@ -294,6 +294,11 @@ func (r *legacyLayerReader) Next() (path string, size int64, fileInfo *winio.Fil
 	return
 }
 
+func (r *legacyLayerReader) LinkInfo() (fileIDInfo *winio.FileIDInfo, err error) {
+	fileIDInfo, err = winio.GetFileID(r.currentFile)
+	return
+}
+
 func (r *legacyLayerReader) Read(b []byte) (int, error) {
 	if r.backupReader == nil {
 		if r.currentFile == nil {