image: optimize file searching

walkFunc is inefficient. If we knows which file to read, there
is no need to call w.walk().
For imageLayout, w.find() will improve the file searching speed

Signed-off-by: Ma Shimiao <[email protected]>

Author: Ma Shimiao

image/config.go

@@ -20,7 +20,6 @@ import (
 	"fmt"
 	"io"
 	"io/ioutil"
-	"os"
 	"path/filepath"
 	"strconv"
 	"strings"
@@ -35,10 +34,7 @@ func findConfig(w walker, d *v1.Descriptor) (*v1.Image, error) {
 	var c v1.Image
 	cpath := filepath.Join("blobs", string(d.Digest.Algorithm()), d.Digest.Hex())
 
-	switch err := w.walk(func(path string, info os.FileInfo, r io.Reader) error {
-		if info.IsDir() || filepath.Clean(path) != cpath {
-			return nil
-		}
+	switch err := w.find(cpath, func(path string, r io.Reader) error {
 		buf, err := ioutil.ReadAll(r)
 		if err != nil {
 			return errors.Wrapf(err, "%s: error reading config", path)

image/descriptor.go

@@ -53,12 +53,9 @@ func listReferences(w walker) ([]v1.Descriptor, error) {
 func findDescriptor(w walker, names []string) ([]v1.Descriptor, error) {
 	var descs []v1.Descriptor
 	var index v1.Index
+	dpath := "index.json"
 
-	if err := w.walk(func(path string, info os.FileInfo, r io.Reader) error {
-		if info.IsDir() || filepath.Clean(path) != indexPath {
-			return nil
-		}
-
+	if err := w.find(dpath, func(path string, r io.Reader) error {
 		if err := json.NewDecoder(r).Decode(&index); err != nil {
 			return err
 		}

image/manifest.go

@@ -39,11 +39,7 @@ func findManifest(w walker, d *v1.Descriptor) (*v1.Manifest, error) {
 	var m v1.Manifest
 	mpath := filepath.Join("blobs", string(d.Digest.Algorithm()), d.Digest.Hex())
 
-	switch err := w.walk(func(path string, info os.FileInfo, r io.Reader) error {
-		if info.IsDir() || filepath.Clean(path) != mpath {
-			return nil
-		}
-
+	switch err := w.find(mpath, func(path string, r io.Reader) error {
 		buf, err := ioutil.ReadAll(r)
 		if err != nil {
 			return errors.Wrapf(err, "%s: error reading manifest", path)
@@ -108,18 +104,10 @@ func unpackManifest(m *v1.Manifest, w walker, dest string) (retErr error) {
 		}
 	}()
 	for _, d := range m.Layers {
-		switch err := w.walk(func(path string, info os.FileInfo, r io.Reader) error {
-			if info.IsDir() {
-				return nil
-			}
-
-			dd, err := filepath.Rel(filepath.Join("blobs", string(d.Digest.Algorithm())), filepath.Clean(path))
-			if err != nil || d.Digest.Hex() != dd {
-				return nil
-			}
-
+		lpath := filepath.Join("blobs", string(d.Digest.Algorithm()), d.Digest.Hex())
+		switch err := w.find(lpath, func(path string, r io.Reader) error {
 			if err := unpackLayer(d.MediaType, path, dest, r); err != nil {
-				return errors.Wrap(err, "error unpack: extracting layer")
+				return errors.Wrap(err, "unpack: error extracting layer")
 			}
 
 			return errEOW

image/walker.go

@@ -34,6 +34,8 @@ var (
 // walkFunc is a function type that gets called for each file or directory visited by the Walker.
 type walkFunc func(path string, _ os.FileInfo, _ io.Reader) error
 
+type findFunc func(path string, r io.Reader) error
+
 // walker is the interface that defines how to access a given archival format
 type walker interface {
 
@@ -43,6 +45,9 @@ type walker interface {
 	// get will copy an arbitrary blob, defined by desc, in to dst. returns
 	// the number of bytes copied on success.
 	get(desc v1.Descriptor, dst io.Writer) (int64, error)
+
+	// find calls findFunc for handling content of path
+	find(path string, ff findFunc) error
 }
 
 // tarWalker exposes access to image layouts in a tar file.
@@ -120,6 +125,34 @@ func (w *tarWalker) get(desc v1.Descriptor, dst io.Writer) (int64, error) {
 	return bytes, nil
 }
 
+func (w *tarWalker) find(path string, ff findFunc) error {
+	done := false
+
+	f := func(relpath string, info os.FileInfo, rdr io.Reader) error {
+		var err error
+		if done {
+			return nil
+		}
+
+		if filepath.Clean(relpath) == path && !info.IsDir() {
+			if err = ff(relpath, rdr); err != nil {
+				return err
+			}
+			done = true
+		}
+		return nil
+	}
+
+	if err := w.walk(f); err != nil {
+		return errors.Wrapf(err, "find failed: unable to walk")
+	}
+	if !done {
+		return os.ErrNotExist
+	}
+
+	return nil
+}
+
 type eofReader struct{}
 
 func (eofReader) Read(_ []byte) (int, error) {
@@ -188,6 +221,27 @@ func (w *pathWalker) get(desc v1.Descriptor, dst io.Writer) (int64, error) {
 	return nbytes, nil
 }
 
+func (w *pathWalker) find(path string, ff findFunc) error {
+	name := filepath.Join(w.root, path)
+
+	info, err := os.Stat(name)
+	if err != nil {
+		return err
+	}
+
+	if info.IsDir() {
+		return fmt.Errorf("object is dir")
+	}
+
+	file, err := os.Open(name)
+	if err != nil {
+		return errors.Wrap(err, "unable to open file") // os.Open includes the path
+	}
+	defer file.Close()
+
+	return ff(name, file)
+}
+
 type zipWalker struct {
 	fileName string
 }
@@ -249,3 +303,31 @@ func (w *zipWalker) get(desc v1.Descriptor, dst io.Writer) (int64, error) {
 
 	return bytes, nil
 }
+
+func (w *zipWalker) find(path string, ff findFunc) error {
+	done := false
+
+	f := func(relpath string, info os.FileInfo, rdr io.Reader) error {
+		var err error
+		if done {
+			return nil
+		}
+
+		if filepath.Clean(relpath) == path && !info.IsDir() {
+			if err = ff(relpath, rdr); err != nil {
+				return err
+			}
+			done = true
+		}
+		return nil
+	}
+
+	if err := w.walk(f); err != nil {
+		return errors.Wrapf(err, "find failed: unable to walk")
+	}
+	if !done {
+		return os.ErrNotExist
+	}
+
+	return nil
+}