formats/packfile: new reader API

13 files changed, 347 insertions, 372 deletions

diff --git a/common/hash.go b/common/hash.go
new file mode 100644
index 0000000..83844c7
--- /dev/null
+++ b/common/hash.go
@@ -0,0 +1,35 @@
+package common
+
+import (
+	"crypto/sha1"
+	"encoding/hex"
+	"strconv"
+)
+
+// Hash SHA1 hased content
+type Hash [20]byte
+
+// ComputeHash compute the hash for a given ObjectType and content
+func ComputeHash(t ObjectType, content []byte) Hash {
+	h := t.Bytes()
+	h = append(h, ' ')
+	h = strconv.AppendInt(h, int64(len(content)), 10)
+	h = append(h, 0)
+	h = append(h, content...)
+
+	return Hash(sha1.Sum(h))
+}
+
+// NewHash return a new Hash from a hexadecimal hash representation
+func NewHash(s string) Hash {
+	b, _ := hex.DecodeString(s)
+
+	var h Hash
+	copy(h[:], b)
+
+	return h
+}
+
+func (h Hash) String() string {
+	return hex.EncodeToString(h[:])
+}
diff --git a/common/hash_test.go b/common/hash_test.go
new file mode 100644
index 0000000..cee0c0f
--- /dev/null
+++ b/common/hash_test.go
@@ -0,0 +1,27 @@
+package common
+
+import (
+	"testing"
+
+	. "gopkg.in/check.v1"
+)
+
+func Test(t *testing.T) { TestingT(t) }
+
+type HashSuite struct{}
+
+var _ = Suite(&HashSuite{})
+
+func (s *HashSuite) TestComputeHash(c *C) {
+	hash := ComputeHash(BlobObject, []byte(""))
+	c.Assert(hash.String(), Equals, "e69de29bb2d1d6434b8b29ae775ad8c2e48c5391")
+
+	hash = ComputeHash(BlobObject, []byte("Hello, World!\n"))
+	c.Assert(hash.String(), Equals, "8ab686eafeb1f44702738c8b0f24f2567c36da6d")
+}
+
+func (s *HashSuite) TestNewHash(c *C) {
+	hash := ComputeHash(BlobObject, []byte("Hello, World!\n"))
+
+	c.Assert(hash, Equals, NewHash(hash.String()))
+}
diff --git a/common/object.go b/common/object.go
new file mode 100644
index 0000000..60c44da
--- /dev/null
+++ b/common/object.go
@@ -0,0 +1,95 @@
+package common
+
+import (
+	"bytes"
+	"io"
+)
+
+// Object is a generic representation of any git object
+type Object interface {
+	Type() ObjectType
+	SetType(ObjectType)
+	Size() int64
+	SetSize(int64)
+	Hash() Hash
+	Reader() io.Reader
+	Writer() io.Writer
+}
+
+// ObjectStorage generic storage of objects
+type ObjectStorage interface {
+	New() Object
+	Set(Object) Hash
+	Get(Hash) (Object, bool)
+}
+
+// ObjectType internal object type's
+type ObjectType int8
+
+const (
+	CommitObject   ObjectType = 1
+	TreeObject     ObjectType = 2
+	BlobObject     ObjectType = 3
+	TagObject      ObjectType = 4
+	OFSDeltaObject ObjectType = 6
+	REFDeltaObject ObjectType = 7
+)
+
+func (t ObjectType) String() string {
+	switch t {
+	case CommitObject:
+		return "commit"
+	case TreeObject:
+		return "tree"
+	case BlobObject:
+		return "blob"
+	default:
+		return "-"
+	}
+}
+
+func (t ObjectType) Bytes() []byte {
+	return []byte(t.String())
+}
+
+type RAWObject struct {
+	b []byte
+	t ObjectType
+	s int64
+}
+
+func (o *RAWObject) Type() ObjectType     { return o.t }
+func (o *RAWObject) SetType(t ObjectType) { o.t = t }
+func (o *RAWObject) Size() int64          { return o.s }
+func (o *RAWObject) SetSize(s int64)      { o.s = s }
+func (o *RAWObject) Reader() io.Reader    { return bytes.NewBuffer(o.b) }
+func (o *RAWObject) Hash() Hash           { return ComputeHash(o.t, o.b) }
+func (o *RAWObject) Writer() io.Writer    { return o }
+func (o *RAWObject) Write(p []byte) (n int, err error) {
+	o.b = append(o.b, p...)
+	return len(p), nil
+}
+
+type RAWObjectStorage struct {
+	Objects map[Hash]*RAWObject
+}
+
+func NewRAWObjectStorage() *RAWObjectStorage {
+	return &RAWObjectStorage{make(map[Hash]*RAWObject, 0)}
+}
+
+func (o *RAWObjectStorage) New() Object {
+	return &RAWObject{}
+}
+
+func (o *RAWObjectStorage) Set(obj Object) Hash {
+	h := obj.Hash()
+	o.Objects[h] = obj.(*RAWObject)
+
+	return h
+}
+
+func (o *RAWObjectStorage) Get(h Hash) (Object, bool) {
+	obj, ok := o.Objects[h]
+	return obj, ok
+}
diff --git a/formats/packfile/common.go b/formats/packfile/common.go
index 4a97dc7..06c63d4 100644
--- a/formats/packfile/common.go
+++ b/formats/packfile/common.go
@@ -6,20 +6,17 @@ import (
 )
 
 type trackingReader struct {
-	r io.Reader
-	n int
+	r        io.Reader
+	position int64
 }
 
-func (t *trackingReader) Pos() int { return t.n }
-
 func (t *trackingReader) Read(p []byte) (n int, err error) {
 	n, err = t.r.Read(p)
 	if err != nil {
 		return 0, err
 	}
 
-	t.n += n
-
+	t.position += int64(n)
 	return n, err
 }
 
@@ -34,6 +31,6 @@ func (t *trackingReader) ReadByte() (c byte, err error) {
 		return 0, fmt.Errorf("read %d bytes, should have read just 1", n)
 	}
 
-	t.n += n // n is 1
+	t.position++
 	return p[0], nil
 }
diff --git a/formats/packfile/common_test.go b/formats/packfile/common_test.go
deleted file mode 100644
index 104a5d2..0000000
--- a/formats/packfile/common_test.go
+++ /dev/null
@@ -1,9 +0,0 @@
-package packfile
-
-import (
-	"testing"
-
-	. "gopkg.in/check.v1"
-)
-
-func Test(t *testing.T) { TestingT(t) }
diff --git a/formats/packfile/delta.go b/formats/packfile/delta.go
index 30703eb..571ccf8 100644
--- a/formats/packfile/delta.go
+++ b/formats/packfile/delta.go
@@ -2,7 +2,7 @@ package packfile
 
 import "io"
 
-const delta_size_min = 4
+const deltaSizeMin = 4
 
 func deltaHeaderSize(b []byte) (uint, []byte) {
 	var size, j uint
@@ -18,8 +18,8 @@ func deltaHeaderSize(b []byte) (uint, []byte) {
 	return size, b[j:]
 }
 
-func PatchDelta(src, delta []byte) []byte {
-	if len(delta) < delta_size_min {
+func patchDelta(src, delta []byte) []byte {
+	if len(delta) < deltaSizeMin {
 		return nil
 	}
 	size, delta := deltaHeaderSize(delta)
@@ -94,12 +94,13 @@ func PatchDelta(src, delta []byte) []byte {
 	return dest
 }
 
-func decodeOffset(src io.ByteReader, steps int) (int, error) {
+func decodeOffset(src io.ByteReader, steps int64) (int64, error) {
 	b, err := src.ReadByte()
 	if err != nil {
 		return 0, err
 	}
-	var offset = int(b & 0x7f)
+
+	var offset = int64(b & 0x7f)
 	for (b & 0x80) != 0 {
 		offset++ // WHY?
 		b, err = src.ReadByte()
@@ -107,7 +108,7 @@ func decodeOffset(src io.ByteReader, steps int) (int, error) {
 			return 0, err
 		}
 
-		offset = (offset << 7) + int(b&0x7f)
+		offset = (offset << 7) + int64(b&0x7f)
 	}
 
 	// offset needs to be aware of the bytes we read for `o.typ` and `o.size`
diff --git a/formats/packfile/doc.go b/formats/packfile/doc.go
index 1fc28da..cb3f542 100644
--- a/formats/packfile/doc.go
+++ b/formats/packfile/doc.go
@@ -1,8 +1,5 @@
 package packfile
 
-// Code from:
-// https://github.com/gitchain/gitchain/tree/master/git @ 4c2fabdf9
-//
 // GIT pack format
 // ===============
 //
diff --git a/formats/packfile/packfile.go b/formats/packfile/packfile.go
deleted file mode 100644
index d70f396..0000000
--- a/formats/packfile/packfile.go
+++ /dev/null
@@ -1,82 +0,0 @@
-package packfile
-
-import "fmt"
-
-type Packfile struct {
-	Version     uint32
-	Size        int64
-	ObjectCount int
-	Checksum    []byte
-	Commits     map[Hash]*Commit
-	Trees       map[Hash]*Tree
-	Blobs       map[Hash]*Blob
-}
-
-func NewPackfile() *Packfile {
-	return &Packfile{
-		Commits: make(map[Hash]*Commit, 0),
-		Trees:   make(map[Hash]*Tree, 0),
-		Blobs:   make(map[Hash]*Blob, 0),
-	}
-}
-
-type BlobEntry struct {
-	path string
-	*Blob
-}
-
-type SubtreeEntry struct {
-	path string
-	*Tree
-	TreeCh
-}
-
-type treeEntry interface {
-	isTreeEntry()
-	Path() string
-}
-
-func (b BlobEntry) isTreeEntry()    {}
-func (b BlobEntry) Path() string    { return b.path }
-func (b SubtreeEntry) isTreeEntry() {}
-func (b SubtreeEntry) Path() string { return b.path }
-
-type TreeCh <-chan treeEntry
-
-func (p *Packfile) WalkCommit(commitHash Hash) (TreeCh, error) {
-	commit, ok := p.Commits[commitHash]
-	if !ok {
-		return nil, fmt.Errorf("Unable to find %q commit", commitHash)
-	}
-
-	return p.WalkTree(p.Trees[commit.Tree]), nil
-}
-
-func (p *Packfile) WalkTree(tree *Tree) TreeCh {
-	return p.walkTree(tree, "")
-}
-
-func (p *Packfile) walkTree(tree *Tree, pathPrefix string) TreeCh {
-	ch := make(chan treeEntry)
-
-	if tree == nil {
-		close(ch)
-		return ch
-	}
-
-	go func() {
-		defer func() {
-			close(ch)
-		}()
-		for _, e := range tree.Entries {
-			path := pathPrefix + e.Name
-			if blob, ok := p.Blobs[e.Hash]; ok {
-				ch <- BlobEntry{path, blob}
-			} else if subtree, ok := p.Trees[e.Hash]; ok {
-				ch <- SubtreeEntry{path, subtree, p.walkTree(subtree, path+"/")}
-			}
-		}
-	}()
-
-	return ch
-}
diff --git a/formats/packfile/reader.go b/formats/packfile/reader.go
index c355e12..6ccf384 100644
--- a/formats/packfile/reader.go
+++ b/formats/packfile/reader.go
@@ -5,15 +5,29 @@ import (
 	"encoding/binary"
 	"fmt"
 	"io"
+	"io/ioutil"
+
+	"gopkg.in/src-d/go-git.v2/common"
 
 	"github.com/klauspost/compress/zlib"
 )
 
 type Format int
 
+var (
+	EmptyRepositoryErr        = newError("empty repository")
+	UnsupportedVersionErr     = newError("unsupported packfile version")
+	MaxObjectsLimitReachedErr = newError("max. objects limit reached")
+	MalformedPackfileErr      = newError("malformed pack file, does not start with 'PACK'")
+	InvalidObjectErr          = newError("invalid git object")
+	PatchingErr               = newError("patching error")
+	PackEntryNotFoundErr      = newError("can't find a pack entry")
+	ObjectNotFoundErr         = newError("can't find a object")
+	ZLibErr                   = newError("zlib reading error")
+)
+
 const (
 	DefaultMaxObjectsLimit = 1 << 20
-	DefaultMaxObjectSize   = 1 << 32 // 4GB
 
 	VersionSupported        = 2
 	UnknownFormat    Format = 0
@@ -21,7 +35,8 @@ const (
 	REFDeltaFormat   Format = 2
 )
 
-type PackfileReader struct {
+// Reader reads a packfile from a binary string splitting it on objects
+type Reader struct {
 	// MaxObjectsLimit is the limit of objects to be load in the packfile, if
 	// a packfile excess this number an error is throw, the default value
 	// is defined by DefaultMaxObjectsLimit, usually the default limit is more
@@ -29,116 +44,98 @@ type PackfileReader struct {
 	// where the number of object is bigger the memory can be exhausted.
 	MaxObjectsLimit uint32
 
-	// MaxObjectSize is the maximum size in bytes, reading objects with a bigger
-	// size cause a error. The default value is defined by DefaultMaxObjectSize
-	MaxObjectSize uint64
-
 	// Format specifies if we are using ref-delta's or ofs-delta's, choosing the
 	// correct format the memory usage is optimized
 	// https://github.com/git/git/blob/8d530c4d64ffcc853889f7b385f554d53db375ed/Documentation/technical/protocol-capabilities.txt#L154
 	Format Format
 
 	r       *trackingReader
-	objects map[Hash]*RAWObject
-	offsets map[int]*RAWObject
+	s       common.ObjectStorage
+	offsets map[int64]common.Hash
 }
 
-func NewPackfileReader(r io.Reader, fn ContentCallback) (*PackfileReader, error) {
-	return &PackfileReader{
+// NewReader returns a new Reader that reads from a io.Reader
+func NewReader(r io.Reader) *Reader {
+	return &Reader{
 		MaxObjectsLimit: DefaultMaxObjectsLimit,
-		MaxObjectSize:   DefaultMaxObjectSize,
 
 		r:       &trackingReader{r: r},
-		objects: make(map[Hash]*RAWObject, 0),
-		offsets: make(map[int]*RAWObject, 0),
-	}, nil
+		offsets: make(map[int64]common.Hash, 0),
+	}
 }
 
-func (pr *PackfileReader) Read() (chan *RAWObject, error) {
-	if err := pr.validateHeader(); err != nil {
+// Read reads the objects and stores it at the ObjectStorage
+func (r *Reader) Read(s common.ObjectStorage) (int64, error) {
+	r.s = s
+	if err := r.validateHeader(); err != nil {
 		if err == io.EOF {
-			// This is an empty repo. It's OK.
-			return nil, nil
+			return -1, EmptyRepositoryErr
 		}
 
-		return nil, err
+		return -1, err
 	}
 
-	version, err := pr.readInt32()
+	version, err := r.readInt32()
 	if err != nil {
-		return nil, err
+		return -1, err
 	}
 
 	if version > VersionSupported {
-		return nil, NewError("unsupported packfile version %d", version)
+		return -1, UnsupportedVersionErr
 	}
 
-	count, err := pr.readInt32()
+	count, err := r.readInt32()
 	if err != nil {
-		return nil, err
+		return -1, err
 	}
 
-	if count > pr.MaxObjectsLimit {
-		return nil, NewError("too many objects %d, limit is %d", count, pr.MaxObjectsLimit)
+	if count > r.MaxObjectsLimit {
+		return -1, MaxObjectsLimitReachedErr
 	}
 
-	ch := make(chan *RAWObject, 1)
-	go pr.readObjects(ch, count)
-
-	// packfile.Size = int64(pr.r.Pos())
-
-	return ch, nil
+	return r.r.position, r.readObjects(count)
 }
 
-func (pr *PackfileReader) validateHeader() error {
+func (r *Reader) validateHeader() error {
 	var header = make([]byte, 4)
-	if _, err := pr.r.Read(header); err != nil {
+	if _, err := r.r.Read(header); err != nil {
 		return err
 	}
 
 	if !bytes.Equal(header, []byte{'P', 'A', 'C', 'K'}) {
-		return NewError("Pack file does not start with 'PACK'")
+		return MalformedPackfileErr
 	}
 
 	return nil
 }
 
-func (pr *PackfileReader) readInt32() (uint32, error) {
+func (r *Reader) readInt32() (uint32, error) {
 	var value uint32
-	if err := binary.Read(pr.r, binary.BigEndian, &value); err != nil {
+	if err := binary.Read(r.r, binary.BigEndian, &value); err != nil {
 		return 0, err
 	}
 
 	return value, nil
 }
 
-func (pr *PackfileReader) readObjects(ch chan *RAWObject, count uint32) error {
+func (r *Reader) readObjects(count uint32) error {
 	// This code has 50-80 µs of overhead per object not counting zlib inflation.
 	// Together with zlib inflation, it's 400-410 µs for small objects.
 	// That's 1 sec for ~2450 objects, ~4.20 MB, or ~250 ms per MB,
 	// of which 12-20 % is _not_ zlib inflation (ie. is our code).
-	defer func() {
-		close(ch)
-	}()
-
 	for i := 0; i < int(count); i++ {
-		var pos = pr.Pos()
-		obj, err := pr.newRAWObject()
+		start := r.r.position
+		obj, err := r.newRAWObject()
 		if err != nil && err != io.EOF {
 			fmt.Println(err)
 			return err
 		}
 
-		if pr.Format == UnknownFormat || pr.Format == OFSDeltaFormat {
-			pr.offsets[pos] = obj
-		}
-
-		if pr.Format == UnknownFormat || pr.Format == REFDeltaFormat {
-			pr.objects[obj.Hash] = obj
+		if r.Format == UnknownFormat || r.Format == OFSDeltaFormat {
+			r.offsets[start] = obj.Hash()
 		}
 
-		ch <- obj
-
+		r.s.Set(obj)
 		if err == io.EOF {
 			break
 		}
@@ -147,154 +144,147 @@ func (pr *PackfileReader) readObjects(ch chan *RAWObject, count uint32) error {
 	return nil
 }
 
-func (pr *PackfileReader) Pos() int { return pr.r.Pos() }
-
-func (pr *PackfileReader) newRAWObject() (*RAWObject, error) {
-	raw := &RAWObject{}
-	steps := 0
+func (r *Reader) newRAWObject() (common.Object, error) {
+	raw := r.s.New()
+	var steps int64
 
 	var buf [1]byte
-	if _, err := pr.r.Read(buf[:]); err != nil {
+	if _, err := r.r.Read(buf[:]); err != nil {
 		return nil, err
 	}
 
-	raw.Type = ObjectType((buf[0] >> 4) & 7)
-	raw.Size = uint64(buf[0] & 15)
+	typ := common.ObjectType((buf[0] >> 4) & 7)
+	size := int64(buf[0] & 15)
 	steps++ // byte we just read to get `o.typ` and `o.size`
 
 	var shift uint = 4
 	for buf[0]&0x80 == 0x80 {
-		if _, err := pr.r.Read(buf[:]); err != nil {
+		if _, err := r.r.Read(buf[:]); err != nil {
 			return nil, err
 		}
 
-		raw.Size += uint64(buf[0]&0x7f) << shift
+		size += int64(buf[0]&0x7f) << shift
 		steps++ // byte we just read to update `o.size`
 		shift += 7
 	}
 
+	raw.SetType(typ)
+	raw.SetSize(size)
+
 	var err error
-	switch raw.Type {
-	case REFDeltaObject:
-		err = pr.readREFDelta(raw)
-	case OFSDeltaObject:
-		err = pr.readOFSDelta(raw, steps)
-	case CommitObject, TreeObject, BlobObject, TagObject:
-		err = pr.readObject(raw)
+	switch raw.Type() {
+	case common.REFDeltaObject:
+		err = r.readREFDelta(raw)
+	case common.OFSDeltaObject:
+		err = r.readOFSDelta(raw, steps)
+	case common.CommitObject, common.TreeObject, common.BlobObject, common.TagObject:
+		err = r.readObject(raw)
 	default:
-		err = NewError("Invalid git object tag %q", raw.Type)
+		err = InvalidObjectErr.n("tag %q", raw.Type)
 	}
 
 	return raw, err
 }
 
-func (pr *PackfileReader) readREFDelta(raw *RAWObject) error {
-	var ref Hash
-	if _, err := pr.r.Read(ref[:]); err != nil {
+func (r *Reader) readREFDelta(raw common.Object) error {
+	var ref common.Hash
+	if _, err := r.r.Read(ref[:]); err != nil {
 		return err
 	}
 
-	buf, err := pr.inflate(raw.Size)
-	if err != nil {
+	buf := bytes.NewBuffer(nil)
+	if err := r.inflate(buf); err != nil {
 		return err
 	}
 
-	referenced, ok := pr.objects[ref]
+	referenced, ok := r.s.Get(ref)
 	if !ok {
-		fmt.Println("not found", ref)
-	} else {
-		patched := PatchDelta(referenced.Bytes, buf[:])
-		if patched == nil {
-			return NewError("error while patching %x", ref)
-		}
+		return ObjectNotFoundErr.n("%s", ref)
+	}
 
-		raw.Type = referenced.Type
-		raw.Bytes = patched
-		raw.Size = uint64(len(patched))
-		raw.Hash = ComputeHash(raw.Type, raw.Bytes)
+	d, _ := ioutil.ReadAll(referenced.Reader())
+	patched := patchDelta(d, buf.Bytes())
+	if patched == nil {
+		return PatchingErr.n("hash %q", ref)
 	}
 
+	raw.SetType(referenced.Type())
+	raw.SetSize(int64(len(patched)))
+	raw.Writer().Write(patched)
+
 	return nil
 }
 
-func (pr *PackfileReader) readOFSDelta(raw *RAWObject, steps int) error {
-	var pos = pr.Pos()
-
-	// read negative offset
-	offset, err := decodeOffset(pr.r, steps)
+func (r *Reader) readOFSDelta(raw common.Object, steps int64) error {
+	start := r.r.position
+	offset, err := decodeOffset(r.r, steps)
 	if err != nil {
 		return err
 	}
 
-	buf, err := pr.inflate(raw.Size)
-	if err != nil {
+	buf := bytes.NewBuffer(nil)
+	if err := r.inflate(buf); err != nil {
 		return err
 	}
 
-	ref, ok := pr.offsets[pos+offset]
+	ref, ok := r.offsets[start+offset]
 	if !ok {
-		return NewError("can't find a pack entry at %d", pos+offset)
+		return PackEntryNotFoundErr.n("offset %d", start+offset)
 	}
 
-	patched := PatchDelta(ref.Bytes, buf)
+	referenced, _ := r.s.Get(ref)
+	d, _ := ioutil.ReadAll(referenced.Reader())
+	patched := patchDelta(d, buf.Bytes())
 	if patched == nil {
-		return NewError("error while patching %q", ref)
+		return PatchingErr.n("hash %q", ref)
 	}
 
-	raw.Type = ref.Type
-	raw.Bytes = patched
-	raw.Size = uint64(len(patched))
-	raw.Hash = ComputeHash(raw.Type, raw.Bytes)
+	raw.SetType(referenced.Type())
+	raw.SetSize(int64(len(patched)))
+	raw.Writer().Write(patched)
 
 	return nil
 }
 
-func (pr *PackfileReader) readObject(raw *RAWObject) error {
-	buf, err := pr.inflate(raw.Size)
-	if err != nil {
-		return err
-	}
-
-	raw.Bytes = buf
-	raw.Hash = ComputeHash(raw.Type, raw.Bytes)
-
-	return nil
+func (r *Reader) readObject(raw common.Object) error {
+	return r.inflate(raw.Writer())
 }
 
-func (pr *PackfileReader) inflate(size uint64) ([]byte, error) {
-	zr, err := zlib.NewReader(pr.r)
+func (r *Reader) inflate(w io.Writer) error {
+	zr, err := zlib.NewReader(r.r)
 	if err != nil {
 		if err == zlib.ErrHeader {
-			return nil, zlib.ErrHeader
+			return zlib.ErrHeader
 		}
 
-		return nil, NewError("error opening packfile's object zlib: %v", err)
+		return ZLibErr.n("%s", err)
 	}
 
 	defer zr.Close()
 
-	if size > pr.MaxObjectSize {
-		return nil, NewError("the object size %q exceeed the allowed limit: %q",
-			size, pr.MaxObjectSize)
-	}
-
-	var buf bytes.Buffer
-	io.Copy(&buf, zr) // also: io.CopyN(&buf, zr, int64(o.size))
-
-	if buf.Len() != int(size) {
-		return nil, NewError(
-			"inflated size mismatch, expected %d, got %d", size, buf.Len())
-	}
-
-	return buf.Bytes(), nil
+	_, err = io.Copy(w, zr)
+	return err
 }
 
 type ReaderError struct {
-	Msg string // description of error
+	reason, additional string
+}
+
+func newError(reason string) *ReaderError {
+	return &ReaderError{reason: reason}
 }
 
-func NewError(format string, args ...interface{}) error {
-	return &ReaderError{Msg: fmt.Sprintf(format, args...)}
+func (e *ReaderError) Error() string {
+	if e.additional == "" {
+		return e.reason
+	}
+
+	return fmt.Sprintf("%s: %s", e.reason, e.additional)
 }
 
-func (e *ReaderError) Error() string { return e.Msg }
+func (e *ReaderError) n(format string, args ...interface{}) *ReaderError {
+	return &ReaderError{
+		reason:     e.reason,
+		additional: fmt.Sprintf(format, args...),
+	}
+}
diff --git a/formats/packfile/reader_test.go b/formats/packfile/reader_test.go
index 917eee1..14c092e 100644
--- a/formats/packfile/reader_test.go
+++ b/formats/packfile/reader_test.go
@@ -6,13 +6,17 @@ import (
 	"fmt"
 	"os"
 	"runtime"
+	"testing"
 	"time"
 
-	"github.com/dustin/go-humanize"
+	"gopkg.in/src-d/go-git.v2/common"
 
+	"github.com/dustin/go-humanize"
 	. "gopkg.in/check.v1"
 )
 
+func Test(t *testing.T) { TestingT(t) }
+
 type ReaderSuite struct{}
 
 var _ = Suite(&ReaderSuite{})
@@ -23,13 +27,13 @@ func (s *ReaderSuite) TestReadPackfile(c *C) {
 	data, _ := base64.StdEncoding.DecodeString(packFileWithEmptyObjects)
 	d := bytes.NewReader(data)
 
-	r, err := NewPackfileReader(d, nil)
-	c.Assert(err, IsNil)
+	r := NewReader(d)
 
-	ch, err := r.Read()
+	storage := common.NewRAWObjectStorage()
+	_, err := r.Read(storage)
 	c.Assert(err, IsNil)
 
-	AssertObjects(c, ch, []string{
+	AssertObjects(c, storage, []string{
 		"778c85ff95b5514fea0ba4c7b6a029d32e2c3b96",
 		"db4002e880a08bf6cc7217512ad937f1ac8824a2",
 		"551fe11a9ef992763b7e0be4500cf7169f2f8575",
@@ -56,14 +60,14 @@ func (s *ReaderSuite) testReadPackfileGitFixture(c *C, file string, f Format) {
 	d, err := os.Open(file)
 	c.Assert(err, IsNil)
 
-	r, err := NewPackfileReader(d, nil)
-	c.Assert(err, IsNil)
-
+	r := NewReader(d)
 	r.Format = f
-	ch, err := r.Read()
+
+	storage := common.NewRAWObjectStorage()
+	_, err = r.Read(storage)
 	c.Assert(err, IsNil)
 
-	AssertObjects(c, ch, []string{
+	AssertObjects(c, storage, []string{
 		"918c48b83bd081e863dbe1b80f8998f058cd8294",
 		"af2d6a6954d532f8ffb47615169c8fdf9d383a1a",
 		"1669dce138d9b841a518c64b10914d88f5e488ea",
@@ -95,14 +99,12 @@ func (s *ReaderSuite) testReadPackfileGitFixture(c *C, file string, f Format) {
 	})
 }
 
-func AssertObjects(c *C, ch chan *RAWObject, expects []string) {
+func AssertObjects(c *C, s *common.RAWObjectStorage, expects []string) {
+	c.Assert(len(expects), Equals, len(s.Objects))
 	for _, expected := range expects {
-		obtained := <-ch
-		c.Assert(obtained.Hash.String(), Equals, expected)
-
-		computed := ComputeHash(obtained.Type, obtained.Bytes)
-		c.Assert(computed.String(), Equals, expected)
-		c.Assert(obtained.Bytes, HasLen, int(obtained.Size))
+		obtained, ok := s.Get(common.NewHash(expected))
+		c.Assert(ok, Equals, true)
+		c.Assert(obtained.Hash().String(), Equals, expected)
 	}
 }
 
@@ -150,7 +152,7 @@ func (s *ReaderSuite) _TestMemoryOFS(c *C) {
 	fmt.Println("HeapAlloc", a.HeapAlloc-b.HeapAlloc, humanize.Bytes(a.HeapAlloc-b.HeapAlloc))
 	fmt.Println("HeapSys", a.HeapSys, humanize.Bytes(a.HeapSys-b.HeapSys))
 
-	fmt.Println("objects", len(p))
+	fmt.Println("objects", len(p.Objects))
 	fmt.Println("time", time.Since(start))
 }
 
@@ -168,26 +170,20 @@ func (s *ReaderSuite) _TestMemoryREF(c *C) {
 	fmt.Println("HeapAlloc", a.HeapAlloc-b.HeapAlloc, humanize.Bytes(a.HeapAlloc-b.HeapAlloc))
 	fmt.Println("HeapSys", a.HeapSys, humanize.Bytes(a.HeapSys-b.HeapSys))
 
-	fmt.Println("objects", len(p))
+	fmt.Println("objects", len(p.Objects))
 	fmt.Println("time", time.Since(start))
 }
 
-func readFromFile(c *C, file string, f Format) []*RAWObject {
+func readFromFile(c *C, file string, f Format) *common.RAWObjectStorage {
 	d, err := os.Open(file)
 	c.Assert(err, IsNil)
 
-	r, err := NewPackfileReader(d, nil)
-	c.Assert(err, IsNil)
-
+	r := NewReader(d)
 	r.Format = f
-	ch, err := r.Read()
-	c.Assert(err, IsNil)
-	c.Assert(ch, NotNil)
 
-	var objs []*RAWObject
-	for o := range ch {
-		objs = append(objs, o)
-	}
+	storage := common.NewRAWObjectStorage()
+	_, err = r.Read(storage)
+	c.Assert(err, IsNil)
 
-	return objs
+	return storage
 }
diff --git a/formats/packfile/objects.go b/objects.go
index 9286090..77bdc2a 100644
--- a/formats/packfile/objects.go
+++ b/objects.go
@@ -1,76 +1,19 @@
-package packfile
+package git
 
 import (
 	"bytes"
-	"crypto/sha1"
 	"encoding/hex"
 	"fmt"
 	"strconv"
 	"time"
-)
-
-type ObjectType int8
 
-const (
-	CommitObject   ObjectType = 1
-	TreeObject     ObjectType = 2
-	BlobObject     ObjectType = 3
-	TagObject      ObjectType = 4
-	OFSDeltaObject ObjectType = 6
-	REFDeltaObject ObjectType = 7
+	"gopkg.in/src-d/go-git.v2/common"
 )
 
-func (t ObjectType) String() string {
-	switch t {
-	case CommitObject:
-		return "commit"
-	case TreeObject:
-		return "tree"
-	case BlobObject:
-		return "blob"
-	default:
-		return "-"
-	}
-}
-
-type RAWObject struct {
-	Hash  Hash
-	Type  ObjectType
-	Size  uint64
-	Bytes []byte
-}
-
 // Object generic object interface
 type Object interface {
-	Type() ObjectType
-	Hash() Hash
-}
-
-// Hash SHA1 hased content
-type Hash [20]byte
-
-// ComputeHash compute the hash for a given objType and content
-func ComputeHash(t ObjectType, content []byte) Hash {
-	h := []byte(t.String())
-	h = append(h, ' ')
-	h = strconv.AppendInt(h, int64(len(content)), 10)
-	h = append(h, 0)
-	h = append(h, content...)
-
-	return Hash(sha1.Sum(h))
-}
-
-func NewHash(s string) Hash {
-	b, _ := hex.DecodeString(s)
-
-	var h Hash
-	copy(h[:], b)
-
-	return h
-}
-
-func (h Hash) String() string {
-	return hex.EncodeToString(h[:])
+	Type() common.ObjectType
+	Hash() common.Hash
 }
 
 // Commit points to a single tree, marking it as what the project looked like
@@ -79,17 +22,17 @@ func (h Hash) String() string {
 // commit, a pointer to the previous commit(s), etc.
 // http://schacon.github.io/gitbook/1_the_git_object_model.html
 type Commit struct {
-	Tree      Hash
-	Parents   []Hash
+	Tree      common.Hash
+	Parents   []common.Hash
 	Author    Signature
 	Committer Signature
 	Message   string
-	hash      Hash
+	hash      common.Hash
 }
 
 // ParseCommit transform a byte slice into a Commit struct
 func ParseCommit(b []byte) (*Commit, error) {
-	o := &Commit{hash: ComputeHash(CommitObject, b)}
+	o := &Commit{hash: common.ComputeHash(common.CommitObject, b)}
 
 	lines := bytes.Split(b, []byte{'\n'})
 	for i := range lines {
@@ -101,7 +44,7 @@ func ParseCommit(b []byte) (*Commit, error) {
 			case "tree":
 				_, err = hex.Decode(o.Tree[:], split[1])
 			case "parent":
-				var h Hash
+				var h common.Hash
 				_, err = hex.Decode(h[:], split[1])
 				if err == nil {
 					o.Parents = append(o.Parents, h)
@@ -126,12 +69,12 @@ func ParseCommit(b []byte) (*Commit, error) {
 }
 
 // Type returns the object type
-func (o *Commit) Type() ObjectType {
-	return CommitObject
+func (o *Commit) Type() common.ObjectType {
+	return common.CommitObject
 }
 
 // Hash returns the computed hash of the commit
-func (o *Commit) Hash() Hash {
+func (o *Commit) Hash() common.Hash {
 	return o.hash
 }
 
@@ -139,18 +82,18 @@ func (o *Commit) Hash() Hash {
 // and/or blobs (i.e. files and sub-directories)
 type Tree struct {
 	Entries []TreeEntry
-	hash    Hash
+	hash    common.Hash
 }
 
 // TreeEntry represents a file
 type TreeEntry struct {
 	Name string
-	Hash Hash
+	Hash common.Hash
 }
 
 // ParseTree transform a byte slice into a Tree struct
 func ParseTree(b []byte) (*Tree, error) {
-	o := &Tree{hash: ComputeHash(TreeObject, b)}
+	o := &Tree{hash: common.ComputeHash(common.TreeObject, b)}
 
 	if len(b) == 0 {
 		return o, nil
@@ -176,41 +119,39 @@ func ParseTree(b []byte) (*Tree, error) {
 }
 
 // Type returns the object type
-func (o *Tree) Type() ObjectType {
-	return TreeObject
+func (o *Tree) Type() common.ObjectType {
+	return common.TreeObject
 }
 
 // Hash returns the computed hash of the tree
-func (o *Tree) Hash() Hash {
+func (o *Tree) Hash() common.Hash {
 	return o.hash
 }
 
 // Blob is used to store file data - it is generally a file.
 type Blob struct {
 	Len  int
-	hash Hash
+	hash common.Hash
 }
 
 // ParseBlob transform a byte slice into a Blob struct
 func ParseBlob(b []byte) (*Blob, error) {
 	return &Blob{
 		Len:  len(b),
-		hash: ComputeHash(BlobObject, b),
+		hash: common.ComputeHash(common.BlobObject, b),
 	}, nil
 }
 
 // Type returns the object type
-func (o *Blob) Type() ObjectType {
-	return BlobObject
+func (o *Blob) Type() common.ObjectType {
+	return common.BlobObject
 }
 
 // Hash returns the computed hash of the blob
-func (o *Blob) Hash() Hash {
+func (o *Blob) Hash() common.Hash {
 	return o.hash
 }
 
-type ContentCallback func(hash Hash, content []byte)
-
 // Signature represents an action signed by a person
 type Signature struct {
 	Name  string
diff --git a/formats/packfile/objects_test.go b/objects_test.go
index 0760653..53da5f0 100644
--- a/formats/packfile/objects_test.go
+++ b/objects_test.go
@@ -1,4 +1,4 @@
-package packfile
+package git
 
 import (
 	"encoding/base64"
@@ -11,20 +11,6 @@ type ObjectsSuite struct{}
 
 var _ = Suite(&ObjectsSuite{})
 
-func (s *ObjectsSuite) TestComputeHash(c *C) {
-	hash := ComputeHash(BlobObject, []byte(""))
-	c.Assert(hash.String(), Equals, "e69de29bb2d1d6434b8b29ae775ad8c2e48c5391")
-
-	hash = ComputeHash(BlobObject, []byte("Hello, World!\n"))
-	c.Assert(hash.String(), Equals, "8ab686eafeb1f44702738c8b0f24f2567c36da6d")
-}
-
-func (s *ObjectsSuite) TestNewHash(c *C) {
-	hash := ComputeHash(BlobObject, []byte("Hello, World!\n"))
-
-	c.Assert(hash, Equals, NewHash(hash.String()))
-}
-
 var CommitFixture = "dHJlZSBjMmQzMGZhOGVmMjg4NjE4ZjY1ZjZlZWQ2ZTE2OGUwZDUxNDg4NmY0CnBhcmVudCBiMDI5NTE3ZjYzMDBjMmRhMGY0YjY1MWI4NjQyNTA2Y2Q2YWFmNDVkCnBhcmVudCBiOGU0NzFmNThiY2JjYTYzYjA3YmRhMjBlNDI4MTkwNDA5YzJkYjQ3CmF1dGhvciBNw6F4aW1vIEN1YWRyb3MgPG1jdWFkcm9zQGdtYWlsLmNvbT4gMTQyNzgwMjQzNCArMDIwMApjb21taXR0ZXIgTcOheGltbyBDdWFkcm9zIDxtY3VhZHJvc0BnbWFpbC5jb20+IDE0Mjc4MDI0MzQgKzAyMDAKCk1lcmdlIHB1bGwgcmVxdWVzdCAjMSBmcm9tIGRyaXBvbGxlcy9mZWF0dXJlCgpDcmVhdGluZyBjaGFuZ2Vsb2c="
 
 func (s *ObjectsSuite) TestParseCommit(c *C) {
diff --git a/remote_test.go b/remote_test.go
index 6cdbce0..5674407 100644
--- a/remote_test.go
+++ b/remote_test.go
@@ -1,6 +1,7 @@
 package git
 
 import (
+	"gopkg.in/src-d/go-git.v2/common"
 	"gopkg.in/src-d/go-git.v2/formats/packfile"
 
 	. "gopkg.in/check.v1"
@@ -40,10 +41,10 @@ func (s *SuiteRemote) TestFetchDefaultBranch(c *C) {
 	reader, err := r.FetchDefaultBranch()
 	c.Assert(err, IsNil)
 
-	pr, err := packfile.NewPackfileReader(reader, nil)
-	c.Assert(err, IsNil)
+	pr := packfile.NewReader(reader)
 
-	ch, err := pr.Read()
+	storage := common.NewRAWObjectStorage()
+	_, err = pr.Read(storage)
 	c.Assert(err, IsNil)
-	c.Assert(ch, NotNil)
+	c.Assert(storage.Objects, HasLen, 28)
 }