object: make Decode take a hash.Hash as parameter

In most cases, we can determine the object ID algorithm in use from its
length without much difficulty.  However, for trees, the data is stored
in a binary format, and consequently it's not possible to correctly
parse a tree without knowing the hash algorithm in use.  Let's pass the
hash algorithm down through the Decode function so we can know the
correct length to parse.

While we're at it, increase the buffer length for the tree parsing code
and make it work for both SHA-1 and SHA-256.

Author: bk2204

blob.go

@@ -3,6 +3,7 @@ package gitobj
 import (
 	"bytes"
 	"fmt"
+	"hash"
 	"io"
 	"os"
 )
@@ -75,7 +76,7 @@ func (b *Blob) Type() ObjectType { return BlobObjectType }
 // stream, which is always zero.
 //
 // If any errors are encountered while reading the blob, they will be returned.
-func (b *Blob) Decode(r io.Reader, size int64) (n int, err error) {
+func (b *Blob) Decode(hash hash.Hash, r io.Reader, size int64) (n int, err error) {
 	b.Size = size
 	b.Contents = io.LimitReader(r, size)
 

blob_test.go

@@ -2,6 +2,7 @@ package gitobj
 
 import (
 	"bytes"
+	"crypto/sha1"
 	"errors"
 	"io/ioutil"
 	"strings"
@@ -48,7 +49,7 @@ func TestBlobDecoding(t *testing.T) {
 	from := strings.NewReader(contents)
 
 	b := new(Blob)
-	n, err := b.Decode(from, int64(len(contents)))
+	n, err := b.Decode(sha1.New(), from, int64(len(contents)))
 
 	assert.Equal(t, 0, n)
 	assert.Nil(t, err)

commit.go

@@ -5,6 +5,7 @@ import (
 	"bytes"
 	"encoding/hex"
 	"fmt"
+	"hash"
 	"io"
 	"strings"
 	"time"
@@ -91,12 +92,12 @@ func (c *Commit) Type() ObjectType { return CommitObjectType }
 //
 // If any error was encountered along the way, that will be returned, along with
 // the number of bytes read up to that point.
-func (c *Commit) Decode(from io.Reader, size int64) (n int, err error) {
+func (c *Commit) Decode(hash hash.Hash, from io.Reader, size int64) (n int, err error) {
 	var finishedHeaders bool
 	var messageParts []string
 
 	s := bufio.NewScanner(from)
-	s.Buffer(nil, 1024 * 1024)
+	s.Buffer(nil, 1024*1024)
 	for s.Scan() {
 		text := s.Text()
 		n = n + len(text+"\n")

commit_test.go

@@ -2,6 +2,7 @@ package gitobj
 
 import (
 	"bytes"
+	"crypto/sha1"
 	"encoding/hex"
 	"fmt"
 	"io"
@@ -77,7 +78,7 @@ func TestCommitDecoding(t *testing.T) {
 	flen := from.Len()
 
 	commit := new(Commit)
-	n, err := commit.Decode(from, int64(flen))
+	n, err := commit.Decode(sha1.New(), from, int64(flen))
 
 	assert.Nil(t, err)
 	assert.Equal(t, flen, n)
@@ -107,7 +108,7 @@ func TestCommitDecodingWithEmptyName(t *testing.T) {
 	flen := from.Len()
 
 	commit := new(Commit)
-	n, err := commit.Decode(from, int64(flen))
+	n, err := commit.Decode(sha1.New(), from, int64(flen))
 
 	assert.Nil(t, err)
 	assert.Equal(t, flen, n)
@@ -138,7 +139,7 @@ func TestCommitDecodingWithLargeCommitMessage(t *testing.T) {
 	flen := from.Len()
 
 	commit := new(Commit)
-	n, err := commit.Decode(from, int64(flen))
+	n, err := commit.Decode(sha1.New(), from, int64(flen))
 
 	assert.Nil(t, err)
 	assert.Equal(t, flen, n)
@@ -164,7 +165,7 @@ func TestCommitDecodingWithMessageKeywordPrefix(t *testing.T) {
 	flen := from.Len()
 
 	commit := new(Commit)
-	n, err := commit.Decode(from, int64(flen))
+	n, err := commit.Decode(sha1.New(), from, int64(flen))
 
 	assert.NoError(t, err)
 	assert.Equal(t, flen, n)
@@ -191,7 +192,7 @@ func TestCommitDecodingWithWhitespace(t *testing.T) {
 	flen := from.Len()
 
 	commit := new(Commit)
-	n, err := commit.Decode(from, int64(flen))
+	n, err := commit.Decode(sha1.New(), from, int64(flen))
 
 	assert.NoError(t, err)
 	assert.Equal(t, flen, n)
@@ -221,7 +222,7 @@ func TestCommitDecodingMultilineHeader(t *testing.T) {
 	flen := from.Len()
 
 	commit := new(Commit)
-	n, err := commit.Decode(from, int64(flen))
+	n, err := commit.Decode(sha1.New(), from, int64(flen))
 
 	require.Nil(t, err)
 	require.Equal(t, flen, n)

object.go

@@ -1,6 +1,9 @@
 package gitobj
 
-import "io"
+import (
+	"hash"
+	"io"
+)
 
 // Object is an interface satisfied by any concrete type that represents a loose
 // Git object.
@@ -29,7 +32,7 @@ type Object interface {
 	//
 	// If an(y) error was encountered, it should be returned immediately,
 	// along with the number of bytes read up to that point.
-	Decode(from io.Reader, size int64) (n int, err error)
+	Decode(hash hash.Hash, from io.Reader, size int64) (n int, err error)
 
 	// Type returns the ObjectType constant that represents an instance of
 	// the implementing type.

object_db.go

@@ -370,7 +370,7 @@ func (o *ObjectDatabase) decode(r *ObjectReader, into Object) error {
 		return &UnexpectedObjectType{Got: typ, Wanted: into.Type()}
 	}
 
-	if _, err = into.Decode(r, size); err != nil {
+	if _, err = into.Decode(o.Hasher(), r, size); err != nil {
 		return err
 	}
 

object_db_test.go

@@ -3,6 +3,7 @@ package gitobj
 import (
 	"bytes"
 	"compress/zlib"
+	"crypto/sha1"
 	"encoding/hex"
 	"fmt"
 	"io"
@@ -270,6 +271,7 @@ func TestWriteCommitWithGPGSignature(t *testing.T) {
 
 	commit := new(Commit)
 	_, err = commit.Decode(
+		sha1.New(),
 		strings.NewReader(roundTripCommit), int64(len(roundTripCommit)))
 	require.NoError(t, err)
 

tag.go

@@ -5,6 +5,7 @@ import (
 	"bytes"
 	"encoding/hex"
 	"fmt"
+	"hash"
 	"io"
 	"strings"
 )
@@ -24,7 +25,7 @@ type Tag struct {
 //
 // If any error was encountered along the way it will be returned, and the
 // receiving *Tag is considered invalid.
-func (t *Tag) Decode(r io.Reader, size int64) (int, error) {
+func (t *Tag) Decode(hash hash.Hash, r io.Reader, size int64) (int, error) {
 	scanner := bufio.NewScanner(io.LimitReader(r, size))
 
 	var (

tag_test.go

@@ -2,6 +2,7 @@ package gitobj
 
 import (
 	"bytes"
+	"crypto/sha1"
 	"fmt"
 	"testing"
 
@@ -52,7 +53,7 @@ func TestTagDecode(t *testing.T) {
 	flen := from.Len()
 
 	tag := new(Tag)
-	n, err := tag.Decode(from, int64(flen))
+	n, err := tag.Decode(sha1.New(), from, int64(flen))
 
 	assert.Nil(t, err)
 	assert.Equal(t, n, flen)

tree.go

@@ -4,6 +4,7 @@ import (
 	"bufio"
 	"bytes"
 	"fmt"
+	"hash"
 	"io"
 	"sort"
 	"strconv"
@@ -27,7 +28,8 @@ func (t *Tree) Type() ObjectType { return TreeObjectType }
 //
 // If any error was encountered along the way, that will be returned, along with
 // the number of bytes read up to that point.
-func (t *Tree) Decode(from io.Reader, size int64) (n int, err error) {
+func (t *Tree) Decode(hash hash.Hash, from io.Reader, size int64) (n int, err error) {
+	hashlen := hash.Size()
 	buf := bufio.NewReader(from)
 
 	var entries []*TreeEntry
@@ -51,15 +53,15 @@ func (t *Tree) Decode(from io.Reader, size int64) (n int, err error) {
 		n += len(fname)
 		fname = strings.TrimSuffix(fname, "\x00")
 
-		var sha [20]byte
-		if _, err = io.ReadFull(buf, sha[:]); err != nil {
+		var sha [32]byte
+		if _, err = io.ReadFull(buf, sha[:hashlen]); err != nil {
 			return n, err
 		}
-		n += 20
+		n += hashlen
 
 		entries = append(entries, &TreeEntry{
 			Name:     fname,
-			Oid:      sha[:],
+			Oid:      sha[:hashlen],
 			Filemode: int32(mode),
 		})
 	}

tree_test.go

@@ -3,6 +3,7 @@ package gitobj
 import (
 	"bufio"
 	"bytes"
+	"crypto/sha1"
 	"fmt"
 	"sort"
 	"strconv"
@@ -68,7 +69,7 @@ func TestTreeDecoding(t *testing.T) {
 	flen := from.Len()
 
 	tree := new(Tree)
-	n, err := tree.Decode(from, int64(flen))
+	n, err := tree.Decode(sha1.New(), from, int64(flen))
 
 	assert.Nil(t, err)
 	assert.Equal(t, flen, n)
@@ -106,7 +107,7 @@ func TestTreeDecodingShaBoundary(t *testing.T) {
 	flen := from.Len()
 
 	tree := new(Tree)
-	n, err := tree.Decode(bufio.NewReaderSize(&from, flen-2), int64(flen))
+	n, err := tree.Decode(sha1.New(), bufio.NewReaderSize(&from, flen-2), int64(flen))
 
 	assert.Nil(t, err)
 	assert.Equal(t, flen, n)