Introduce our own types into the git package

git/repo.go → git/go-git.go

@@ -14,7 +14,7 @@ var (
 	endError = fmt.Errorf("Finished") // not really an error
 )
-type Repo struct {
+type goGitRepository struct {
 	Repo *git.Repository
 	FromHash plumbing.Hash
@@ -24,8 +24,8 @@ type Repo struct {
 	ToCommit   *object.Commit
 }
-func NewRepo(projectPath string, fromSHA string, toSHA string) (*Repo, error) {
-	out := &Repo{}
+func NewGoGitRepository(projectPath string, fromSHA string, toSHA string) (*goGitRepository, error) {
+	out := &goGitRepository{}
 	repo, err := git.PlainOpen(projectPath)
 	if err != nil {
@@ -67,7 +67,7 @@ func NewRepo(projectPath string, fromSHA string, toSHA string) (*Repo, error) {
 	return out, nil
 }
-func (r *Repo) Diff() (object.Changes, error) {
+func (r *goGitRepository) diff() (object.Changes, error) {
 	var fromTree, toTree *object.Tree
 	if r.FromCommit != nil {
@@ -87,14 +87,32 @@ func (r *Repo) Diff() (object.Changes, error) {
 	return object.DiffTree(fromTree, toTree)
 }
-type FileFunc func(file *object.File, fromCommit, toCommit *object.Commit) error
-func (r *Repo) EachFileChange(ins, mod, del FileFunc) error {
-	changes, err := r.Diff()
+func goGitBuildSignature(sig object.Signature) Signature {
+	return Signature{
+		Name:  sig.Name,
+		Email: sig.Email,
+		When:  sig.When,
+	}
+}
+
+func goGitBuildFile(change object.ChangeEntry, file *object.File) *File {
+	return &File{
+		Path: change.Name,
+		Oid:  file.ID().String(),
+		Blob: file.Blob.Reader,
+		Size: file.Size,
+	}
+}
+func (r *goGitRepository) EachFileChange(ins, mod, del FileFunc) error {
+	changes, err := r.diff()
 	if err != nil {
 		return err
 	}
+	fromCommitStr := r.FromHash.String()
+	toCommitStr := r.ToHash.String()
+
 	for _, change := range changes {
 		// FIXME(nick): submodules may need better support
 		// https://github.com/src-d/go-git/issues/317
@@ -114,14 +132,11 @@ func (r *Repo) EachFileChange(ins, mod, del FileFunc) error {
 		switch action {
 		case merkletrie.Insert:
-			toF.Name = change.To.Name
-			err = ins(toF, r.FromCommit, r.ToCommit)
+			err = ins(goGitBuildFile(change.To, toF), fromCommitStr, toCommitStr)
 		case merkletrie.Modify:
-			toF.Name = change.To.Name
-			err = mod(toF, r.FromCommit, r.ToCommit)
+			err = mod(goGitBuildFile(change.To, toF), fromCommitStr, toCommitStr)
 		case merkletrie.Delete:
-			fromF.Name = change.From.Name
-			err = del(fromF, r.FromCommit, r.ToCommit)
+			err = del(goGitBuildFile(change.From, fromF), fromCommitStr, toCommitStr)
 		default:
 			err = fmt.Errorf("Unrecognised change calculating diff: %+v", change)
 		}
@@ -136,9 +151,16 @@ func (r *Repo) EachFileChange(ins, mod, del FileFunc) error {
 // EachCommit runs `f` for each commit within `fromSHA`...`toSHA` (i.e., the
 // inclusive set).
-func (r *Repo) EachCommit(f func(*object.Commit) error) error {
+func (r *goGitRepository) EachCommit(f CommitFunc) error {
 	err := object.WalkCommitHistory(r.ToCommit, func(c *object.Commit) error {
-		if err := f(c); err != nil {
+
+		commit := &Commit{
+			Message:   c.Message,
+			Hash:      c.Hash.String(),
+			Author:    goGitBuildSignature(c.Author),
+			Committer: goGitBuildSignature(c.Committer),
+		}
+		if err := f(commit); err != nil {
 			return err
 		}

git/repository.go

+package git
+
+import (
+	"io"
+	"time"
+)
+
+type File struct {
+	Path string
+	Blob func() (io.ReadCloser, error)
+	Oid  string
+	Size int64
+}
+
+type Signature struct {
+	Name  string
+	Email string
+	When  time.Time
+}
+
+type Commit struct {
+	Author    Signature
+	Committer Signature
+	Message   string
+	Hash      string
+}
+
+type Repository interface {
+	EachFileChange(ins, mod, del FileFunc) error
+	EachCommit(f CommitFunc) error
+}
+
+type FileFunc func(file *File, fromCommit, toCommit string) error
+type CommitFunc func(commit *Commit) error

indexer/blob.go

@@ -5,8 +5,7 @@ import (
 	"fmt"
 	"io/ioutil"
-	"gopkg.in/src-d/go-git.v4/plumbing/object"
-
+	"gitlab.com/gitlab-org/es-git-go/git"
 	"gitlab.com/gitlab-org/es-git-go/linguist"
 )
@@ -59,14 +58,14 @@ func GenerateBlobID(parentID, filename string) string {
 	return fmt.Sprintf("%s_%s", parentID, filename)
 }
-func (i *Indexer) BuildBlob(file *object.File, commitSHA string) (*Blob, error) {
+func (i *Indexer) BuildBlob(file *git.File, commitSHA string) (*Blob, error) {
 	parentID := i.Submitter.ParentID()
-	if file.Blob.Size > maxBlobSize {
+	if file.Size > maxBlobSize {
 		return nil, skipTooLargeBlob
 	}
-	reader, err := file.Blob.Reader()
+	reader, err := file.Blob()
 	if err != nil {
 		return nil, err
 	}
@@ -85,12 +84,12 @@ func (i *Indexer) BuildBlob(file *object.File, commitSHA string) (*Blob, error)
 	}
 	content := tryEncodeBytes(b)
-	filename := tryEncodeString(file.Name)
+	filename := tryEncodeString(file.Path)
 	return &Blob{
 		Type:      "blob",
-		ID:        GenerateBlobID(parentID, file.Name),
-		OID:       file.Blob.Hash.String(),
+		ID:        GenerateBlobID(parentID, filename),
+		OID:       file.Oid,
 		RepoID:    parentID,
 		CommitSHA: commitSHA,
 		Content:   content,

indexer/commit.go

@@ -3,7 +3,7 @@ package indexer
 import (
 	"fmt"
-	"gopkg.in/src-d/go-git.v4/plumbing/object"
+	"gitlab.com/gitlab-org/es-git-go/git"
 )
 type Commit struct {
@@ -20,8 +20,8 @@ func GenerateCommitID(parentID, commitSHA string) string {
 	return fmt.Sprintf("%s_%s", parentID, commitSHA)
 }
-func (i *Indexer) BuildCommit(c *object.Commit) *Commit {
-	sha := c.Hash.String()
+func (i *Indexer) BuildCommit(c *git.Commit) *Commit {
+	sha := c.Hash
 	parentID := i.Submitter.ParentID()
 	return &Commit{

indexer/indexer.go

@@ -4,8 +4,6 @@ import (
 	"fmt"
 	"log"
-	"gopkg.in/src-d/go-git.v4/plumbing/object"
-
 	"gitlab.com/gitlab-org/es-git-go/git"
 )
@@ -19,48 +17,46 @@ type Submitter interface {
 }
 type Indexer struct {
-	Repo *git.Repo
+	git.Repository
 	Submitter
 }
-// FIXME: none of the indexers worry about encoding right now
-
-func (i *Indexer) SubmitCommit(c *object.Commit) error {
+func (i *Indexer) SubmitCommit(c *git.Commit) error {
 	commit := i.BuildCommit(c)
 	i.Submitter.Index(commit.ID, map[string]interface{}{"commit": commit})
 	return nil
 }
-func (i *Indexer) SubmitBlob(f *object.File, _, toCommit *object.Commit) error {
+func (i *Indexer) SubmitBlob(f *git.File, _, toCommit string) error {
 	// FIXME(nick): Not sure commitSHA is right, or how it works at all
-	blob, err := i.BuildBlob(f, toCommit.Hash.String())
+	blob, err := i.BuildBlob(f, toCommit)
 	if err != nil {
 		if isSkipBlobErr(err) {
 			return nil
 		}
-		return fmt.Errorf("Blob %s: %s", f.Name, err)
+		return fmt.Errorf("Blob %s: %s", f.Path, err)
 	}
 	i.Submitter.Index(blob.ID, map[string]interface{}{"blob": blob})
 	return nil
 }
-func (i *Indexer) RemoveBlob(file *object.File, _, toCommit *object.Commit) error {
-	blobID := GenerateBlobID(toCommit.Hash.String(), i.Submitter.ParentID())
+func (i *Indexer) RemoveBlob(file *git.File, _, toCommit string) error {
+	blobID := GenerateBlobID(toCommit, i.Submitter.ParentID())
 	i.Submitter.Remove(blobID)
 	return nil
 }
 func (i *Indexer) IndexCommits() error {
-	return i.Repo.EachCommit(i.SubmitCommit)
+	return i.Repository.EachCommit(i.SubmitCommit)
 }
 func (i *Indexer) IndexBlobs() error {
-	return i.Repo.EachFileChange(i.SubmitBlob, i.SubmitBlob, i.RemoveBlob)
+	return i.Repository.EachFileChange(i.SubmitBlob, i.SubmitBlob, i.RemoveBlob)
 }
 func (i *Indexer) Index() error {

indexer/person.go

@@ -3,7 +3,7 @@ package indexer
 import (
 	"time"
-	"gopkg.in/src-d/go-git.v4/plumbing/object"
+	"gitlab.com/gitlab-org/es-git-go/git"
 )
 const (
@@ -20,7 +20,7 @@ func GenerateDate(t time.Time) string {
 	return t.Format(elasticTimeFormat)
 }
-func BuildPerson(p object.Signature) *Person {
+func BuildPerson(p git.Signature) *Person {
 	return &Person{
 		Name:  tryEncodeString(p.Name),
 		Email: tryEncodeString(p.Email),

main.go

@@ -19,7 +19,7 @@ func main() {
 	fromSHA := os.Getenv("FROM_SHA")
 	toSHA := os.Getenv("TO_SHA")
-	repo, err := git.NewRepo(projectPath, fromSHA, toSHA)
+	repo, err := git.NewGoGitRepository(projectPath, fromSHA, toSHA)
 	if err != nil {
 		log.Fatalf("Failed to open %s: %s", projectPath, err)
 	}
@@ -30,8 +30,8 @@ func main() {
 	}
 	idx := &indexer.Indexer{
-		Submitter: esClient,
-		Repo:      repo,
+		Submitter:  esClient,
+		Repository: repo,
 	}
 	log.Printf("Indexing from %s to %s", repo.FromHash, repo.ToHash)
@@ -39,6 +39,5 @@ func main() {
 	if err := idx.Index(); err != nil {
 		log.Fatalln("Indexing error: ", err)
-
 	}
 }