git: add support for cloning empty repos

Signed-off-by: Sanskar Jaiswal <[email protected]>

git/client.go

@@ -24,6 +24,8 @@ import (
 // on a Git repository.
 type RepositoryReader interface {
 	// Clone clones a repository from the provided url using the options provided.
+	// It returns a Commit object describing the Git commit that the repository
+	// HEAD points to. If the repository is empty, it returns a nil Commit.
 	Clone(ctx context.Context, url string, cloneOpts CloneOptions) (*Commit, error)
 	// IsClean returns whether the working tree is clean.
 	IsClean() (bool, error)

git/gogit/clone.go

@@ -20,6 +20,7 @@ import (
 	"context"
 	"fmt"
 	"io"
+	"os"
 	"sort"
 	"strings"
 	"time"
@@ -93,13 +94,48 @@ func (g *Client) cloneBranch(ctx context.Context, url, branch string, opts git.C
 
 	repo, err := extgogit.CloneContext(ctx, g.storer, g.worktreeFS, cloneOpts)
 	if err != nil {
-		if err == transport.ErrEmptyRemoteRepository || err == transport.ErrRepositoryNotFound || isRemoteBranchNotFoundErr(err, ref.String()) {
+		if err == transport.ErrRepositoryNotFound || isRemoteBranchNotFoundErr(err, ref.String()) {
 			return nil, git.ErrRepositoryNotFound{
 				Message: fmt.Sprintf("unable to clone: %s", err),
 				URL:     url,
 			}
 		}
-		return nil, fmt.Errorf("unable to clone '%s': %w", url, gitutil.GoGitError(err))
+		// Directly cloning an empty Git repo to a directory fails with this error.
+		// We check for the error, init a new Git repo in that directory and then
+		// do a fresh clone.
+		if err == transport.ErrEmptyRemoteRepository {
+			if err = os.RemoveAll(g.path); err == nil {
+				if err = g.Init(ctx, url, branch); err == nil {
+					commit, err := g.Clone(ctx, url, opts)
+					if err == nil {
+						return commit, nil
+					}
+				}
+			}
+		}
+		if err != nil {
+			return nil, fmt.Errorf("unable to clone '%s': %w", url, gitutil.GoGitError(err))
+		}
+	}
+
+	// Fetch the log and count the entries to detect if the repository we cloned is empty.
+	// Is there a better way?
+	log, err := repo.Log(&extgogit.LogOptions{
+		All: true,
+	})
+	if err != nil {
+		return nil, fmt.Errorf("unable to resole Git repository log: %w", err)
+	}
+
+	var l int
+	log.ForEach(func(c *object.Commit) error {
+		l += 1
+		return nil
+	})
+	// Since there are no logs, the repo is empty, hence we exit early.
+	if l == 0 {
+		g.repository = repo
+		return nil, nil
 	}
 
 	head, err := repo.Head()

git/gogit/clone_test.go

@@ -46,7 +46,7 @@ import (
 const testRepositoryPath = "../testdata/git/repo"
 
 func TestClone_cloneBranch(t *testing.T) {
-	repo, path, err := initRepo(t)
+	repo, repoPath, err := initRepo(t)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -65,6 +65,11 @@ func TestClone_cloneBranch(t *testing.T) {
 		t.Fatal(err)
 	}
 
+	_, emptyRepoPath, err := initRepo(t)
+	if err != nil {
+		t.Fatal(err)
+	}
+
 	tests := []struct {
 		name                   string
 		branch                 string
@@ -73,6 +78,7 @@ func TestClone_cloneBranch(t *testing.T) {
 		expectedCommit         string
 		expectedConcreteCommit bool
 		expectedErr            string
+		expectedEmpty          bool
 	}{
 		{
 			name:                   "Default branch",
@@ -102,6 +108,11 @@ func TestClone_cloneBranch(t *testing.T) {
 			branch:      "invalid",
 			expectedErr: "couldn't find remote ref \"refs/heads/invalid\"",
 		},
+		{
+			name:          "empty repo",
+			branch:        "master",
+			expectedEmpty: true,
+		},
 	}
 
 	for _, tt := range tests {
@@ -112,7 +123,14 @@ func TestClone_cloneBranch(t *testing.T) {
 			ggc, err := NewClient(tmpDir, &git.AuthOptions{Transport: git.HTTP})
 			g.Expect(err).ToNot(HaveOccurred())
 
-			cc, err := ggc.Clone(context.TODO(), path, git.CloneOptions{
+			var upstreamPath string
+			if tt.expectedEmpty {
+				upstreamPath = emptyRepoPath
+			} else {
+				upstreamPath = repoPath
+			}
+
+			cc, err := ggc.Clone(context.TODO(), upstreamPath, git.CloneOptions{
 				CheckoutStrategy: git.CheckoutStrategy{
 					Branch: tt.branch,
 				},
@@ -125,6 +143,14 @@ func TestClone_cloneBranch(t *testing.T) {
 				g.Expect(cc).To(BeNil())
 				return
 			}
+
+			if tt.expectedEmpty {
+				g.Expect(cc).To(BeNil())
+				g.Expect(err).ToNot(HaveOccurred())
+				g.Expect(filepath.Join(ggc.path, ".git")).To(BeADirectory())
+				return
+			}
+
 			g.Expect(err).ToNot(HaveOccurred())
 			g.Expect(cc.String()).To(Equal(tt.branch + "/" + tt.expectedCommit))
 			g.Expect(git.IsConcreteCommit(*cc)).To(Equal(tt.expectedConcreteCommit))

git/libgit2/clone.go

@@ -79,6 +79,14 @@ func (l *Client) cloneBranch(ctx context.Context, url, branch string, opts git.C
 		return nil, fmt.Errorf("unable to fetch remote '%s': %w", url, gitutil.LibGit2Error(err))
 	}
 
+	isEmpty, err := l.repository.IsEmpty()
+	if err != nil {
+		return nil, fmt.Errorf("unable to check if cloned repo '%s' is empty: %w", url, err)
+	}
+	if isEmpty {
+		return nil, nil
+	}
+
 	branchRef, err := l.repository.References.Lookup(fmt.Sprintf("refs/remotes/origin/%s", branch))
 	if err != nil {
 		return nil, fmt.Errorf("unable to lookup branch '%s' for '%s': %w", branch, url, gitutil.LibGit2Error(err))