git: Add logic to detect object hash for a repositorypks-git-objectids-sha256

Add a new funciton `DetectObjectHash()` that, given a repository, can
detect the object hash that's used by the repository by inspecting the
`extensions.objectFormat` config entry.

3 files changed, 127 insertions, 0 deletions

diff --git a/internal/git/gittest/repo.go b/internal/git/gittest/repo.go
index a9ff1e75e..1b4b08bb3 100644
--- a/internal/git/gittest/repo.go
+++ b/internal/git/gittest/repo.go
@@ -239,6 +239,8 @@ func RewrittenRepository(ctx context.Context, t testing.TB, cfg config.Cfg, repo
 type InitRepoOpts struct {
 	// WithRelativePath determines the relative path of this repository.
 	WithRelativePath string
+	// ObjectFormat overrides the object format used by the repository.
+	ObjectFormat string
 }
 
 // InitRepo creates a new empty repository in the given storage. You can either pass no or exactly
@@ -260,6 +262,9 @@ func InitRepo(t testing.TB, cfg config.Cfg, storage config.Storage, opts ...Init
 	args := []string{"init", "--bare"}
 	args = append(args, initRepoExtraArgs...)
 	args = append(args, repoPath)
+	if opt.ObjectFormat != "" {
+		args = append(args, "--object-format", opt.ObjectFormat)
+	}
 
 	Exec(t, cfg, args...)
 
diff --git a/internal/git/object_id.go b/internal/git/object_id.go
index fcaf7225c..4d6083a12 100644
--- a/internal/git/object_id.go
+++ b/internal/git/object_id.go
@@ -1,10 +1,15 @@
 package git
 
 import (
+	"bytes"
+	"context"
 	"encoding/hex"
 	"errors"
 	"fmt"
 	"regexp"
+
+	"gitlab.com/gitlab-org/gitaly/v15/internal/command"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/helper/text"
 )
 
 var (
@@ -36,6 +41,32 @@ type ObjectHash struct {
 	ZeroOID ObjectID
 }
 
+// DetectObjectHash detects the object-hash used by the given repository.
+func DetectObjectHash(ctx context.Context, repoExecutor RepositoryExecutor) (ObjectHash, error) {
+	var stdout bytes.Buffer
+
+	if err := repoExecutor.ExecAndWait(ctx, SubCmd{
+		Name: "config",
+		Args: []string{"extensions.objectFormat"},
+	}, WithStdout(&stdout)); err != nil {
+		if status, ok := command.ExitStatus(err); ok && status == 1 {
+			return ObjectHashSHA1, nil
+		}
+
+		return ObjectHash{}, fmt.Errorf("reading object format: %w", err)
+	}
+
+	objectFormat := text.ChompBytes(stdout.Bytes())
+	switch objectFormat {
+	case "sha1":
+		return ObjectHashSHA1, nil
+	case "sha256":
+		return ObjectHashSHA256, nil
+	default:
+		return ObjectHash{}, fmt.Errorf("unknown object format: %q", objectFormat)
+	}
+}
+
 // FromHex constructs a new ObjectID from the given hex representation of the object ID. Returns
 // ErrInvalidObjectID if the given object ID is not valid.
 func (h ObjectHash) FromHex(hex string) (ObjectID, error) {
diff --git a/internal/git/object_id_test.go b/internal/git/object_id_test.go
index f637fec3e..a6b62ffdc 100644
--- a/internal/git/object_id_test.go
+++ b/internal/git/object_id_test.go
@@ -6,13 +6,104 @@ import (
 	"bytes"
 	"encoding/hex"
 	"fmt"
+	"path/filepath"
 	"strings"
 	"testing"
 
 	"github.com/stretchr/testify/require"
 	"gitlab.com/gitlab-org/gitaly/v15/internal/git"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/git/gittest"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/git/localrepo"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/helper/text"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/testhelper"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/testhelper/testcfg"
+	"gitlab.com/gitlab-org/gitaly/v15/proto/go/gitalypb"
 )
 
+func TestDetectObjectHash(t *testing.T) {
+	cfg := testcfg.Build(t)
+	ctx := testhelper.Context(t)
+
+	for _, tc := range []struct {
+		desc         string
+		setup        func(t *testing.T) *gitalypb.Repository
+		expectedErr  error
+		expectedHash git.ObjectHash
+	}{
+		{
+			desc: "defaults to SHA1",
+			setup: func(t *testing.T) *gitalypb.Repository {
+				repo, repoPath := gittest.InitRepo(t, cfg, cfg.Storages[0], gittest.InitRepoOpts{
+					ObjectFormat: "sha1",
+				})
+
+				// Verify that the repo doesn't explicitly mention it's using SHA1
+				// as object hash.
+				content := testhelper.MustReadFile(t, filepath.Join(repoPath, "config"))
+				require.NotContains(t, text.ChompBytes(content), "sha1")
+
+				return repo
+			},
+			expectedHash: git.ObjectHashSHA1,
+		},
+		{
+			desc: "explicitly set to SHA1",
+			setup: func(t *testing.T) *gitalypb.Repository {
+				repo, repoPath := gittest.InitRepo(t, cfg, cfg.Storages[0], gittest.InitRepoOpts{
+					ObjectFormat: "sha1",
+				})
+
+				// Explicitly set the object format to SHA1.
+				gittest.Exec(t, cfg, "-C", repoPath, "config", "extensions.objectFormat", "sha1")
+
+				return repo
+			},
+			expectedHash: git.ObjectHashSHA1,
+		},
+		{
+			desc: "explicitly set to SHA256",
+			setup: func(t *testing.T) *gitalypb.Repository {
+				repo, repoPath := gittest.InitRepo(t, cfg, cfg.Storages[0], gittest.InitRepoOpts{
+					ObjectFormat: "sha256",
+				})
+
+				require.Equal(t,
+					"sha256",
+					text.ChompBytes(gittest.Exec(t, cfg, "-C", repoPath, "config", "extensions.objectFormat")),
+				)
+
+				return repo
+			},
+			expectedHash: git.ObjectHashSHA256,
+		},
+		{
+			desc: "unknown hash",
+			setup: func(t *testing.T) *gitalypb.Repository {
+				repo, repoPath := gittest.InitRepo(t, cfg, cfg.Storages[0])
+
+				// Explicitly set the object format to something unknown.
+				gittest.Exec(t, cfg, "-C", repoPath, "config", "extensions.objectFormat", "blake2")
+
+				return repo
+			},
+			expectedErr: fmt.Errorf("reading object format: exit status 128"),
+		},
+	} {
+		t.Run(tc.desc, func(t *testing.T) {
+			repoProto := tc.setup(t)
+			repo := localrepo.NewTestRepo(t, cfg, repoProto)
+
+			hash, err := git.DetectObjectHash(ctx, repo)
+			if tc.expectedErr != nil {
+				require.EqualError(t, err, tc.expectedErr.Error())
+			} else {
+				require.NoError(t, err)
+			}
+			require.Equal(t, tc.expectedHash, hash)
+		})
+	}
+}
+
 func TestObjectHash_ValidateHex(t *testing.T) {
 	for _, hash := range []struct {
 		desc     string