linguist: Split function to initialize languageStats

The functionality of newLanguageStats() was a bit overloaded. Split it
up and stop returning pointers.

3 files changed, 36 insertions, 32 deletions

diff --git a/internal/gitaly/linguist/language_stats.go b/internal/gitaly/linguist/language_stats.go
index cefa3b644..5ae08c3d1 100644
--- a/internal/gitaly/linguist/language_stats.go
+++ b/internal/gitaly/linguist/language_stats.go
@@ -28,7 +28,7 @@ type languageStats struct {
 	CommitID string `json:"commit_id"`
 
 	// m will protect concurrent writes to Totals & ByFile maps
-	m sync.Mutex
+	m *sync.Mutex
 
 	// Totals contains the total statistics for the CommitID
 	Totals ByteCountPerLanguage `json:"totals"`
@@ -37,43 +37,47 @@ type languageStats struct {
 	ByFile map[string]ByteCountPerLanguage `json:"by_file"`
 }
 
-// newLanguageStats creates a languageStats object and tries to load the
-// optionally available stats from file.
-func newLanguageStats(repo *localrepo.Repo) (*languageStats, error) {
-	stats := languageStats{
+func newLanguageStats() languageStats {
+	return languageStats{
 		Totals: ByteCountPerLanguage{},
 		ByFile: make(map[string]ByteCountPerLanguage),
+		m:      &sync.Mutex{},
 	}
+}
 
+// initLanguageStats tries to load the optionally available stats from file or
+// returns a blank languageStats struct.
+func initLanguageStats(repo *localrepo.Repo) (languageStats, error) {
 	objPath, err := repo.Path()
 	if err != nil {
-		return &stats, fmt.Errorf("new language stats get repo path: %w", err)
+		return newLanguageStats(), fmt.Errorf("new language stats get repo path: %w", err)
 	}
 
 	file, err := os.Open(filepath.Join(objPath, languageStatsFilename))
 	if err != nil {
 		if os.IsNotExist(err) {
-			return &stats, nil
+			return newLanguageStats(), nil
 		}
-		return &stats, fmt.Errorf("new language stats open: %w", err)
+		return newLanguageStats(), fmt.Errorf("new language stats open: %w", err)
 	}
 	defer file.Close()
 
 	r, err := zlib.NewReader(file)
 	if err != nil {
-		return &stats, fmt.Errorf("new language stats zlib reader: %w", err)
+		return newLanguageStats(), fmt.Errorf("new language stats zlib reader: %w", err)
 	}
 
 	var loaded languageStats
 	if err = json.NewDecoder(r).Decode(&loaded); err != nil {
-		return &stats, fmt.Errorf("new language stats json decode: %w", err)
+		return newLanguageStats(), fmt.Errorf("new language stats json decode: %w", err)
 	}
 
 	if loaded.Version != languageStatsVersion {
-		return &stats, fmt.Errorf("new language stats version mismatch %s vs %s", languageStatsVersion, loaded.Version)
+		return newLanguageStats(), fmt.Errorf("new language stats version mismatch %s vs %s", languageStatsVersion, loaded.Version)
 	}
 
-	return &loaded, nil
+	loaded.m = &sync.Mutex{}
+	return loaded, nil
 }
 
 // add the statistics for the given filename
diff --git a/internal/gitaly/linguist/language_stats_test.go b/internal/gitaly/linguist/language_stats_test.go
index 9199f6eb8..02ec820c1 100644
--- a/internal/gitaly/linguist/language_stats_test.go
+++ b/internal/gitaly/linguist/language_stats_test.go
@@ -17,7 +17,7 @@ import (
 	"gitlab.com/gitlab-org/gitaly/v15/internal/testhelper/testcfg"
 )
 
-func TestNewLanguageStats(t *testing.T) {
+func TestInitLanguageStats(t *testing.T) {
 	t.Parallel()
 
 	ctx := testhelper.Context(t)
@@ -30,7 +30,7 @@ func TestNewLanguageStats(t *testing.T) {
 		{
 			desc: "non-existing cache",
 			run: func(t *testing.T, repo *localrepo.Repo, repoPath string) {
-				stats, err := newLanguageStats(repo)
+				stats, err := initLanguageStats(repo)
 				require.NoError(t, err)
 				require.Empty(t, stats.Totals)
 				require.Empty(t, stats.ByFile)
@@ -39,7 +39,7 @@ func TestNewLanguageStats(t *testing.T) {
 		{
 			desc: "pre-existing cache",
 			run: func(t *testing.T, repo *localrepo.Repo, repoPath string) {
-				stats, err := newLanguageStats(repo)
+				stats, err := initLanguageStats(repo)
 				require.NoError(t, err)
 
 				stats.Totals["C"] = 555
@@ -53,7 +53,7 @@ func TestNewLanguageStats(t *testing.T) {
 			run: func(t *testing.T, repo *localrepo.Repo, repoPath string) {
 				require.NoError(t, os.WriteFile(filepath.Join(repoPath, languageStatsFilename), []byte("garbage"), 0o644))
 
-				stats, err := newLanguageStats(repo)
+				stats, err := initLanguageStats(repo)
 				require.Errorf(t, err, "new language stats zlib reader: invalid header")
 				require.Empty(t, stats.Totals)
 				require.Empty(t, stats.ByFile)
@@ -62,7 +62,7 @@ func TestNewLanguageStats(t *testing.T) {
 		{
 			desc: "incorrect version cache",
 			run: func(t *testing.T, repo *localrepo.Repo, repoPath string) {
-				stats, err := newLanguageStats(repo)
+				stats, err := initLanguageStats(repo)
 				require.NoError(t, err)
 
 				stats.Totals["C"] = 555
@@ -77,7 +77,7 @@ func TestNewLanguageStats(t *testing.T) {
 				require.NoError(t, file.Sync())
 				require.NoError(t, file.Close())
 
-				newStats, err := newLanguageStats(repo)
+				newStats, err := initLanguageStats(repo)
 				require.Error(t, err, fmt.Errorf("new language stats version mismatch %s vs %s", languageStatsVersion, "faulty"))
 				require.Empty(t, newStats.Totals)
 			},
@@ -106,11 +106,11 @@ func TestLanguageStats_add(t *testing.T) {
 
 	for _, tc := range []struct {
 		desc string
-		run  func(*testing.T, *languageStats)
+		run  func(*testing.T, languageStats)
 	}{
 		{
 			desc: "adds to the total",
-			run: func(t *testing.T, s *languageStats) {
+			run: func(t *testing.T, s languageStats) {
 				s.add("main.go", "Go", 100)
 
 				require.Equal(t, uint64(100), s.Totals["Go"])
@@ -120,7 +120,7 @@ func TestLanguageStats_add(t *testing.T) {
 		},
 		{
 			desc: "accumulates",
-			run: func(t *testing.T, s *languageStats) {
+			run: func(t *testing.T, s languageStats) {
 				s.add("main.go", "Go", 100)
 				s.add("main_test.go", "Go", 80)
 
@@ -132,7 +132,7 @@ func TestLanguageStats_add(t *testing.T) {
 		},
 		{
 			desc: "languages don't interfere",
-			run: func(t *testing.T, s *languageStats) {
+			run: func(t *testing.T, s languageStats) {
 				s.add("main.go", "Go", 60)
 				s.add("Makefile", "Make", 30)
 
@@ -145,7 +145,7 @@ func TestLanguageStats_add(t *testing.T) {
 		},
 		{
 			desc: "updates the stat for a file",
-			run: func(t *testing.T, s *languageStats) {
+			run: func(t *testing.T, s languageStats) {
 				s.add("main.go", "Go", 60)
 				s.add("main.go", "Go", 30)
 
@@ -156,7 +156,7 @@ func TestLanguageStats_add(t *testing.T) {
 		},
 	} {
 		t.Run(tc.desc, func(t *testing.T) {
-			s, err := newLanguageStats(repo)
+			s, err := initLanguageStats(repo)
 			require.NoError(t, err)
 
 			tc.run(t, s)
@@ -177,11 +177,11 @@ func TestLanguageStats_drop(t *testing.T) {
 
 	for _, tc := range []struct {
 		desc string
-		run  func(*testing.T, *languageStats)
+		run  func(*testing.T, languageStats)
 	}{
 		{
 			desc: "existing file",
-			run: func(t *testing.T, s *languageStats) {
+			run: func(t *testing.T, s languageStats) {
 				s.drop("main.go")
 
 				require.Equal(t, uint64(20), s.Totals["Go"])
@@ -191,7 +191,7 @@ func TestLanguageStats_drop(t *testing.T) {
 		},
 		{
 			desc: "non-existing file",
-			run: func(t *testing.T, s *languageStats) {
+			run: func(t *testing.T, s languageStats) {
 				s.drop("foo.go")
 
 				require.Equal(t, uint64(100), s.Totals["Go"])
@@ -202,7 +202,7 @@ func TestLanguageStats_drop(t *testing.T) {
 		},
 		{
 			desc: "all files",
-			run: func(t *testing.T, s *languageStats) {
+			run: func(t *testing.T, s languageStats) {
 				s.drop("main.go", "main_test.go")
 
 				require.Empty(t, s.Totals)
@@ -211,7 +211,7 @@ func TestLanguageStats_drop(t *testing.T) {
 		},
 	} {
 		t.Run(tc.desc, func(t *testing.T) {
-			s, err := newLanguageStats(repo)
+			s, err := initLanguageStats(repo)
 			require.NoError(t, err)
 
 			s.Totals["Go"] = 100
@@ -233,7 +233,7 @@ func TestLanguageStats_save(t *testing.T) {
 	})
 	repo := localrepo.NewTestRepo(t, cfg, repoProto)
 
-	s, err := newLanguageStats(repo)
+	s, err := initLanguageStats(repo)
 	require.NoError(t, err)
 
 	s.Totals["Go"] = 100
@@ -244,7 +244,7 @@ func TestLanguageStats_save(t *testing.T) {
 	require.NoError(t, err)
 	require.FileExists(t, filepath.Join(repoPath, languageStatsFilename))
 
-	loaded, err := newLanguageStats(repo)
+	loaded, err := initLanguageStats(repo)
 	require.NoError(t, err)
 
 	require.Equal(t, "buzz", loaded.CommitID)
diff --git a/internal/gitaly/linguist/linguist.go b/internal/gitaly/linguist/linguist.go
index 95610c904..af758e8f3 100644
--- a/internal/gitaly/linguist/linguist.go
+++ b/internal/gitaly/linguist/linguist.go
@@ -107,7 +107,7 @@ func (inst *Instance) startGitLinguist(ctx context.Context, commitID string) (*c
 }
 
 func (inst *Instance) enryStats(ctx context.Context, commitID string) (ByteCountPerLanguage, error) {
-	stats, err := newLanguageStats(inst.repo)
+	stats, err := initLanguageStats(inst.repo)
 	if err != nil {
 		ctxlogrus.Extract(ctx).WithError(err).Info("linguist load from cache")
 	}