Stop using git count-objects and use raw directory size for repository (#8848)

* Migrate from git count-objects to a raw directory size
* As per @guillep2k ignore unusual files

4 files changed, 26 insertions, 8 deletions

diff --git a/models/migrations/v28.go b/models/migrations/v28.go
index 587e944ce6..a849fea3c2 100644
--- a/models/migrations/v28.go
+++ b/models/migrations/v28.go
@@ -60,9 +60,9 @@ func addRepoSize(x *xorm.Engine) (err error) {
 			}
 
 			repoPath := filepath.Join(setting.RepoRootPath, strings.ToLower(user.Name), strings.ToLower(repo.Name)) + ".git"
-			countObject, err := git.GetRepoSize(repoPath)
+			countObject, err := git.CountObjects(repoPath)
 			if err != nil {
-				log.Warn("GetRepoSize: %v", err)
+				log.Warn("CountObjects: %v", err)
 				continue
 			}
 
diff --git a/models/repo.go b/models/repo.go
index 90918025fb..d79a8fdf61 100644
--- a/models/repo.go
+++ b/models/repo.go
@@ -36,6 +36,7 @@ import (
 	api "code.gitea.io/gitea/modules/structs"
 	"code.gitea.io/gitea/modules/sync"
 	"code.gitea.io/gitea/modules/timeutil"
+	"code.gitea.io/gitea/modules/util"
 
 	"github.com/mcuadros/go-version"
 	"github.com/unknwon/com"
@@ -708,17 +709,17 @@ func (repo *Repository) IsOwnedBy(userID int64) bool {
 }
 
 func (repo *Repository) updateSize(e Engine) error {
-	repoInfoSize, err := git.GetRepoSize(repo.repoPath(e))
+	size, err := util.GetDirectorySize(repo.repoPath(e))
 	if err != nil {
 		return fmt.Errorf("UpdateSize: %v", err)
 	}
 
-	repo.Size = repoInfoSize.Size + repoInfoSize.SizePack
+	repo.Size = size
 	_, err = e.ID(repo.ID).Cols("size").Update(repo)
 	return err
 }
 
-// UpdateSize updates the repository size, calculating it using git.GetRepoSize
+// UpdateSize updates the repository size, calculating it using util.GetDirectorySize
 func (repo *Repository) UpdateSize() error {
 	return repo.updateSize(x)
 }
diff --git a/modules/git/repo.go b/modules/git/repo.go
index 4c6690b913..80f6109772 100644
--- a/modules/git/repo.go
+++ b/modules/git/repo.go
@@ -304,8 +304,8 @@ const (
 	statSizeGarbage  = "size-garbage: "
 )
 
-// GetRepoSize returns disk consumption for repo in path
-func GetRepoSize(repoPath string) (*CountObject, error) {
+// CountObjects returns the results of git count-objects on the repoPath
+func CountObjects(repoPath string) (*CountObject, error) {
 	cmd := NewCommand("count-objects", "-v")
 	stdout, err := cmd.RunInDir(repoPath)
 	if err != nil {
diff --git a/modules/util/path.go b/modules/util/path.go
index f79334209c..2b198eb6dc 100644
--- a/modules/util/path.go
+++ b/modules/util/path.go
@@ -4,7 +4,10 @@
 
 package util
 
-import "path/filepath"
+import (
+	"os"
+	"path/filepath"
+)
 
 // EnsureAbsolutePath ensure that a path is absolute, making it
 // relative to absoluteBase if necessary
@@ -14,3 +17,17 @@ func EnsureAbsolutePath(path string, absoluteBase string) string {
 	}
 	return filepath.Join(absoluteBase, path)
 }
+
+const notRegularFileMode os.FileMode = os.ModeDir | os.ModeSymlink | os.ModeNamedPipe | os.ModeSocket | os.ModeDevice | os.ModeCharDevice | os.ModeIrregular
+
+// GetDirectorySize returns the dumb disk consumption for a given path
+func GetDirectorySize(path string) (int64, error) {
+	var size int64
+	err := filepath.Walk(path, func(_ string, info os.FileInfo, err error) error {
+		if info != nil && (info.Mode()&notRegularFileMode) == 0 {
+			size += info.Size()
+		}
+		return err
+	})
+	return size, err
+}