Adjust gitea doctor --run storages to check all storage types (#21785)

The doctor check `storages` currently only checks the attachment
storage. This PR adds some basic garbage collection functionality for
the other types of storage.

Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>

10 files changed, 296 insertions, 31 deletions

diff --git a/models/git/lfs.go b/models/git/lfs.go
index 74721dabb1..87e07d7a5e 100644
--- a/models/git/lfs.go
+++ b/models/git/lfs.go
@@ -235,9 +235,9 @@ func LFSObjectAccessible(user *user_model.User, oid string) (bool, error) {
 	return count > 0, err
 }
 
-// LFSObjectIsAssociated checks if a provided Oid is associated
-func LFSObjectIsAssociated(oid string) (bool, error) {
-	return db.GetEngine(db.DefaultContext).Exist(&LFSMetaObject{Pointer: lfs.Pointer{Oid: oid}})
+// ExistsLFSObject checks if a provided Oid exists within the DB
+func ExistsLFSObject(ctx context.Context, oid string) (bool, error) {
+	return db.GetEngine(ctx).Exist(&LFSMetaObject{Pointer: lfs.Pointer{Oid: oid}})
 }
 
 // LFSAutoAssociate auto associates accessible LFSMetaObjects
diff --git a/models/packages/package_blob.go b/models/packages/package_blob.go
index 8c701d4285..fcb71a96ec 100644
--- a/models/packages/package_blob.go
+++ b/models/packages/package_blob.go
@@ -62,6 +62,13 @@ func GetBlobByID(ctx context.Context, blobID int64) (*PackageBlob, error) {
 	return pb, nil
 }
 
+// ExistPackageBlobWithSHA returns if a package blob exists with the provided sha
+func ExistPackageBlobWithSHA(ctx context.Context, blobSha256 string) (bool, error) {
+	return db.GetEngine(ctx).Exist(&PackageBlob{
+		HashSHA256: blobSha256,
+	})
+}
+
 // FindExpiredUnreferencedBlobs gets all blobs without associated files older than the specific duration
 func FindExpiredUnreferencedBlobs(ctx context.Context, olderThan time.Duration) ([]*PackageBlob, error) {
 	pbs := make([]*PackageBlob, 0, 10)
diff --git a/models/repo/archiver.go b/models/repo/archiver.go
index 003911943f..84358ce0dc 100644
--- a/models/repo/archiver.go
+++ b/models/repo/archiver.go
@@ -7,11 +7,14 @@ package repo
 import (
 	"context"
 	"fmt"
+	"strconv"
+	"strings"
 	"time"
 
 	"code.gitea.io/gitea/models/db"
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/timeutil"
+	"code.gitea.io/gitea/modules/util"
 
 	"xorm.io/builder"
 )
@@ -44,6 +47,28 @@ func (archiver *RepoArchiver) RelativePath() string {
 	return fmt.Sprintf("%d/%s/%s.%s", archiver.RepoID, archiver.CommitID[:2], archiver.CommitID, archiver.Type.String())
 }
 
+// repoArchiverForRelativePath takes a relativePath created from (archiver *RepoArchiver) RelativePath() and creates a shell repoArchiver struct representing it
+func repoArchiverForRelativePath(relativePath string) (*RepoArchiver, error) {
+	parts := strings.SplitN(relativePath, "/", 3)
+	if len(parts) != 3 {
+		return nil, util.SilentWrap{Message: fmt.Sprintf("invalid storage path: %s", relativePath), Err: util.ErrInvalidArgument}
+	}
+	repoID, err := strconv.ParseInt(parts[0], 10, 64)
+	if err != nil {
+		return nil, util.SilentWrap{Message: fmt.Sprintf("invalid storage path: %s", relativePath), Err: util.ErrInvalidArgument}
+	}
+	nameExts := strings.SplitN(parts[2], ".", 2)
+	if len(nameExts) != 2 {
+		return nil, util.SilentWrap{Message: fmt.Sprintf("invalid storage path: %s", relativePath), Err: util.ErrInvalidArgument}
+	}
+
+	return &RepoArchiver{
+		RepoID:   repoID,
+		CommitID: parts[1] + nameExts[0],
+		Type:     git.ToArchiveType(nameExts[1]),
+	}, nil
+}
+
 var delRepoArchiver = new(RepoArchiver)
 
 // DeleteRepoArchiver delete archiver
@@ -65,6 +90,17 @@ func GetRepoArchiver(ctx context.Context, repoID int64, tp git.ArchiveType, comm
 	return nil, nil
 }
 
+// ExistsRepoArchiverWithStoragePath checks if there is a RepoArchiver for a given storage path
+func ExistsRepoArchiverWithStoragePath(ctx context.Context, storagePath string) (bool, error) {
+	// We need to invert the path provided func (archiver *RepoArchiver) RelativePath() above
+	archiver, err := repoArchiverForRelativePath(storagePath)
+	if err != nil {
+		return false, err
+	}
+
+	return db.GetEngine(ctx).Exist(archiver)
+}
+
 // AddRepoArchiver adds an archiver
 func AddRepoArchiver(ctx context.Context, archiver *RepoArchiver) error {
 	_, err := db.GetEngine(ctx).Insert(archiver)
diff --git a/models/repo/attachment.go b/models/repo/attachment.go
index 180d7730ba..df7528df09 100644
--- a/models/repo/attachment.go
+++ b/models/repo/attachment.go
@@ -122,9 +122,9 @@ func GetAttachmentsByUUIDs(ctx context.Context, uuids []string) ([]*Attachment,
 	return attachments, db.GetEngine(ctx).In("uuid", uuids).Find(&attachments)
 }
 
-// ExistAttachmentsByUUID returns true if attachment is exist by given UUID
-func ExistAttachmentsByUUID(uuid string) (bool, error) {
-	return db.GetEngine(db.DefaultContext).Where("`uuid`=?", uuid).Exist(new(Attachment))
+// ExistAttachmentsByUUID returns true if attachment exists with the given UUID
+func ExistAttachmentsByUUID(ctx context.Context, uuid string) (bool, error) {
+	return db.GetEngine(ctx).Where("`uuid`=?", uuid).Exist(new(Attachment))
 }
 
 // GetAttachmentsByIssueID returns all attachments of an issue.
diff --git a/models/repo/avatar.go b/models/repo/avatar.go
index 1bc37598fe..84b9f5ac21 100644
--- a/models/repo/avatar.go
+++ b/models/repo/avatar.go
@@ -24,6 +24,13 @@ func (repo *Repository) CustomAvatarRelativePath() string {
 	return repo.Avatar
 }
 
+// ExistsWithAvatarAtStoragePath returns true if there is a user with this Avatar
+func ExistsWithAvatarAtStoragePath(ctx context.Context, storagePath string) (bool, error) {
+	// See func (repo *Repository) CustomAvatarRelativePath()
+	// repo.Avatar is used directly as the storage path - therefore we can check for existence directly using the path
+	return db.GetEngine(ctx).Where("`avatar`=?", storagePath).Exist(new(Repository))
+}
+
 // RelAvatarLink returns a relative link to the repository's avatar.
 func (repo *Repository) RelAvatarLink() string {
 	return repo.relAvatarLink(db.DefaultContext)
diff --git a/models/user/avatar.go b/models/user/avatar.go
index 102206f3a2..f523766746 100644
--- a/models/user/avatar.go
+++ b/models/user/avatar.go
@@ -111,3 +111,10 @@ func (u *User) IsUploadAvatarChanged(data []byte) bool {
 	avatarID := fmt.Sprintf("%x", md5.Sum([]byte(fmt.Sprintf("%d-%x", u.ID, md5.Sum(data)))))
 	return u.Avatar != avatarID
 }
+
+// ExistsWithAvatarAtStoragePath returns true if there is a user with this Avatar
+func ExistsWithAvatarAtStoragePath(ctx context.Context, storagePath string) (bool, error) {
+	// See func (u *User) CustomAvatarRelativePath()
+	// u.Avatar is used directly as the storage path - therefore we can check for existence directly using the path
+	return db.GetEngine(ctx).Where("`avatar`=?", storagePath).Exist(new(User))
+}
diff --git a/modules/doctor/storage.go b/modules/doctor/storage.go
index dafd989cf0..8ae9168ea6 100644
--- a/modules/doctor/storage.go
+++ b/modules/doctor/storage.go
@@ -6,71 +6,255 @@ package doctor
 
 import (
 	"context"
+	"errors"
+	"io/fs"
+	"strings"
 
-	repo_model "code.gitea.io/gitea/models/repo"
+	"code.gitea.io/gitea/models/git"
+	"code.gitea.io/gitea/models/packages"
+	"code.gitea.io/gitea/models/repo"
+	"code.gitea.io/gitea/models/user"
+	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/log"
+	packages_module "code.gitea.io/gitea/modules/packages"
 	"code.gitea.io/gitea/modules/storage"
+	"code.gitea.io/gitea/modules/util"
 )
 
-func checkAttachmentStorageFiles(logger log.Logger, autofix bool) error {
-	var total, garbageNum int
-	var deletePaths []string
-	if err := storage.Attachments.IterateObjects(func(p string, obj storage.Object) error {
+type commonStorageCheckOptions struct {
+	storer     storage.ObjectStorage
+	isOrphaned func(path string, obj storage.Object, stat fs.FileInfo) (bool, error)
+	name       string
+}
+
+func commonCheckStorage(ctx context.Context, logger log.Logger, autofix bool, opts *commonStorageCheckOptions) error {
+	totalCount, orphanedCount := 0, 0
+	totalSize, orphanedSize := int64(0), int64(0)
+
+	var pathsToDelete []string
+	if err := opts.storer.IterateObjects(func(p string, obj storage.Object) error {
 		defer obj.Close()
 
-		total++
+		totalCount++
 		stat, err := obj.Stat()
 		if err != nil {
 			return err
 		}
-		exist, err := repo_model.ExistAttachmentsByUUID(stat.Name())
+		totalSize += stat.Size()
+
+		orphaned, err := opts.isOrphaned(p, obj, stat)
 		if err != nil {
 			return err
 		}
-		if !exist {
-			garbageNum++
+		if orphaned {
+			orphanedCount++
+			orphanedSize += stat.Size()
 			if autofix {
-				deletePaths = append(deletePaths, p)
+				pathsToDelete = append(pathsToDelete, p)
 			}
 		}
 		return nil
 	}); err != nil {
-		logger.Error("storage.Attachments.IterateObjects failed: %v", err)
+		logger.Error("Error whilst iterating %s storage: %v", opts.name, err)
 		return err
 	}
 
-	if garbageNum > 0 {
+	if orphanedCount > 0 {
 		if autofix {
 			var deletedNum int
-			for _, p := range deletePaths {
-				if err := storage.Attachments.Delete(p); err != nil {
-					log.Error("Delete attachment %s failed: %v", p, err)
+			for _, p := range pathsToDelete {
+				if err := opts.storer.Delete(p); err != nil {
+					log.Error("Error whilst deleting %s from %s storage: %v", p, opts.name, err)
 				} else {
 					deletedNum++
 				}
 			}
-			logger.Info("%d missed information attachment detected, %d deleted.", garbageNum, deletedNum)
+			logger.Info("Deleted %d/%d orphaned %s(s)", deletedNum, orphanedCount, opts.name)
 		} else {
-			logger.Warn("Checked %d attachment, %d missed information.", total, garbageNum)
+			logger.Warn("Found %d/%d (%s/%s) orphaned %s(s)", orphanedCount, totalCount, base.FileSize(orphanedSize), base.FileSize(totalSize), opts.name)
 		}
+	} else {
+		logger.Info("Found %d (%s) %s(s)", totalCount, base.FileSize(totalSize), opts.name)
 	}
 	return nil
 }
 
-func checkStorageFiles(ctx context.Context, logger log.Logger, autofix bool) error {
-	if err := storage.Init(); err != nil {
-		logger.Error("storage.Init failed: %v", err)
-		return err
+type checkStorageOptions struct {
+	All          bool
+	Attachments  bool
+	LFS          bool
+	Avatars      bool
+	RepoAvatars  bool
+	RepoArchives bool
+	Packages     bool
+}
+
+// checkStorage will return a doctor check function to check the requested storage types for "orphaned" stored object/files and optionally delete them
+func checkStorage(opts *checkStorageOptions) func(ctx context.Context, logger log.Logger, autofix bool) error {
+	return func(ctx context.Context, logger log.Logger, autofix bool) error {
+		if err := storage.Init(); err != nil {
+			logger.Error("storage.Init failed: %v", err)
+			return err
+		}
+
+		if opts.Attachments || opts.All {
+			if err := commonCheckStorage(ctx, logger, autofix,
+				&commonStorageCheckOptions{
+					storer: storage.Attachments,
+					isOrphaned: func(path string, obj storage.Object, stat fs.FileInfo) (bool, error) {
+						exists, err := repo.ExistAttachmentsByUUID(ctx, stat.Name())
+						return !exists, err
+					},
+					name: "attachment",
+				}); err != nil {
+				return err
+			}
+		}
+
+		if opts.LFS || opts.All {
+			if err := commonCheckStorage(ctx, logger, autofix,
+				&commonStorageCheckOptions{
+					storer: storage.LFS,
+					isOrphaned: func(path string, obj storage.Object, stat fs.FileInfo) (bool, error) {
+						// The oid of an LFS stored object is the name but with all the path.Separators removed
+						oid := strings.ReplaceAll(path, "/", "")
+						exists, err := git.ExistsLFSObject(ctx, oid)
+						return !exists, err
+					},
+					name: "LFS file",
+				}); err != nil {
+				return err
+			}
+		}
+
+		if opts.Avatars || opts.All {
+			if err := commonCheckStorage(ctx, logger, autofix,
+				&commonStorageCheckOptions{
+					storer: storage.Avatars,
+					isOrphaned: func(path string, obj storage.Object, stat fs.FileInfo) (bool, error) {
+						exists, err := user.ExistsWithAvatarAtStoragePath(ctx, path)
+						return !exists, err
+					},
+					name: "avatar",
+				}); err != nil {
+				return err
+			}
+		}
+
+		if opts.RepoAvatars || opts.All {
+			if err := commonCheckStorage(ctx, logger, autofix,
+				&commonStorageCheckOptions{
+					storer: storage.RepoAvatars,
+					isOrphaned: func(path string, obj storage.Object, stat fs.FileInfo) (bool, error) {
+						exists, err := repo.ExistsWithAvatarAtStoragePath(ctx, path)
+						return !exists, err
+					},
+					name: "repo avatar",
+				}); err != nil {
+				return err
+			}
+		}
+
+		if opts.RepoArchives || opts.All {
+			if err := commonCheckStorage(ctx, logger, autofix,
+				&commonStorageCheckOptions{
+					storer: storage.RepoAvatars,
+					isOrphaned: func(path string, obj storage.Object, stat fs.FileInfo) (bool, error) {
+						exists, err := repo.ExistsRepoArchiverWithStoragePath(ctx, path)
+						if err == nil || errors.Is(err, util.ErrInvalidArgument) {
+							// invalid arguments mean that the object is not a valid repo archiver and it should be removed
+							return !exists, nil
+						}
+						return !exists, err
+					},
+					name: "repo archive",
+				}); err != nil {
+				return err
+			}
+		}
+
+		if opts.Packages || opts.All {
+			if err := commonCheckStorage(ctx, logger, autofix,
+				&commonStorageCheckOptions{
+					storer: storage.Packages,
+					isOrphaned: func(path string, obj storage.Object, stat fs.FileInfo) (bool, error) {
+						key, err := packages_module.RelativePathToKey(path)
+						if err != nil {
+							// If there is an error here then the relative path does not match a valid package
+							// Therefore it is orphaned by default
+							return true, nil
+						}
+
+						exists, err := packages.ExistPackageBlobWithSHA(ctx, string(key))
+
+						return !exists, err
+					},
+					name: "package blob",
+				}); err != nil {
+				return err
+			}
+		}
+
+		return nil
 	}
-	return checkAttachmentStorageFiles(logger, autofix)
 }
 
 func init() {
 	Register(&Check{
-		Title:                      "Check if there is garbage storage files",
+		Title:                      "Check if there are orphaned storage files",
 		Name:                       "storages",
 		IsDefault:                  false,
-		Run:                        checkStorageFiles,
+		Run:                        checkStorage(&checkStorageOptions{All: true}),
+		AbortIfFailed:              false,
+		SkipDatabaseInitialization: false,
+		Priority:                   1,
+	})
+
+	Register(&Check{
+		Title:                      "Check if there are orphaned attachments in storage",
+		Name:                       "storage-attachments",
+		IsDefault:                  false,
+		Run:                        checkStorage(&checkStorageOptions{Attachments: true}),
+		AbortIfFailed:              false,
+		SkipDatabaseInitialization: false,
+		Priority:                   1,
+	})
+
+	Register(&Check{
+		Title:                      "Check if there are orphaned lfs files in storage",
+		Name:                       "storage-lfs",
+		IsDefault:                  false,
+		Run:                        checkStorage(&checkStorageOptions{LFS: true}),
+		AbortIfFailed:              false,
+		SkipDatabaseInitialization: false,
+		Priority:                   1,
+	})
+
+	Register(&Check{
+		Title:                      "Check if there are orphaned avatars in storage",
+		Name:                       "storage-avatars",
+		IsDefault:                  false,
+		Run:                        checkStorage(&checkStorageOptions{Avatars: true, RepoAvatars: true}),
+		AbortIfFailed:              false,
+		SkipDatabaseInitialization: false,
+		Priority:                   1,
+	})
+
+	Register(&Check{
+		Title:                      "Check if there are orphaned archives in storage",
+		Name:                       "storage-archives",
+		IsDefault:                  false,
+		Run:                        checkStorage(&checkStorageOptions{RepoArchives: true}),
+		AbortIfFailed:              false,
+		SkipDatabaseInitialization: false,
+		Priority:                   1,
+	})
+
+	Register(&Check{
+		Title:                      "Check if there are orphaned package blobs in storage",
+		Name:                       "storage-packages",
+		IsDefault:                  false,
+		Run:                        checkStorage(&checkStorageOptions{Packages: true}),
 		AbortIfFailed:              false,
 		SkipDatabaseInitialization: false,
 		Priority:                   1,
diff --git a/modules/git/repo_archive.go b/modules/git/repo_archive.go
index a0cbfba5d9..13be2004ca 100644
--- a/modules/git/repo_archive.go
+++ b/modules/git/repo_archive.go
@@ -38,6 +38,18 @@ func (a ArchiveType) String() string {
 	return "unknown"
 }
 
+func ToArchiveType(s string) ArchiveType {
+	switch s {
+	case "zip":
+		return ZIP
+	case "tar.gz":
+		return TARGZ
+	case "bundle":
+		return BUNDLE
+	}
+	return 0
+}
+
 // CreateArchive create archive content to the target path
 func (repo *Repository) CreateArchive(ctx context.Context, format ArchiveType, target io.Writer, usePrefix bool, commitID string) error {
 	if format.String() == "unknown" {
diff --git a/modules/packages/content_store.go b/modules/packages/content_store.go
index a3a5d1a666..be416ac269 100644
--- a/modules/packages/content_store.go
+++ b/modules/packages/content_store.go
@@ -7,8 +7,10 @@ package packages
 import (
 	"io"
 	"path"
+	"strings"
 
 	"code.gitea.io/gitea/modules/storage"
+	"code.gitea.io/gitea/modules/util"
 )
 
 // BlobHash256Key is the key to address a blob content
@@ -45,3 +47,13 @@ func (s *ContentStore) Delete(key BlobHash256Key) error {
 func KeyToRelativePath(key BlobHash256Key) string {
 	return path.Join(string(key)[0:2], string(key)[2:4], string(key))
 }
+
+// RelativePathToKey converts a relative path aa/bb/aabb000000... to the sha256 key aabb000000...
+func RelativePathToKey(relativePath string) (BlobHash256Key, error) {
+	parts := strings.SplitN(relativePath, "/", 3)
+	if len(parts) != 3 || len(parts[0]) != 2 || len(parts[1]) != 2 || len(parts[2]) < 4 || parts[0]+parts[1] != parts[2][0:4] {
+		return "", util.ErrInvalidArgument
+	}
+
+	return BlobHash256Key(parts[2]), nil
+}
diff --git a/routers/web/repo/lfs.go b/routers/web/repo/lfs.go
index 67cb6837a5..9bf4307bfe 100644
--- a/routers/web/repo/lfs.go
+++ b/routers/web/repo/lfs.go
@@ -478,7 +478,7 @@ func LFSPointerFiles(ctx *context.Context) {
 						return err
 					}
 					if !result.Associatable {
-						associated, err := git_model.LFSObjectIsAssociated(pointerBlob.Oid)
+						associated, err := git_model.ExistsLFSObject(ctx, pointerBlob.Oid)
 						if err != nil {
 							return err
 						}