Improve handling of moved and added video files (stashapp#4598)

* If old file path is not in library, treat as move
* Use existing phash if file with same oshash exists

internal/manager/manager_tasks.go

@@ -19,14 +19,17 @@ import (
 )
 
 func useAsVideo(pathname string) bool {
-	if instance.Config.IsCreateImageClipsFromVideos() && config.StashConfigs.GetStashFromDirPath(instance.Config.GetStashPaths(), pathname).ExcludeVideo {
+	stash := config.StashConfigs.GetStashFromDirPath(instance.Config.GetStashPaths(), pathname)
+
+	if instance.Config.IsCreateImageClipsFromVideos() && stash != nil && stash.ExcludeVideo {
 		return false
 	}
 	return isVideo(pathname)
 }
 
 func useAsImage(pathname string) bool {
-	if instance.Config.IsCreateImageClipsFromVideos() && config.StashConfigs.GetStashFromDirPath(instance.Config.GetStashPaths(), pathname).ExcludeVideo {
+	stash := config.StashConfigs.GetStashFromDirPath(instance.Config.GetStashPaths(), pathname)
+	if instance.Config.IsCreateImageClipsFromVideos() && stash != nil && stash.ExcludeVideo {
 		return isImage(pathname) || isVideo(pathname)
 	}
 	return isImage(pathname)

internal/manager/task_generate_phash.go

@@ -25,19 +25,38 @@ func (t *GeneratePhashTask) Start(ctx context.Context) {
 		return
 	}
 
-	hash, err := videophash.Generate(instance.FFMpeg, t.File)
-	if err != nil {
-		logger.Errorf("error generating phash: %s", err.Error())
-		logErrorOutput(err)
-		return
+	var hash int64
+	set := false
+
+	// #4393 - if there is a file with the same oshash, we can use the same phash
+	// only use this if we're not overwriting
+	if !t.Overwrite {
+		existing, err := t.findExistingPhash(ctx)
+		if err != nil {
+			logger.Warnf("Error finding existing phash: %v", err)
+		} else if existing != nil {
+			logger.Infof("Using existing phash for %s", t.File.Path)
+			hash = existing.(int64)
+			set = true
+		}
+	}
+
+	if !set {
+		generated, err := videophash.Generate(instance.FFMpeg, t.File)
+		if err != nil {
+			logger.Errorf("Error generating phash: %v", err)
+			logErrorOutput(err)
+			return
+		}
+
+		hash = int64(*generated)
 	}
 
 	r := t.repository
 	if err := r.WithTxn(ctx, func(ctx context.Context) error {
-		hashValue := int64(*hash)
 		t.File.Fingerprints = t.File.Fingerprints.AppendUnique(models.Fingerprint{
 			Type:        models.FingerprintTypePhash,
-			Fingerprint: hashValue,
+			Fingerprint: hash,
 		})
 
 		return r.File.Update(ctx, t.File)
@@ -46,6 +65,36 @@ func (t *GeneratePhashTask) Start(ctx context.Context) {
 	}
 }
 
+func (t *GeneratePhashTask) findExistingPhash(ctx context.Context) (interface{}, error) {
+	r := t.repository
+	var ret interface{}
+	if err := r.WithReadTxn(ctx, func(ctx context.Context) error {
+		oshash := t.File.Fingerprints.Get(models.FingerprintTypeOshash)
+
+		// find other files with the same oshash
+		files, err := r.File.FindByFingerprint(ctx, models.Fingerprint{
+			Type:        models.FingerprintTypeOshash,
+			Fingerprint: oshash,
+		})
+		if err != nil {
+			return fmt.Errorf("finding files by oshash: %w", err)
+		}
+
+		// find the first file with a phash
+		for _, file := range files {
+			if phash := file.Base().Fingerprints.Get(models.FingerprintTypePhash); phash != nil {
+				ret = phash
+				return nil
+			}
+		}
+		return nil
+	}); err != nil {
+		return nil, err
+	}
+
+	return ret, nil
+}
+
 func (t *GeneratePhashTask) required() bool {
 	if t.Overwrite {
 		return true

internal/manager/task_scan.go

@@ -264,6 +264,12 @@ func (f *scanFilter) Accept(ctx context.Context, path string, info fs.FileInfo)
 		return false
 	}
 
+	s := f.stashPaths.GetStashFromDirPath(path)
+	if s == nil {
+		logger.Debugf("Skipping %s as it is not in the stash library", path)
+		return false
+	}
+
 	isVideoFile := useAsVideo(path)
 	isImageFile := useAsImage(path)
 	isZipFile := fsutil.MatchExtension(path, f.zipExt)
@@ -288,13 +294,6 @@ func (f *scanFilter) Accept(ctx context.Context, path string, info fs.FileInfo)
 		return false
 	}
 
-	s := f.stashPaths.GetStashFromDirPath(path)
-
-	if s == nil {
-		logger.Debugf("Skipping %s as it is not in the stash library", path)
-		return false
-	}
-
 	// shortcut: skip the directory entirely if it matches both exclusion patterns
 	// add a trailing separator so that it correctly matches against patterns like path/.*
 	pathExcludeTest := path + string(filepath.Separator)

pkg/file/scan.go

@@ -867,16 +867,22 @@ func (s *scanJob) handleRename(ctx context.Context, f models.File, fp []models.F
 			continue
 		}
 
-		if _, err := fs.Lstat(other.Base().Path); err != nil {
+		info, err := fs.Lstat(other.Base().Path)
+		switch {
+		case err != nil:
 			missing = append(missing, other)
-		} else if strings.EqualFold(f.Base().Path, other.Base().Path) {
+		case strings.EqualFold(f.Base().Path, other.Base().Path):
 			// #1426 - if file exists but is a case-insensitive match for the
 			// original filename, and the filesystem is case-insensitive
 			// then treat it as a move
 			if caseSensitive, _ := fs.IsPathCaseSensitive(other.Base().Path); !caseSensitive {
 				// treat as a move
 				missing = append(missing, other)
 			}
+		case !s.acceptEntry(ctx, other.Base().Path, info):
+			// #4393 - if the file is no longer in the configured library paths, treat it as a move
+			logger.Debugf("File %q no longer in library paths. Treating as a move.", other.Base().Path)
+			missing = append(missing, other)
 		}
 	}