Clean missing captions during scan (#3240)

internal/manager/task_scan.go

@@ -21,6 +21,7 @@ import (
 	"github.com/stashapp/stash/pkg/models"
 	"github.com/stashapp/stash/pkg/scene"
 	"github.com/stashapp/stash/pkg/scene/generate"
+	"github.com/stashapp/stash/pkg/txn"
 )
 
 type scanner interface {
@@ -111,9 +112,11 @@ type sceneFinder interface {
 // handlerRequiredFilter returns true if a File's handler needs to be executed despite the file not being updated.
 type handlerRequiredFilter struct {
 	extensionConfig
+	txnManager     txn.Manager
 	SceneFinder    sceneFinder
 	ImageFinder    fileCounter
 	GalleryFinder  galleryFinder
+	CaptionUpdater video.CaptionUpdater
 
 	FolderCache *lru.LRU
 
@@ -126,9 +129,11 @@ func newHandlerRequiredFilter(c *config.Instance) *handlerRequiredFilter {
 
 	return &handlerRequiredFilter{
 		extensionConfig:          newExtensionConfig(c),
+		txnManager:               db,
 		SceneFinder:              db.Scene,
 		ImageFinder:              db.Image,
 		GalleryFinder:            db.Gallery,
+		CaptionUpdater:           db.File,
 		FolderCache:              lru.New(processes * 2),
 		videoFileNamingAlgorithm: c.GetVideoFileNamingAlgorithm(),
 	}
@@ -205,6 +210,15 @@ func (f *handlerRequiredFilter) Accept(ctx context.Context, ff file.File) bool {
 				return true
 			}
 		}
+
+		// clean captions - scene handler handles this as well, but
+		// unchanged files aren't processed by the scene handler
+		videoFile, _ := ff.(*file.VideoFile)
+		if videoFile != nil {
+			if err := video.CleanCaptions(ctx, videoFile, f.txnManager, f.CaptionUpdater); err != nil {
+				logger.Errorf("Error cleaning captions: %v", err)
+			}
+		}
 	}
 
 	return false
@@ -329,6 +343,7 @@ func getScanHandlers(options ScanMetadataInput, taskQueue *job.TaskQueue, progre
 			Handler: &scene.ScanHandler{
 				CreatorUpdater: db.Scene,
 				PluginCache:    pluginCache,
+				CaptionUpdater: db.File,
 				CoverGenerator: &coverGenerator{},
 				ScanGenerator: &sceneGenerators{
 					input:     options,

pkg/file/video/caption.go

@@ -2,6 +2,7 @@ package video
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -59,23 +60,6 @@ func IsLangInCaptions(lang string, ext string, captions []*models.VideoCaption)
 	return false
 }
 
-// CleanCaptions removes non existent/accessible language codes from captions
-func CleanCaptions(scenePath string, captions []*models.VideoCaption) (cleanedCaptions []*models.VideoCaption, changed bool) {
-	changed = false
-	for _, caption := range captions {
-		found := false
-		f := caption.Path(scenePath)
-		if _, er := os.Stat(f); er == nil {
-			cleanedCaptions = append(cleanedCaptions, caption)
-			found = true
-		}
-		if !found {
-			changed = true
-		}
-	}
-	return
-}
-
 // getCaptionPrefix returns the prefix used to search for video files for the provided caption path
 func getCaptionPrefix(captionPath string) string {
 	basename := strings.TrimSuffix(captionPath, filepath.Ext(captionPath)) // caption filename without the extension
@@ -148,3 +132,52 @@ func AssociateCaptions(ctx context.Context, captionPath string, txnMgr txn.Manag
 		logger.Error(err.Error())
 	}
 }
+
+// CleanCaptions removes non existent/accessible language codes from captions
+func CleanCaptions(ctx context.Context, f *file.VideoFile, txnMgr txn.Manager, w CaptionUpdater) error {
+	captions, err := w.GetCaptions(ctx, f.ID)
+	if err != nil {
+		return fmt.Errorf("getting captions for file %s: %w", f.Path, err)
+	}
+
+	if len(captions) == 0 {
+		return nil
+	}
+
+	filePath := f.Path
+
+	changed := false
+	var newCaptions []*models.VideoCaption
+
+	for _, caption := range captions {
+		captionPath := caption.Path(filePath)
+		_, err := os.Stat(captionPath)
+		if errors.Is(err, os.ErrNotExist) {
+			logger.Infof("Removing non existent caption %s for %s", caption.Filename, f.Path)
+			changed = true
+		} else {
+			// other errors are ignored for the purposes of cleaning
+			newCaptions = append(newCaptions, caption)
+		}
+	}
+
+	if changed {
+		fn := func(ctx context.Context) error {
+			return w.UpdateCaptions(ctx, f.ID, newCaptions)
+		}
+
+		// possible that we are already in a transaction and txnMgr is nil
+		// in that case just call the function directly
+		if txnMgr == nil {
+			err = fn(ctx)
+		} else {
+			err = txn.WithTxn(ctx, txnMgr, fn)
+		}
+
+		if err != nil {
+			return fmt.Errorf("updating captions for file %s: %w", f.Path, err)
+		}
+	}
+
+	return nil
+}

pkg/scene/scan.go

@@ -7,6 +7,7 @@ import (
 	"time"
 
 	"github.com/stashapp/stash/pkg/file"
+	"github.com/stashapp/stash/pkg/file/video"
 	"github.com/stashapp/stash/pkg/logger"
 	"github.com/stashapp/stash/pkg/models"
 	"github.com/stashapp/stash/pkg/models/paths"
@@ -36,6 +37,7 @@ type ScanHandler struct {
 
 	CoverGenerator CoverGenerator
 	ScanGenerator  ScanGenerator
+	CaptionUpdater video.CaptionUpdater
 	PluginCache    *plugin.Cache
 
 	FileNamingAlgorithm models.HashAlgorithm
@@ -52,6 +54,9 @@ func (h *ScanHandler) validate() error {
 	if h.ScanGenerator == nil {
 		return errors.New("ScanGenerator is required")
 	}
+	if h.CaptionUpdater == nil {
+		return errors.New("CaptionUpdater is required")
+	}
 	if !h.FileNamingAlgorithm.IsValid() {
 		return errors.New("FileNamingAlgorithm is required")
 	}
@@ -72,6 +77,12 @@ func (h *ScanHandler) Handle(ctx context.Context, f file.File, oldFile file.File
 		return ErrNotVideoFile
 	}
 
+	if oldFile != nil {
+		if err := video.CleanCaptions(ctx, videoFile, nil, h.CaptionUpdater); err != nil {
+			return fmt.Errorf("cleaning captions: %w", err)
+		}
+	}
+
 	// try to match the file to a scene
 	existing, err := h.CreatorUpdater.FindByFileID(ctx, f.Base().ID)
 	if err != nil {

ui/v2.5/src/docs/en/Changelog/v0190.md

@@ -7,3 +7,6 @@
 
 ### 🎨 Improvements
 * Changed performer aliases to be a list, rather than a string field. ([#3113](https://github.com/stashapp/stash/pull/3113))
+
+### 🐛 Bug fixes
+* Fixed missing captions not being removed during scan. ([#3240](https://github.com/stashapp/stash/pull/3240))