fix: preserve stream language metadata using ffprobe and -map_metadata -1

cmd/videnc/main.go

@@ -89,6 +89,11 @@ func processFile(inputPath string, cfg *types.Config, enc *encoder.Encoder, meta
 		return err
 	}
 
+	streamLangs, err := enc.GetStreamLanguages(inputPath)
+	if err != nil {
+		log.ErrorFile(inputPath, "Getting stream languages", err.Error())
+	}
+
 	mediaType := watcher.DetectMediaType(width, height)
 	crf := cfg.Encoding.DVD.CRF
 	preset := cfg.Encoding.DVD.Preset
@@ -128,7 +133,7 @@ func processFile(inputPath string, cfg *types.Config, enc *encoder.Encoder, meta
 		DeleteOrigin: deleteOrigin,
 	}
 
-	if err := enc.Transcode(inputPath, job, meta, interlaced); err != nil {
+	if err := enc.Transcode(inputPath, job, meta, interlaced, streamLangs); err != nil {
 		log.ErrorFile(inputPath, "Transcoding", err.Error())
 		mover.MoveToFailed(inputPath, cfg.Paths.Failed)
 		outputPath := filepath.Join(filepath.Dir(inputPath), "output.mkv")

internal/encoder/encoder.go

@@ -30,6 +30,54 @@ type ProbeResult struct {
 	Streams []StreamInfo `json:"streams"`
 }
 
+type StreamMetadata struct {
+	Index     int    `json:"index"`
+	CodecType string `json:"codec_type"`
+	CodecName string `json:"codec_name"`
+	Language  string `json:"tags"`
+	Title     string `json:"title"`
+}
+
+func (e *Encoder) GetStreamLanguages(path string) ([]StreamMetadata, error) {
+	cmd := exec.Command(e.ffprobePath, "-v", "error", "-show_streams", "-print_format", "json", path)
+	output, err := cmd.CombinedOutput()
+	if err != nil {
+		return nil, fmt.Errorf("ffprobe error: %w", err)
+	}
+
+	var result struct {
+		Streams []struct {
+			Index     int               `json:"index"`
+			CodecType string            `json:"codec_type"`
+			CodecName string            `json:"codec_name"`
+			Tags      map[string]string `json:"tags"`
+		} `json:"streams"`
+	}
+	if err := json.Unmarshal(output, &result); err != nil {
+		return nil, fmt.Errorf("parsing ffprobe: %w", err)
+	}
+
+	var streams []StreamMetadata
+	for _, s := range result.Streams {
+		lang := ""
+		title := ""
+		if s.Tags != nil {
+			lang = s.Tags["language"]
+			title = s.Tags["title"]
+		}
+		streams = append(streams, StreamMetadata{
+			Index:     s.Index,
+			CodecType: s.CodecType,
+			CodecName: s.CodecName,
+			Language:  lang,
+			Title:     title,
+		})
+	}
+
+	fmt.Printf("DEBUG stream languages: %+v\n", streams)
+	return streams, nil
+}
+
 func New() *Encoder {
 	return &Encoder{
 		ffmpegPath:  "ffmpeg",
@@ -131,7 +179,7 @@ func (e *Encoder) calculateZscaleWidth(path string, originalHeight int) (string,
 	return zscale, newWidth, nil
 }
 
-func (e *Encoder) Transcode(input string, job *types.Job, metadata *types.Metadata, interlaced bool) error {
+func (e *Encoder) Transcode(input string, job *types.Job, metadata *types.Metadata, interlaced bool, streamLangs []StreamMetadata) error {
 	dir := filepath.Dir(input)
 	audioWavs, err := e.extractAudio(input, dir)
 	if err != nil {
@@ -145,7 +193,7 @@ func (e *Encoder) Transcode(input string, job *types.Job, metadata *types.Metada
 	}
 	defer e.cleanupOpus(opusFiles)
 
-	if err := e.encodeVideo(input, opusFiles, job, metadata, interlaced); err != nil {
+	if err := e.encodeVideo(input, opusFiles, job, metadata, interlaced, streamLangs); err != nil {
 		return fmt.Errorf("encoding video: %w", err)
 	}
 
@@ -177,7 +225,7 @@ func (e *Encoder) encodeOpus(wavs []string, dir string) ([]string, error) {
 	return opusFiles, nil
 }
 
-func (e *Encoder) encodeVideo(input string, opusFiles []string, job *types.Job, metadata *types.Metadata, interlaced bool) error {
+func (e *Encoder) encodeVideo(input string, opusFiles []string, job *types.Job, metadata *types.Metadata, interlaced bool, streamLangs []StreamMetadata) error {
 	dir := filepath.Dir(input)
 	outFile := filepath.Join(dir, "output.mkv")
 
@@ -221,12 +269,30 @@ func (e *Encoder) encodeVideo(input string, opusFiles []string, job *types.Job,
 
 	args = append(args, "-c:a", "copy")
 	args = append(args, "-c:s", "copy")
-	args = append(args, "-map_metadata", "0")
+	args = append(args, "-map_metadata", "-1")
 
-	args = append(args, "-metadata", fmt.Sprintf("TITLE=%s", metadata.Title))
+	for _, stream := range streamLangs {
-	args = append(args, "-metadata", fmt.Sprintf("DATE_RELEASED=%s", metadata.DateReleased))
+		if stream.CodecType == "audio" && stream.Language != "" {
-	args = append(args, "-metadata", fmt.Sprintf("IMDBID=%s", metadata.IMDBID))
+			idx := 1
-	args = append(args, "-metadata", fmt.Sprintf("ORIGINAL_MEDIA_TYPE=%s", metadata.OriginalMedia))
+			for _, s := range streamLangs {
+				if s.CodecType == "audio" && s.Index < stream.Index {
+					idx++
+				}
+			}
+			args = append(args, fmt.Sprintf("-metadata:s:a:%d", idx-1), fmt.Sprintf("language=%s", stream.Language))
+			fmt.Printf("DEBUG setting audio language: stream %d -> language=%s\n", idx-1, stream.Language)
+		}
+		if stream.CodecType == "subtitle" && stream.Language != "" {
+			idx := 0
+			for _, s := range streamLangs {
+				if s.CodecType == "subtitle" && s.Index < stream.Index {
+					idx++
+				}
+			}
+			args = append(args, fmt.Sprintf("-metadata:s:s:%d", idx), fmt.Sprintf("language=%s", stream.Language))
+			fmt.Printf("DEBUG setting subtitle language: stream %d -> language=%s\n", idx, stream.Language)
+		}
+	}
 
 	if metadata.IsSeries {
 		args = append(args, "-metadata", fmt.Sprintf("COLLECTION=%s", metadata.Collection))