Deduce bit depth, sample rate, and bitrate from file sizes; add comprehensive parser tests

2026-05-09 22:09:17 +02:00
parent 7fa859e815
commit 1e8506f146
2 changed files with 542 additions and 68 deletions
@@ -196,6 +196,7 @@ func (p *GenericParser) fillFromTorrent(r *release.Release, torrentData []byte)
 	}

 	p.fillFromTitle(r, info.Name)
+	p.deduceFromFileSize(r)

 	log.Trace().
 		Str("hash", r.InfoHash).
@@ -205,6 +206,9 @@ func (p *GenericParser) fillFromTorrent(r *release.Release, torrentData []byte)
 		Bool("cover", r.HasCoverArt).
 		Bool("cue", r.HasCueSheet).
 		Bool("log", r.HasRipLog).
+		Int("bit_depth", r.BitDepth).
+		Int("sample_rate", r.SampleRate).
+		Str("bitrate", r.Bitrate).
 		Msg("filled from torrent")
 }

@@ -267,6 +271,75 @@ func (p *GenericParser) fillFromTitle(r *release.Release, title string) {
 		Msg("filled from title")
 }

+func (p *GenericParser) deduceFromFileSize(r *release.Release) {
+	if r.AudioFileCount == 0 || r.TotalAudioSize == 0 {
+		return
+	}
+
+	avgFileSize := r.TotalAudioSize / int64(r.AudioFileCount)
+	avgFileSizeMB := float64(avgFileSize) / (1024 * 1024)
+
+	switch {
+	case r.Format.IsLossless():
+		if r.BitDepth > 0 && r.SampleRate > 0 {
+			return
+		}
+
+		// Average FLAC file size per ~4 min track:
+		//   16/44.1 ≈ 25-35 MB    24/48 ≈ 40-60 MB
+		//   24/96   ≈ 80-120 MB   24/192 ≈ 160-240 MB
+		switch {
+		case avgFileSizeMB >= 130:
+			p.setIfMissing(r, 24, 192000)
+		case avgFileSizeMB >= 65:
+			p.setIfMissing(r, 24, 96000)
+		case avgFileSizeMB >= 38:
+			p.setIfMissing(r, 24, 48000)
+		default:
+			p.setIfMissing(r, 16, 44100)
+		}
+
+		log.Trace().
+			Float64("avg_file_mb", avgFileSizeMB).
+			Int("deduced_bit_depth", r.BitDepth).
+			Int("deduced_sample_rate", r.SampleRate).
+			Msg("deduced lossless quality from file size")
+
+	case r.Format == release.FormatMP3:
+		if r.Bitrate != "" {
+			return
+		}
+
+		// Average MP3 file size per ~4 min track:
+		//   128 kbps ≈ 3.5-4 MB   192 kbps ≈ 5-6 MB
+		//   256 kbps ≈ 7-8 MB     320 kbps ≈ 9-10 MB
+		switch {
+		case avgFileSizeMB >= 8.5:
+			r.Bitrate = "320 kbps"
+		case avgFileSizeMB >= 6.5:
+			r.Bitrate = "256 kbps"
+		case avgFileSizeMB >= 4.5:
+			r.Bitrate = "192 kbps"
+		default:
+			r.Bitrate = "128 kbps"
+		}
+
+		log.Trace().
+			Float64("avg_file_mb", avgFileSizeMB).
+			Str("deduced_bitrate", r.Bitrate).
+			Msg("deduced mp3 bitrate from file size")
+	}
+}
+
+func (p *GenericParser) setIfMissing(r *release.Release, bitDepth int, sampleRate int) {
+	if r.BitDepth == 0 {
+		r.BitDepth = bitDepth
+	}
+	if r.SampleRate == 0 {
+		r.SampleRate = sampleRate
+	}
+}
+
 var trackNumberPrefix = regexp.MustCompile(`^\d{1,3}[\s.\-]+`)

 func cleanTrackName(name string) string {