music-agregator/internal/tracker/generic_parser.go

package tracker

import (
	"path/filepath"
	"regexp"
	"strconv"
	"strings"

	"github.com/anacrolix/torrent/metainfo"
	"github.com/rs/zerolog/log"

	metadataPb "homelab.lan/music-agregator/gen/metadata/v1"
	"homelab.lan/music-agregator/internal/release"
)

var (
	bitratePattern = regexp.MustCompile(`(?i)(\d{2,3})\s*kbps`)
	hiResPatterns  = []*regexp.Regexp{
		regexp.MustCompile(`(?i)(\d{1,2})\s*[Bb]it\s*[-/]?\s*(\d{2,3}(?:\.\d)?)\s*[kK][Hh][Zz]`),
		regexp.MustCompile(`(?i)\[?\s*(?:FLAC|Flac)\s+(\d{1,2})\s*[-/]\s*(\d{2,3}(?:\.\d)?)\s*\]?`),
		regexp.MustCompile(`(?i)\[?\s*(\d{1,2})\s*[Bb]it\s*\]?`),
	}
	sourcePattern  = regexp.MustCompile(`(?i)\[(CD|WEB|Vinyl|LP|Cassette|MC|DVD|Blu-?Ray|SACD|DAT)\]`)
	ripTypePattern = regexp.MustCompile(`(?i)(vinyl\s*rip|SACD[- ]?R|HDCD|DSD\d*|tape\s*rip)`)
)

var audioExtensions = map[string]release.AudioFormat{
	".flac": release.FormatFLAC,
	".mp3":  release.FormatMP3,
	".aac":  release.FormatAAC,
	".m4a":  release.FormatAAC,
	".ape":  release.FormatAPE,
	".wv":   release.FormatWavPack,
	".alac": release.FormatALAC,
	".ogg":  release.FormatOGG,
	".wav":  release.FormatWAV,
}

type GenericParser struct{}

func NewGenericParser() *GenericParser {
	return &GenericParser{}
}

func (p *GenericParser) ParseTorrent(torrentData []byte, album *metadataPb.Album) *release.Release {
	r := &release.Release{}

	p.fillFromMetadata(r, album)
	p.fillFromTorrent(r, torrentData)

	r.ParsedSuccessfully = r.Artist != "" && r.Album != ""
	if !r.ParsedSuccessfully {
		r.ParseErrors = append(r.ParseErrors, "missing artist or album")
	}

	return r
}

func (p *GenericParser) Parse(title string) *release.Release {
	r := &release.Release{RawTitle: title}

	p.fillFromTitle(r, title)

	r.ParsedSuccessfully = r.Artist != "" && r.Album != ""
	if !r.ParsedSuccessfully {
		r.ParseErrors = append(r.ParseErrors, "missing artist or album")
	}

	return r
}

func (p *GenericParser) fillFromMetadata(r *release.Release, album *metadataPb.Album) {
	if album == nil {
		return
	}

	r.Album = album.GetTitle()

	if len(album.GetArtists()) > 0 {
		r.Artist = album.GetArtists()[0].GetArtist().GetName()
	}

	if album.GetReleaseDate() != "" {
		if year, err := strconv.Atoi(album.GetReleaseDate()[:4]); err == nil {
			r.Year = year
		}
	}

	switch strings.ToLower(album.GetAlbumType()) {
	case "album":
		r.Type = release.TypeAlbum
	case "ep":
		r.Type = release.TypeEP
	case "single":
		r.Type = release.TypeSingle
	case "compilation":
		r.Type = release.TypeCompilation
	case "soundtrack":
		r.Type = release.TypeSoundtrack
	case "live":
		r.Type = release.TypeLive
	}

	for _, g := range album.GetGenres() {
		r.Genres = append(r.Genres, g.GetName())
	}

	if album.GetLabel() != nil {
		r.Label = album.GetLabel().GetName()
	}

	r.TrackCount = int(album.GetTotalTracks())
	r.ReleaseCount = int(album.GetTotalDiscs())

	log.Trace().
		Str("artist", r.Artist).
		Str("album", r.Album).
		Int("year", r.Year).
		Str("type", r.Type.String()).
		Msg("filled from metadata")
}

func (p *GenericParser) fillFromTorrent(r *release.Release, torrentData []byte) {
	if len(torrentData) == 0 {
		return
	}

	mi, err := metainfo.Load(strings.NewReader(string(torrentData)))
	if err != nil {
		log.Error().Err(err).Msg("failed to parse torrent data")
		r.ParseErrors = append(r.ParseErrors, "failed to parse torrent: "+err.Error())
		return
	}

	info, err := mi.UnmarshalInfo()
	if err != nil {
		log.Error().Err(err).Msg("failed to unmarshal torrent info")
		r.ParseErrors = append(r.ParseErrors, "failed to unmarshal torrent info: "+err.Error())
		return
	}

	r.RawTitle = info.Name
	r.InfoHash = mi.HashInfoBytes().HexString()

	formatCounts := make(map[release.AudioFormat]int)
	formatSizes := make(map[release.AudioFormat]int64)

	if len(info.Files) == 0 {
		ext := strings.ToLower(filepath.Ext(info.Name))
		if fmt, ok := audioExtensions[ext]; ok {
			r.Format = fmt
			r.AudioFileCount = 1
			r.TotalAudioSize = info.Length
		}
	} else {
		for _, f := range info.Files {
			path := filepath.Join(f.Path...)
			ext := strings.ToLower(filepath.Ext(path))
			name := strings.TrimSuffix(filepath.Base(path), ext)

			if fmt, ok := audioExtensions[ext]; ok {
				formatCounts[fmt]++
				formatSizes[fmt] += f.Length
				r.TrackNames = append(r.TrackNames, cleanTrackName(name))
			}

			switch ext {
			case ".jpg", ".jpeg", ".png":
				r.HasCoverArt = true
			case ".cue":
				r.HasCueSheet = true
			case ".log":
				r.HasRipLog = true
			}
		}

		var dominantFormat release.AudioFormat
		var maxCount int
		for fmt, count := range formatCounts {
			if count > maxCount {
				maxCount = count
				dominantFormat = fmt
			}
		}
		r.Format = dominantFormat
		r.AudioFileCount = maxCount
		r.TotalAudioSize = formatSizes[dominantFormat]
	}

	if r.HasRipLog {
		r.Source = release.SourceCD
	}

	if r.TrackCount == 0 {
		r.TrackCount = r.AudioFileCount
	}

	p.fillFromTitle(r, info.Name)

	log.Trace().
		Str("hash", r.InfoHash).
		Str("format", r.Format.String()).
		Int("audio_files", r.AudioFileCount).
		Int64("audio_size", r.TotalAudioSize).
		Bool("cover", r.HasCoverArt).
		Bool("cue", r.HasCueSheet).
		Bool("log", r.HasRipLog).
		Msg("filled from torrent")
}

func (p *GenericParser) fillFromTitle(r *release.Release, title string) {
	if title == "" {
		return
	}

	if m := bitratePattern.FindStringSubmatch(title); len(m) > 1 {
		r.Bitrate = m[1] + " kbps"
	}

	for _, pattern := range hiResPatterns {
		m := pattern.FindStringSubmatch(title)
		if len(m) < 2 {
			continue
		}
		if r.BitDepth == 0 {
			if bd, err := strconv.Atoi(m[1]); err == nil {
				r.BitDepth = bd
			}
		}
		if len(m) > 2 && r.SampleRate == 0 {
			if sr, err := strconv.ParseFloat(m[2], 64); err == nil {
				r.SampleRate = int(sr * 1000)
			}
		}
		if r.BitDepth > 0 {
			break
		}
	}

	if m := sourcePattern.FindStringSubmatch(title); len(m) > 1 && r.Source == release.SourceUnknown {
		switch strings.ToUpper(m[1]) {
		case "CD":
			r.Source = release.SourceCD
		case "WEB":
			r.Source = release.SourceWEB
		case "VINYL", "LP":
			r.Source = release.SourceVinyl
		case "CASSETTE", "MC":
			r.Source = release.SourceCassette
		case "DVD":
			r.Source = release.SourceDVD
		case "BLU-RAY", "BLURAY":
			r.Source = release.SourceBluRay
		}
	}

	if m := ripTypePattern.FindStringSubmatch(title); len(m) > 1 {
		r.RipType = m[1]
	}

	log.Trace().
		Str("bitrate", r.Bitrate).
		Int("bit_depth", r.BitDepth).
		Int("sample_rate", r.SampleRate).
		Str("source", r.Source.String()).
		Str("rip_type", r.RipType).
		Msg("filled from title")
}

var trackNumberPrefix = regexp.MustCompile(`^\d{1,3}[\s.\-]+`)

func cleanTrackName(name string) string {
	cleaned := trackNumberPrefix.ReplaceAllString(name, "")
	if cleaned == "" {
		return name
	}
	return strings.TrimSpace(cleaned)
}