fix(scanner): refactor a bit and fix the issue of repeatedly adding and removing tracks 😎

fixes #26
fixes #63
This commit is contained in:
sentriz
2021-09-09 23:19:02 +01:00
parent 569f62c08c
commit 93608d04b4
2 changed files with 140 additions and 98 deletions

View File

@@ -62,20 +62,20 @@ type Scanner struct {
musicPath string musicPath string
isFull bool isFull bool
genreSplit string genreSplit string
// these two are for the transaction we do for every folder. // these two are for the transaction we do for every album.
// the boolean is there so we dont begin or commit multiple // the boolean is there so we dont begin or commit multiple
// times in the handle folder or post children callback // times in the handle album or post children callback
trTx *db.DB trTx *db.DB
trTxOpen bool trTxOpen bool
// these two are for keeping state between noted in the tree. // these two are for keeping state between noted in the tree.
// eg. keep track of a parents folder or the path to a cover // eg. keep track of a parents album or the path to a cover
// we just saw that we need to commit in the post children // we just saw that we need to commit in the post children
// callback // callback
curFolders *stack.Stack curAlbums *stack.Stack
curCover string curCover string
// then the rest are for stats and cleanup at the very end // then the rest are for stats and cleanup at the very end
seenTracks map[int]struct{} // set of p keys seenTracks map[int]struct{} // set of p keys
seenFolders map[int]struct{} // set of p keys seenAlbums map[int]struct{} // set of p keys
seenTracksNew int // n tracks not seen before seenTracksNew int // n tracks not seen before
} }
@@ -117,7 +117,7 @@ func (s *Scanner) cleanTracks() error {
return nil return nil
} }
func (s *Scanner) cleanFolders() error { func (s *Scanner) cleanAlbums() error {
start := time.Now() start := time.Now()
var previous []int var previous []int
var missing []int64 var missing []int64
@@ -129,7 +129,7 @@ func (s *Scanner) cleanFolders() error {
return fmt.Errorf("plucking ids: %w", err) return fmt.Errorf("plucking ids: %w", err)
} }
for _, prev := range previous { for _, prev := range previous {
if _, ok := s.seenFolders[prev]; !ok { if _, ok := s.seenAlbums[prev]; !ok {
missing = append(missing, int64(prev)) missing = append(missing, int64(prev))
} }
} }
@@ -139,7 +139,7 @@ func (s *Scanner) cleanFolders() error {
if err != nil { if err != nil {
return err return err
} }
log.Printf("finished clean folders in %s, %d removed", durSince(start), len(missing)) log.Printf("finished clean albums in %s, %d removed", durSince(start), len(missing))
return nil return nil
} }
@@ -203,8 +203,8 @@ func (s *Scanner) Start(opts ScanOptions) error {
// reset state vars for the new scan // reset state vars for the new scan
s.isFull = opts.IsFull s.isFull = opts.IsFull
s.seenTracks = map[int]struct{}{} s.seenTracks = map[int]struct{}{}
s.seenFolders = map[int]struct{}{} s.seenAlbums = map[int]struct{}{}
s.curFolders = &stack.Stack{} s.curAlbums = &stack.Stack{}
s.seenTracksNew = 0 s.seenTracksNew = 0
// begin walking // begin walking
@@ -235,8 +235,8 @@ func (s *Scanner) Start(opts ScanOptions) error {
if err := s.cleanTracks(); err != nil { if err := s.cleanTracks(); err != nil {
return fmt.Errorf("clean tracks: %w", err) return fmt.Errorf("clean tracks: %w", err)
} }
if err := s.cleanFolders(); err != nil { if err := s.cleanAlbums(); err != nil {
return fmt.Errorf("clean folders: %w", err) return fmt.Errorf("clean albums: %w", err)
} }
if err := s.cleanArtists(); err != nil { if err := s.cleanArtists(); err != nil {
return fmt.Errorf("clean artists: %w", err) return fmt.Errorf("clean artists: %w", err)
@@ -309,7 +309,7 @@ func (s *Scanner) callbackItem(fullPath string, info *godirwalk.Dirent) error {
return fmt.Errorf("stating link to dir: %w", err) return fmt.Errorf("stating link to dir: %w", err)
} }
if isDir { if isDir {
return s.handleFolder(it) return s.handleAlbum(it)
} }
if isCover(filename) { if isCover(filename) {
s.curCover = filename s.curCover = filename
@@ -333,20 +333,20 @@ func (s *Scanner) callbackPost(fullPath string, info *godirwalk.Dirent) error {
s.trTx.Commit() s.trTx.Commit()
s.trTxOpen = false s.trTxOpen = false
} }
// begin taking the current folder off the stack and add it's // begin taking the current album off the stack and add it's
// parent, cover that we found, etc. // parent, cover that we found, etc.
folder := s.curFolders.Pop() album := s.curAlbums.Pop()
if folder.ParentID != 0 { if album.ParentID != 0 {
return nil return nil
} }
folder.ParentID = s.curFolders.PeekID() album.ParentID = s.curAlbums.PeekID()
folder.Cover = s.curCover album.Cover = s.curCover
if err := s.db.Save(folder).Error; err != nil { if err := s.db.Save(album).Error; err != nil {
return fmt.Errorf("writing albums table: %w", err) return fmt.Errorf("writing albums table: %w", err)
} }
// we only log changed folders // we only log changed albums
log.Printf("processed folder `%s`\n", log.Printf("processed folder `%s`\n",
path.Join(folder.LeftPath, folder.RightPath)) path.Join(album.LeftPath, album.RightPath))
return nil return nil
} }
@@ -361,38 +361,38 @@ func (s *Scanner) itemUnchanged(statModTime, updatedInDB time.Time) bool {
return statModTime.Before(updatedInDB) return statModTime.Before(updatedInDB)
} }
func (s *Scanner) handleFolder(it *item) error { func (s *Scanner) handleAlbum(it *item) error {
if s.trTxOpen { if s.trTxOpen {
// a transaction still being open when we handle a folder can // a transaction still being open when we handle an album can
// happen if there is a folder that contains /both/ tracks and // happen if there is a album that contains /both/ tracks and
// sub folders // sub albums
s.trTx.Commit() s.trTx.Commit()
s.trTxOpen = false s.trTxOpen = false
} }
folder := &db.Album{} album := &db.Album{}
defer func() { defer func() {
// folder's id will come from early return // album's id will come from early return
// or save at the end // or save at the end
s.seenFolders[folder.ID] = struct{}{} s.seenAlbums[album.ID] = struct{}{}
s.curFolders.Push(folder) s.curAlbums.Push(album)
}() }()
err := s.db. err := s.db.
Where(db.Album{ Where(db.Album{
LeftPath: it.directory, LeftPath: it.directory,
RightPath: it.filename, RightPath: it.filename,
}). }).
First(folder). First(album).
Error Error
if !gorm.IsRecordNotFoundError(err) && if !gorm.IsRecordNotFoundError(err) &&
s.itemUnchanged(it.stat.ModTime(), folder.UpdatedAt) { s.itemUnchanged(it.stat.ModTime(), album.UpdatedAt) {
// we found the record but it hasn't changed // we found the record but it hasn't changed
return nil return nil
} }
folder.LeftPath = it.directory album.LeftPath = it.directory
folder.RightPath = it.filename album.RightPath = it.filename
folder.RightPathUDec = decoded(it.filename) album.RightPathUDec = decoded(it.filename)
folder.ModifiedAt = it.stat.ModTime() album.ModifiedAt = it.stat.ModTime()
if err := s.db.Save(folder).Error; err != nil { if err := s.db.Save(album).Error; err != nil {
return fmt.Errorf("writing albums table: %w", err) return fmt.Errorf("writing albums table: %w", err)
} }
return nil return nil
@@ -404,78 +404,127 @@ func (s *Scanner) handleTrack(it *item) error {
s.trTxOpen = true s.trTxOpen = true
} }
// set track basics // init empty track and mark its ID (from lookup or save)
track := &db.Track{} // for later cleanup later
var track db.Track
defer func() { defer func() {
// folder's id will come from early return
// or save at the end
s.seenTracks[track.ID] = struct{}{} s.seenTracks[track.ID] = struct{}{}
}() }()
album := s.curAlbums.Peek()
err := s.trTx. err := s.trTx.
Select("id, updated_at"). Select("id, updated_at").
Where(db.Track{ Where(db.Track{
AlbumID: s.curFolders.PeekID(), AlbumID: album.ID,
Filename: it.filename, Filename: it.filename,
}). }).
First(track). First(&track).
Error Error
if !gorm.IsRecordNotFoundError(err) && if !gorm.IsRecordNotFoundError(err) &&
s.itemUnchanged(it.stat.ModTime(), track.UpdatedAt) { s.itemUnchanged(it.stat.ModTime(), track.UpdatedAt) {
// we found the record but it hasn't changed // we found the record but it hasn't changed
return nil return nil
} }
track.Filename = it.filename
track.FilenameUDec = decoded(it.filename) trags, err := tags.New(it.fullPath)
track.Size = int(it.stat.Size())
track.AlbumID = s.curFolders.PeekID()
trTags, err := tags.New(it.fullPath)
if err != nil { if err != nil {
return ErrReadingTags return ErrReadingTags
} }
track.TagTitle = trTags.Title()
track.TagTitleUDec = decoded(trTags.Title())
track.TagTrackArtist = trTags.Artist()
track.TagTrackNumber = trTags.TrackNumber()
track.TagDiscNumber = trTags.DiscNumber()
track.TagBrainzID = trTags.BrainzID()
track.Length = trTags.Length() // these two should be calculated
track.Bitrate = trTags.Bitrate() // ...from the file instead of tags
// set album artist basics genreIDs, err := s.populateGenres(&track, trags)
artistName := firstTag("Unknown Artist", trTags.AlbumArtist, trTags.Artist) if err != nil {
artist := &db.Artist{} return fmt.Errorf("populate genres: %w", err)
s.trTx. }
// create album and album artist records for first track in album
if album.TagTitle == "" {
albumArtist, err := s.populateAlbumArtist(trags)
if err != nil {
return fmt.Errorf("populate artist: %w", err)
}
albumName := trags.SomeAlbum()
album.TagTitle = albumName
album.TagTitleUDec = decoded(albumName)
album.TagBrainzID = trags.AlbumBrainzID()
album.TagYear = trags.Year()
album.TagArtistID = albumArtist.ID
if err := s.populateAlbumGenres(album, genreIDs); err != nil {
return fmt.Errorf("populate album genres: %w", err)
}
}
track.Filename = it.filename
track.FilenameUDec = decoded(it.filename)
track.Size = int(it.stat.Size())
track.AlbumID = album.ID
track.ArtistID = album.TagArtistID
track.TagTitle = trags.Title()
track.TagTitleUDec = decoded(trags.Title())
track.TagTrackArtist = trags.Artist()
track.TagTrackNumber = trags.TrackNumber()
track.TagDiscNumber = trags.DiscNumber()
track.TagBrainzID = trags.BrainzID()
track.Length = trags.Length() // these two should be calculated
track.Bitrate = trags.Bitrate() // ...from the file instead of tags
if err := s.trTx.Save(&track).Error; err != nil {
return fmt.Errorf("writing track table: %w", err)
}
s.seenTracksNew++
if err := s.populateTrackGenres(&track, genreIDs); err != nil {
return fmt.Errorf("populating track genres : %w", err)
}
return nil
}
func (s *Scanner) populateAlbumArtist(trags *tags.Tags) (*db.Artist, error) {
var artist db.Artist
artistName := trags.SomeAlbumArtist()
err := s.trTx.
Where("name=?", artistName). Where("name=?", artistName).
Assign(db.Artist{ Assign(db.Artist{
Name: artistName, Name: artistName,
NameUDec: decoded(artistName), NameUDec: decoded(artistName),
}). }).
FirstOrCreate(artist) FirstOrCreate(&artist).
track.ArtistID = artist.ID Error
if err != nil {
return nil, fmt.Errorf("find or create artist: %w", err)
}
return &artist, nil
}
// set genre func (s *Scanner) populateGenres(track *db.Track, trags *tags.Tags) ([]int, error) {
genreTag := firstTag("Unknown Genre", trTags.Genre) var genreIDs []int
genreNames := strings.Split(genreTag, s.genreSplit) genreNames := strings.Split(trags.SomeGenre(), s.genreSplit)
genreIDs := []int{}
for _, genreName := range genreNames { for _, genreName := range genreNames {
genre := &db.Genre{} genre := &db.Genre{}
s.trTx.FirstOrCreate(genre, db.Genre{ q := s.trTx.FirstOrCreate(genre, db.Genre{
Name: genreName, Name: genreName,
}) })
if err := q.Error; err != nil {
return nil, err
}
genreIDs = append(genreIDs, genre.ID) genreIDs = append(genreIDs, genre.ID)
} }
return genreIDs, nil
}
if err := s.trTx.Save(track).Error; err != nil { func (s *Scanner) populateTrackGenres(track *db.Track, genreIDs []int) error {
return fmt.Errorf("writing track table: %w", err) err := s.trTx.
}
err = s.trTx.
Where("track_id=?", track.ID). Where("track_id=?", track.ID).
Delete(db.TrackGenre{}). Delete(db.TrackGenre{}).
Error Error
if err != nil { if err != nil {
return fmt.Errorf("delete old track genre records: %w", err) return fmt.Errorf("delete old track genre records: %w", err)
} }
err = s.trTx.InsertBulkLeftMany( err = s.trTx.InsertBulkLeftMany(
"track_genres", "track_genres",
[]string{"track_id", "genre_id"}, []string{"track_id", "genre_id"},
@@ -485,15 +534,12 @@ func (s *Scanner) handleTrack(it *item) error {
if err != nil { if err != nil {
return fmt.Errorf("insert bulk track genres: %w", err) return fmt.Errorf("insert bulk track genres: %w", err)
} }
s.seenTracksNew++
// set album if this is the first track in the folder
folder := s.curFolders.Peek()
if folder.TagTitle != "" {
return nil return nil
} }
err = s.trTx.
Where("album_id=?", folder.ID). func (s *Scanner) populateAlbumGenres(album *db.Album, genreIDs []int) error {
err := s.trTx.
Where("album_id=?", album.ID).
Delete(db.AlbumGenre{}). Delete(db.AlbumGenre{}).
Error Error
if err != nil { if err != nil {
@@ -502,25 +548,11 @@ func (s *Scanner) handleTrack(it *item) error {
err = s.trTx.InsertBulkLeftMany( err = s.trTx.InsertBulkLeftMany(
"album_genres", "album_genres",
[]string{"album_id", "genre_id"}, []string{"album_id", "genre_id"},
folder.ID, album.ID,
genreIDs, genreIDs,
) )
if err != nil { if err != nil {
return fmt.Errorf("insert bulk album genres: %w", err) return fmt.Errorf("insert bulk album genres: %w", err)
} }
folder.TagTitle = trTags.Album()
folder.TagTitleUDec = decoded(trTags.Album())
folder.TagBrainzID = trTags.AlbumBrainzID()
folder.TagYear = trTags.Year()
folder.TagArtistID = artist.ID
return nil return nil
} }
func firstTag(fallback string, tags ...func() string) string {
for _, f := range tags {
if tag := f(); tag != "" {
return tag
}
}
return fallback
}

View File

@@ -49,8 +49,18 @@ func (t *Tags) TrackNumber() int { return intSep(t.firstTag("tracknumber"),
func (t *Tags) DiscNumber() int { return intSep(t.firstTag("discnumber"), "/") } // eg. 1/2 func (t *Tags) DiscNumber() int { return intSep(t.firstTag("discnumber"), "/") } // eg. 1/2
func (t *Tags) Length() int { return t.props.Length } func (t *Tags) Length() int { return t.props.Length }
func (t *Tags) Bitrate() int { return t.props.Bitrate } func (t *Tags) Bitrate() int { return t.props.Bitrate }
func (t *Tags) Year() int { return intSep(t.firstTag("originaldate", "date", "year"), "-") }
func (t *Tags) Year() int { func (t *Tags) SomeAlbum() string { return first("Unknown Album", t.Album()) }
// eg. 2019-6-11 func (t *Tags) SomeArtist() string { return first("Unknown Artist", t.Artist()) }
return intSep(t.firstTag("originaldate", "date", "year"), "-") func (t *Tags) SomeAlbumArtist() string { return first("Unknown Artist", t.AlbumArtist(), t.Artist()) }
func (t *Tags) SomeGenre() string { return first("Unknown Genre", t.Genre()) }
func first(or string, strs ...string) string {
for _, str := range strs {
if str != "" {
return str
}
}
return or
} }