def __get_lastfm_tags(self, release_title: str, release_artists: List[str]): flattened_artist = flatten_artists(release_artists) lastfm_release = self.lastfm.get_release(flattened_artist, release_title) lastfm_tags = [ x for x in tag_filter_all(lastfm_release.tags, release_artists + [release_title], True) ] if not lastfm_tags: filtered_artists = [ x for x in release_artists if x.lower() not in ["various artist", "various artists", "va"] ] artists_tags = [ self.lastfm.get_artist(artist).tags for artist in filtered_artists ] weighted_tags = OrderedDict() for artist_tags in artists_tags: for tag in artist_tags: if tag not in weighted_tags or ( tag in weighted_tags and weighted_tags[tag] < artist_tags[tag]): weighted_tags[tag] = artist_tags[tag] lastfm_tags = [ x for x in tag_filter_all(weighted_tags, filtered_artists + [release_title], True) ] return [x for x in lastfm_tags]
def get_folder_name(self, codec_short: bool = True, group_by_category: bool = False): assert self.validate_release_date(), "Release date validation failed" assert self.validate_release_title(), "Release title validation failed" assert self.validate_release_artists(), "Release artists validation failed" assert self.validate_codec(), "Codec validation failed" track1 = self.tracks[next(iter(self.tracks))] # clean release name, and category release_name, _ = get_category_fix_name(self) release_artist = flatten_artists(track1.release_artists) year = track1.date.split("-")[0] release_category_str = "[{0}] ".format(self.category.value) if self.category != ReleaseCategory.ALBUM else "" release_source_str = "[{0}] ".format(self.source.value) if self.source != ReleaseSource.CD else "" title_first_categories = {ReleaseCategory.COMPILATION, ReleaseCategory.MIX, ReleaseCategory.MIXTAPE, ReleaseCategory.VIDEO_GAME_MUSIC, ReleaseCategory.SOUNDTRACK} # folder name if group_by_category is False and self.category in title_first_categories: return normalize_path_chars( "VA - {release_name} - {year} - {release_artist} " "{release_category_str}{release_source_str}[{release_codec}]" .format(release_name=release_name, year=year, release_artist=release_artist, release_category_str=release_category_str, release_codec=self.get_release_codec_setting(codec_short), release_source_str=release_source_str)) elif self.category in title_first_categories: return normalize_path_chars( "{release_name} - {year} - {release_artist} " "{release_category_str}{release_source_str}[{release_codec}]" .format(release_name=release_name, year=year, release_artist=release_artist, release_category_str=release_category_str, release_codec=self.get_release_codec_setting(codec_short), release_source_str=release_source_str)) else: """self.category in {ReleaseCategory.ALBUM, ReleaseCategory.ANTHOLOGY, ReleaseCategory.BOOTLEG, ReleaseCategory.CONCERT_RECORDING, ReleaseCategory.DEMO, ReleaseCategory.EP, ReleaseCategory.INTERVIEW, ReleaseCategory.LIVE_ALBUM, ReleaseCategory.REMIX, ReleaseCategory.SINGLE, ReleaseCategory.UNKNOWN}:""" return normalize_path_chars( "{release_artist} - {year} - {release_name} " "{release_category_str}{release_source_str}[{release_codec}]" .format(release_artist=release_artist, year=year, release_name=release_name, release_category_str=release_category_str, release_source_str=release_source_str, release_codec=self.get_release_codec_setting(codec_short)))
def __lastfm_fixes(self, release: Release, release_artists: List[str]) -> None: # lastfm fixes release_title = release.validate_release_title() if not self.lastfm or not len(release_artists) or not release_title: return # extract (edition info) from release titles release_title, release_edition = split_release_title( normalize_release_title(release_title)) flattened_artist = flatten_artists(release_artists) lastfm_release = None while True: try: lastfm_release = self.lastfm.get_release( flattened_artist, release_title) break except LastfmCache.ReleaseNotFoundError as e: logging.getLogger(__name__).error(e) break except LastfmCache.UpgradeRequiredError: logging.getLogger(__name__).error(upgrade_message) exit(1) except LastfmCache.ConnectionError: logging.getLogger(__name__).error( "Connection error while retrieving release, retrying...") time.sleep(1) except LastfmCache.LastfmCacheError: logging.getLogger(__name__).error( "Server error while retrieving release, retrying...") time.sleep(1) self.__lastfm_release_fixes(release, lastfm_release, release_artists, release_title, release_edition) # fix track artists using lastfm self.__lastfm_fix_track_artists(release) # fix release artists using lastfm self.__lastfm_fix_release_artists(release)
def move_rename_folder(release: Release, unique_releases: Set[Tuple], curr_dir: str, dest_folder: str, duplicate_folder: str, args: argparse.Namespace) -> str: """Rename a release folder, and move to a destination folder""" # if a dry run,or the folder name cannot be validated, do nothing if args.dry_run or not release.can_validate_folder_name(): return curr_dir moved_dir = curr_dir # rename the release folder codec_short = not args.full_codec_names fixed_dir = os.path.join( os.path.split(curr_dir)[0], release.get_folder_name(codec_short=codec_short, group_by_category=args.group_by_category)) if curr_dir != fixed_dir: if not os.path.exists(fixed_dir) or os.path.normcase( curr_dir) == os.path.normcase(fixed_dir): while True: try: os.rename(curr_dir, fixed_dir) break except PermissionError: logging.getLogger(__name__).error( "PermissionError: could not rename directory to {0}". format(fixed_dir)) time.sleep(1) moved_dir = fixed_dir else: logging.getLogger(__name__).error( "Release folder already exists: {0}".format(fixed_dir)) # move the release folder to a destination moved_duplicate = False if dest_folder and release.num_violations == 0: artist_folder = flatten_artists(release.validate_release_artists()) \ if args.group_by_artist and not release.is_va() else "" category_folder = str( release.category.value) if args.group_by_category else "" curr_dest_parent_folder = os.path.join(dest_folder, category_folder, artist_folder) curr_dest_folder = os.path.join( curr_dest_parent_folder, release.get_folder_name(codec_short=codec_short, group_by_category=args.group_by_category)) if os.path.normcase(moved_dir) != os.path.normcase(curr_dest_folder): if not os.path.exists(curr_dest_parent_folder): os.makedirs(curr_dest_parent_folder, exist_ok=True) if not os.path.exists(curr_dest_folder): os.rename(moved_dir, curr_dest_folder) moved_dir = curr_dest_folder # clean up empty directories curr_src_parent_folder = os.path.split(fixed_dir)[0] while not os.listdir(curr_src_parent_folder): os.rmdir(curr_src_parent_folder) curr_src_parent_folder = os.path.split( curr_src_parent_folder)[0] else: if duplicate_folder: release_folder_name = release.get_folder_name( codec_short=codec_short, group_by_category=args.group_by_category) moved_dir = move_duplicate(duplicate_folder, moved_dir, release_folder_name) moved_duplicate = True else: logging.getLogger(__name__).error( "Destination folder already exists: {0}".format( fixed_dir)) # deduplicate versions of the same release unique_release = UniqueRelease( release.validate_release_artists(), release.validate_release_date().split("-")[0], release.validate_release_title(), release.validate_codec(), release.get_codec_rank(), moved_dir) if duplicate_folder and release.num_violations == 0 and not moved_duplicate: if unique_release in unique_releases: existing = [x for x in unique_releases if x == unique_release][0] if unique_release > existing: # move the existing one release_folder_name = os.path.split(existing.path)[1] moved_dir = move_duplicate(duplicate_folder, existing.path, release_folder_name) unique_releases.remove(unique_release) unique_releases.add(unique_release) else: # move the current one release_folder_name = release.get_folder_name( codec_short=codec_short, group_by_category=args.group_by_category) moved_dir = move_duplicate(duplicate_folder, moved_dir, release_folder_name) else: unique_releases.add(unique_release) return moved_dir
def validate(self, release: Release) -> List[Violation]: violations = OrderedSet() # leading/trailing whitespace for filename, track in release.tracks.items(): if track.artists != track.strip_whitespace_artists(): violations.add( Violation( ViolationType.ARTIST_WHITESPACE, "File '{0}' has leading/trailing whitespace in its Artist(s)" .format(filename))) for filename, track in release.tracks.items(): if track.release_artists != track.strip_whitespace_release_artists( ): violations.add( Violation( ViolationType.RELEASE_ARTIST_WHITESPACE, "File '{0}' has leading/trailing whitespace in its Album/Release Artist(s)" .format(filename))) for filename, track in release.tracks.items(): if track.date != track.strip_whitespace_date(): violations.add( Violation( ViolationType.DATE_WHITESPACE, "File '{0}' has leading/trailing whitespace in its Year/Date" .format(filename))) for filename, track in release.tracks.items(): if track.release_title != track.strip_whitespace_release_title(): violations.add( Violation( ViolationType.RELEASE_TITLE_WHITESPACE, "File '{0}' has leading/trailing whitespace in its Album/Release Title" .format(filename))) for filename, track in release.tracks.items(): if track.track_title != track.strip_whitespace_track_title(): violations.add( Violation( ViolationType.TRACK_TITLE_WHITESPACE, "File '{0}' has leading/trailing whitespace in its Track Title" .format(filename))) for filename, track in release.tracks.items(): if track.genres != track.strip_whitespace_genres(): violations.add( Violation( ViolationType.GENRE_WHITESPACE, "File '{0}' has leading/trailing whitespace in its Genre(s)" .format(filename))) # release date if not release.validate_release_date(): violations.add( Violation( ViolationType.DATE_INCONSISTENT, "Release contains blank or inconsistent 'Date' tags")) # artists if release.blank_artists(): violations.add( Violation( ViolationType.ARTIST_BLANK, "Release contains {0} tracks with missing 'Artist' tags". format(release.blank_artists()))) # track titles if release.blank_track_titles(): violations.add( Violation( ViolationType.TRACK_TITLE_BLANK, "Release contains {0} tracks with missing 'Track Title' tags" .format(release.blank_track_titles()))) # release artist release_artists = release.validate_release_artists() if not release_artists: violations.add( Violation( ViolationType.RELEASE_ARTIST_INCONSISTENT, "Release contains blank or inconsistent 'Album/Release Artist' tags" )) # if the lastfmcache is present, validate the release artist validated_release_artists = release_artists if self.lastfm and len(release_artists) == 1: validated_release_artists = [] for artist in release_artists: try: validated_release_artist = self.lastfm.get_artist( artist.strip()).artist_name if validated_release_artist != artist: violations.add( Violation( ViolationType.RELEASE_ARTIST_SPELLING, "Incorrectly spelled Album/Release Artist '{0}' (should be '{1}')" .format(artist, validated_release_artist))) validated_release_artists.append(validated_release_artist) except LastfmCache.ArtistNotFoundError: violations.add( Violation( ViolationType.ARTIST_LOOKUP, "Lookup failed of release artist '{release_artist}'" .format(release_artist=artist.strip()))) # release title release_title = release.validate_release_title() if not release_title: violations.add( Violation( ViolationType.RELEASE_TITLE_INCONSISTENT, "Release contains blank or inconsistent 'Album/Release Title' tags" )) bracket_pairs = [["[", "]"], ["(", ")"], ["{", "}"]] if release_title: # check if "[Source]" is contained in the release title for source in ReleaseSource: for brackets in bracket_pairs: curr_source = "{0}{1}{2}".format(brackets[0], source.value, brackets[1]) if curr_source.lower() in release_title.lower(): violations.add( Violation( ViolationType.RELEASE_TITLE_SOURCE, "Release title contains source {0}".format( curr_source))) # check if the release title ends with a space and a source name, without brackets for source in [x for x in ReleaseSource]: if release_title.lower().endswith(" {0}".format( source.value.lower())): violations.add( Violation( ViolationType.RELEASE_TITLE_SOURCE, "Release title ends with source {0}".format( source.value))) # check if "[Category]" is contained in the release title for category in ReleaseCategory: for brackets in bracket_pairs: curr_category = "{0}{1}{2}".format(brackets[0], category.value, brackets[1]) if curr_category.lower() in release_title.lower(): violations.add( Violation( ViolationType.RELEASE_TITLE_CATEGORY, "Release title contains category {0}".format( curr_category))) # check if the release title ends with a space and a category name, without brackets (except Album) for category in [ x for x in ReleaseCategory if x is not ReleaseCategory.ALBUM ]: if release_title.lower().endswith(" {0}".format( category.value.lower())): violations.add( Violation( ViolationType.RELEASE_TITLE_CATEGORY, "Release title ends with category {0}".format( category.value))) # lastfm artist validations if self.lastfm and release_title and len(validated_release_artists): # extract (edition info) from release titles release_title, _ = split_release_title( normalize_release_title(release_title)) flattened_artist = flatten_artists(validated_release_artists) lastfm_release = None try: lastfm_release = self.lastfm.get_release( flattened_artist, release_title) except LastfmCache.ReleaseNotFoundError as e: logging.getLogger(__name__).error(e) if lastfm_release: # release title if lastfm_release.release_name != release_title and \ ReleaseValidator.__lastfm_can_fix_release_title(release_title, lastfm_release.release_name): violations.add( Violation( ViolationType.RELEASE_TITLE_SPELLING, "Incorrectly spelled Album/Release name '{0}' (should be '{1}')" .format(release_title, lastfm_release.release_name))) # dates if lastfm_release.release_date: date = next(iter(release.tracks.values())).date if lastfm_release.release_date != date and \ (not date or len(lastfm_release.release_date) >= len(date)): violations.add( Violation( ViolationType.DATE_INCORRECT, "Incorrect Release Date '{0}' (should be '{1}')" .format(date, lastfm_release.release_date))) # tags/genres (only fail if 0-1 genres - i.e. lastfm tags have never been applied) release_genres = release.validate_genres() lastfm_tags = self.__get_lastfm_tags( release_title, validated_release_artists) if len(release_genres) < 2 <= len(lastfm_tags): violations.add( Violation( ViolationType.BAD_GENRES, "Bad release genres: [{0}] (should be [{1}])". format(", ".join(release_genres), ", ".join(lastfm_tags)))) # match and validate track titles (intersection only) if self.lastfm_track_title_validation: for track in release.tracks.values(): if track.track_number in lastfm_release.tracks: lastfm_title = normalize_track_title( lastfm_release.tracks[ track.track_number].track_name) if not track.track_title or track.track_title.lower( ) != lastfm_title.lower(): violations.add( Violation( ViolationType.INCORRECT_TRACK_TITLE, "Incorrect track title '{0}' should be: '{1}'" .format(track.track_title, lastfm_title))) # track artists for track in release.tracks.values(): for artist in track.artists: while True: try: validated_artist = self.lastfm.get_artist( normalize_artist_name(artist)).artist_name if validated_artist != artist: violations.add( Violation( ViolationType.TRACK_ARTIST_SPELLING, "Incorrectly spelled Track Artist '{0}' (should be '{1}')" .format(artist, validated_artist))) break except LastfmCache.ArtistNotFoundError: # as e: # violations.add(str(e)) break except LastfmCache.LastfmCacheError: time.sleep(1) # release artists for track in release.tracks.values(): for artist in track.release_artists: while True: try: validated_artist = self.lastfm.get_artist( normalize_artist_name(artist)).artist_name if validated_artist != artist: violations.add( Violation( ViolationType.RELEASE_ARTIST_SPELLING, "Incorrectly spelled Release Artist '{0}' (should be '{1}')" .format(artist, validated_artist))) break except LastfmCache.ArtistNotFoundError: # as e: # violations.add(str(e)) break except LastfmCache.LastfmCacheError: time.sleep(1) validated_track_numbers = release.validate_track_numbers() if validated_track_numbers: flattened_track_nums = [] for disc in validated_track_numbers: flattened_track_nums.append( "\nDisc " + str(disc) + ": " + ",".join(str(i) for i in validated_track_numbers[disc])) violations.add( Violation( ViolationType.MISSING_TRACKS, "Release does not have a full set of tracks:{0}".format( "".join(flattened_track_nums)))) validated_total_tracks = release.validate_total_tracks() for disc in validated_total_tracks: violations.add( Violation( ViolationType.TOTAL_TRACKS_INCONSISTENT, "Release disc {0} has blank, inconsistent or incorrect 'Total Tracks' tags" .format(disc))) # disc number validated_disc_numbers = release.validate_disc_numbers() if validated_disc_numbers: violations.add( Violation( ViolationType.MISSING_DISCS, "Release does not have a full set of discs: {0}".format( ", ".join(str(i) for i in validated_disc_numbers)))) # total discs if not release.validate_total_discs(): violations.add( Violation(ViolationType.TOTAL_DISCS_INCONSISTENT, "Release has incorrect 'Total Discs' tags")) # file type if len(release.get_tag_types()) != 1: violations.add( Violation( ViolationType.TAG_TYPES_INCONSISTENT, "Release has inconsistent tag types: {0}".format(", ".join( [str(x) for x in release.get_tag_types()])))) # bitrate - CBR/VBR/Vx/APS/APE if len(release.get_codecs()) != 1: violations.add( Violation( ViolationType.CODECS_INCONSISTENT, "Release has inconsistent codecs: [{0}]".format(", ".join( release.get_codecs())))) if len(unique([int(x / 1000) for x in release.get_cbr_bitrates()])) > 1: violations.add( Violation( ViolationType.CBR_INCONSISTENT, "Release has inconsistent CBR bitrates: {0}".format( ", ".join([str(x) for x in release.get_cbr_bitrates()])))) # track titles for filename in release.tracks: correct_filename = release.tracks[filename].get_filename( release.is_va()) if correct_filename and filename != correct_filename: violations.add( Violation( ViolationType.FILENAME, "Invalid filename: {0} - should be '{1}'".format( filename, correct_filename))) # forbidden comment substrings for track in release.tracks.values(): if not track.comment: continue for substr in self.forbidden_comment_substrings: if substr in track.comment.lower(): violations.add( Violation( ViolationType.COMMENT_SUBSTRING, "Invalid comment: contains forbidden substring '{0}'" .format(substr))) release.num_violations = len(violations) return list(violations)
def __repr__(self): track0 = self.tracks[next(iter(self.tracks))] date = "<date not found>" if not track0.date else track0.date.split("-")[0] return "{0} - {1} - {2}"\ .format(flatten_artists(track0.release_artists), date, track0.release_title)