Enrich Music with Metadata

plugin/API.pm

@@ -57,14 +57,33 @@ sub getCategories {
 sub getVideoDetails {
 	my ( $class, $cb, $ids ) = @_;
 
+	# topicDetails for genre
 	_call('videos', {
-		part => 'snippet,contentDetails',
+		part => 'snippet,contentDetails,topicDetails',
 		id   => $ids,
 		# cache video details a bit longer
 		_cache_ttl => 7 * 86400,
 	}, $cb);
 }
 
+# single-shot fetch of the first N items of a playlist, no paging. used by
+# the music-playlist filter where we just want a small sample to classify.
+# pass extra (eg. access_token + _noKey) to read private playlists.
+sub getPlaylistSample {
+	my ( $class, $cb, $playlistId, $max, $extra ) = @_;
+
+	my %args = (
+		part       => 'snippet',
+		playlistId => $playlistId,
+		maxResults => $max || 5,
+		_noRegion  => 1,
+		_cache_ttl => 86400,
+		%{ $extra || {} },
+	);
+
+	_call('playlistItems', \%args, $cb);
+}
+
 sub _pagedCall {
 	my ( $method, $args, $cb ) = @_;
 	my $wantedItems = $args->{_quantity} || $prefs->get('max_items');

plugin/HTML/EN/plugins/YouTube/settings/basic.html

@@ -159,6 +159,14 @@
 		<input name="pref_highres_icons" type="checkbox" [% IF prefs.pref_highres_icons %] checked [% END %]>
 	[% END %]
 
+	[% WRAPPER setting title="PLUGIN_YOUTUBE_MUSICONLY" desc="PLUGIN_YOUTUBE_MUSICONLY_DESC" %]
+		<input name="pref_music_playlists_only" type="checkbox" [% IF prefs.pref_music_playlists_only %] checked [% END %]>
+	[% END %]
+
+	[% WRAPPER setting title="PLUGIN_YOUTUBE_RICHMETA" desc="PLUGIN_YOUTUBE_RICHMETA_DESC" %]
+		<input name="pref_rich_metadata" type="checkbox" [% IF prefs.pref_rich_metadata %] checked [% END %]>
+	[% END %]
+
 	[% WRAPPER setting title="PLUGIN_YOUTUBE_MAXITEMS" desc="PLUGIN_YOUTUBE_MAXITEMS_DESC" %]
 		<input type="text" class="stdedit" name="pref_max_items" id="pref_max_items" value="[% prefs.pref_max_items %]" size="5">
 	[% END %]

plugin/Metadata.pm

@@ -0,0 +1,153 @@
+package Plugins::YouTube::Metadata;
+
+# helpers for v3 video items. v3 has no album/year/artist fields,
+# we dig artist/album/year out of the description block when YT
+# Music auto-generated it.
+
+use strict;
+use warnings;
+
+use Slim::Utils::Log;
+
+my $log = logger('plugin.youtube');
+
+use constant CATEGORY_MUSIC => 10;
+
+# "Artist - Title" split. returns (artist, title, fulltitle) or
+# (undef, $title, undef).
+sub parseDashTitle {
+	my ($title) = @_;
+	return (undef, $title, undef) unless defined $title;
+
+	if ($title =~ /^(.+?)\s+-\s+(.+)$/) {
+		return ($1, $2, $title);
+	}
+	return (undef, $title, undef);
+}
+
+# the auto-generated YT Music "Provided to YouTube by..." block.
+# shape:
+#   Provided to YouTube by <distributor>
+#   <title> . <artist1> . <artist2>
+#   <album>
+#   . YYYY <label>
+#   Released on: YYYY-MM-DD
+# middot is U+00B7. we also accept " - " and " / " because some
+# uploaders use those.
+sub parseYTMusicDescription {
+	my ($desc) = @_;
+	return {} unless defined $desc && length $desc;
+	return {} unless $desc =~ /Provided to YouTube by/i;
+
+	my @lines = grep { /\S/ } split /\n/, $desc;
+	shift @lines if @lines && $lines[0] =~ /Provided to YouTube by/i;
+
+	return {} unless @lines >= 2;
+
+	my $titleLine = $lines[0];
+	my $album     = $lines[1];
+
+	my @parts = split /\s*(?:\x{00b7}|\xc2\xb7|·| - | \/ )\s*/, $titleLine;
+	return {} unless @parts >= 2;
+
+	my $title  = shift @parts;
+	my $artist = join(', ', @parts);
+
+	s/^\s+|\s+$//g for ($title, $artist, $album);
+
+	my $year;
+	for my $ln (@lines) {
+		if ($ln =~ /Released on:\s*(\d{4})/) {
+			$year = $1;
+			last;
+		}
+		if ($ln =~ /^\s*(?:\x{00b7}|\xc2\xb7|·|\xe2\x84\x97|\(P\))\s*(\d{4})\b/) {
+			$year ||= $1;
+		}
+	}
+
+	return {
+		title  => $title,
+		artist => $artist,
+		album  => $album,
+		year   => $year,
+	};
+}
+
+# topic categories are wikipedia URLs like .../wiki/Pop_music.
+# trim to "Pop music". drop the generic "Music" entry, useless tag.
+sub topicCategoriesToGenre {
+	my ($topicDetails) = @_;
+	return undef unless ref($topicDetails) eq 'HASH';
+	my $cats = $topicDetails->{topicCategories};
+	return undef unless ref($cats) eq 'ARRAY' && @$cats;
+
+	my @names;
+	for my $url (@$cats) {
+		next unless $url =~ m{/wiki/([^/]+)$};
+		my $name = $1;
+		$name =~ s/_/ /g;
+		$name =~ s/%([0-9A-Fa-f]{2})/chr(hex($1))/ge;
+		next if $name eq 'Music';
+		push @names, $name;
+	}
+
+	return undef unless @names;
+	return join(', ', @names);
+}
+
+# ISO 8601 2024-08-23T14:00:00Z to "2024"
+sub yearFromPublishedAt {
+	my ($publishedAt) = @_;
+	return undef unless defined $publishedAt;
+	return $1 if $publishedAt =~ /^(\d{4})/;
+	return undef;
+}
+
+# v3 item to flat metadata hash. each field is best-effort.
+sub buildFromV3Item {
+	my ($item) = @_;
+	return {} unless ref($item) eq 'HASH';
+
+	my $snippet = $item->{snippet}    || {};
+	my $content = $item->{contentDetails} || {};
+	my $topics  = $item->{topicDetails}   || {};
+
+	my ($artist, $title, $fulltitle) = parseDashTitle($snippet->{title});
+
+	# description block beats the dash split if its there
+	my $ytm = parseYTMusicDescription($snippet->{description});
+	my $album;
+	my $year;
+
+	if ($ytm->{title}) {
+		$title     = $ytm->{title};
+		$fulltitle = $snippet->{title};
+		$artist    = $ytm->{artist} if $ytm->{artist};
+		$album     = $ytm->{album};
+		$year      = $ytm->{year};
+	}
+
+	$artist ||= $snippet->{channelTitle};
+	$year   ||= yearFromPublishedAt($snippet->{publishedAt});
+
+	my $genre = topicCategoriesToGenre($topics);
+	$genre ||= 'Music' if ($snippet->{categoryId} || 0) == CATEGORY_MUSIC;
+
+	my $duration = $content->{duration};
+	my ($misc, $hour, $min, $sec) = $duration =~ /P(?:([^T]*))T(?:(\d+)H)?(?:(\d+)M)?(?:(\d+)S)?/ if $duration;
+	$duration = ($sec || 0) + (($min || 0) * 60) + (($hour || 0) * 3600);
+
+	return {
+		title      => $title || '',
+		artist     => $artist || '',
+		album      => $album,
+		year       => $year,
+		genre      => $genre,
+		duration   => $duration || 0,
+		_fulltitle => $fulltitle,
+		_snippet   => $snippet,
+	};
+}
+
+1;