ispedals/youtube-video-explorer.pl

## youtube-video-explorer.pl
#perl
use v5.16;
use warnings;
use XML::Smart;
use Mojolicious::Lite;

my %videos; #videos to be saved; structured <id:[title, author_url, id, duration, [thumbnails]]>
my %seen; #videos that have been seen and either saved or rejected; keys are video ids
my %authors; #authors whose uploads have been seen; keys are author urls

my @to_process = (
    [
    ]
); #initialize with videos

$seen{$_->[2]} = 1 for @to_process;

sub get_videos_from_feed {
    my $url = shift;

    my $ret = [];

    #actually 97% faster than Mojo::DOM
    my $x   = XML::Smart->new($url);
    for ( @{ $x->{feed}->{entry} } ) {
        my $title  = $_->{title}->content;
        my $author = $_->{author}->{uri}->content . '/uploads?v=2';
        my $id     = $_->{'media:group'}->{'yt:videoid'}->content;
        my @thumbs =
          map { $_->{url} }
          @{ $_->{'media:group'}->{'media:thumbnail'} }[ -3 .. -1 ]; #last three thumbs are start, middle and end of video
        my $duration = $_->{'media:group'}->{'yt:duration'}->{seconds};
        next if $videos{$id} || $seen{$id};
        push( @$ret, [ $title, $author, $id, $duration, \@thumbs ] );
    }
    if ( $x->{feed}->{link}( 'rel', 'eq', 'next' ) ) {
        $url = $x->{feed}{link}( 'rel', 'eq', 'next' )->{href}->content;
        my @next = get_videos_from_feed($url);
        return @$ret, @next;
    }
    else {
        return @$ret;
    }

    #use Mojo::UserAgent;
    #state $ua = Mojo::UserAgent->new;
    #my $dom = $ua->get($url)->res->dom;
    #for ( $dom->find('media\:group')->each ) {
    #    my $title  = $_->at('media\:title')->text;
    #    my $author = $_->at('yt\:uploaderId')->text;
    #    $author =
    #      sprintf 'http://gdata.youtube.com/feeds/api/users/%s/uploads?v=2',
    #      $author;
    #    my $id = $_->at('yt\:videoid')->text;
    #    my @thumbs =
    #      map { $_->{url} } ( $_->find('media\:thumbnail')->each )[ -3 .. -1 ];
    #    my $duration = $_->at('yt\:duration')->{seconds};
    #    next if $videos{$id} || $seen{$id};
    #    push( @$ret, [ $title, $author, $id, $duration, \@thumbs ] );
    #}
    #return @$ret unless ( $url = $dom->at('link[rel="next"]') );

    #my @next = get_videos_from_feed( $url->{href} );
    #return @$ret, @next;
}

sub process {
    my $vid = shift @to_process;

    my @ret;

    my @related_videos =
        get_videos_from_feed "http://gdata.youtube.com/feeds/api/videos/"
      . $vid->[2]
      . "/related?v=2";
    my @author_uploads =
      $authors{ $vid->[1] } ? () : get_videos_from_feed $vid->[1];

    for ( ( @related_videos, @author_uploads ) ) {
        next if $videos{ $_->[2] } || $seen{ $_->[2] } || $authors{ $_->[1] };
        $seen{ $_->[2] } = 1;
        push @ret, $_;
    }

    $authors{ $vid->[1] } = 1;
    return @ret;

}

my @to_be_selected; #contains all the videos derived from a single video, filtered in post
get '/' => sub {
    my $self = shift;
    @to_be_selected = process;
    my $picked    = scalar keys %videos;
    my $processed = scalar keys %seen;
    my $out = "<!html><head><title>Picked $picked out of $processed</title></head><body><form method='post'>";
    for (@to_be_selected) {
        my ( $title, $id, $duration ) = @{$_}[ 0, 2, 3 ];
        $out .=
"<div><p>$title ($duration seconds)</p><input type='checkbox' name='$id'>";
        $out .= "<img src='$_'>" for @{ $_->[4] };
        $out .= '</div>';
    }
    $out .= '<input type="submit" value="Submit">';
    $self->render( text => $out );
};

post '/' => sub {
    my $self = shift;
    my %wanted = map { $_ => 1 } $self->param; #params contains the ids of the video that were selected to be saved
    @to_be_selected = grep { $wanted{ $_->[2] } } @to_be_selected;
    $videos{ $_->[2] } = $_ for @to_be_selected;
    push @to_process, @to_be_selected;
    @to_be_selected = ();
    $self->redirect_to('/');
};

get '/view' => sub {
    my $self = shift;
    my $picked    = scalar keys %videos;
    my $processed = scalar keys %seen;
    my $out = "<!html><head><title>Picked $picked out of $processed</title></head><body>";
    for ( values %videos ) {
        my ( $title, $id, $duration ) = @{$_}[ 0, 2, 3 ];
        $out .= "<div><p><a href='https://www.youtube.com/watch?v=$id'>$title</a> ($duration seconds)</p>";
        $out .= '<img src="' . $_ . '">' for @{ $_->[4] };
        $out .= '</div>';
    }
    $out .= "http://www.youtube.com/watch?v=$_<br>" for keys %videos;
    $out .= '<hr><h1>Seen</h1>';
    $out .= "http://www.youtube.com/watch?v=$_<br>" for keys %seen;
    $out .= '<hr><h1>Authors</h1>';
    $out .= "$_<br>"                                for keys %authors;
    $self->render( text => $out );
};

app->start;
	#perl
	use v5.16;
	use warnings;
	use XML::Smart;
	use Mojolicious::Lite;

	my %videos; #videos to be saved; structured <id:[title, author_url, id, duration, [thumbnails]]>
	my %seen; #videos that have been seen and either saved or rejected; keys are video ids
	my %authors; #authors whose uploads have been seen; keys are author urls

	my @to_process = (
	[
	]
	); #initialize with videos

	$seen{$_->[2]} = 1 for @to_process;

	sub get_videos_from_feed {
	my $url = shift;

	my $ret = [];

	#actually 97% faster than Mojo::DOM
	my $x = XML::Smart->new($url);
	for ( @{ $x->{feed}->{entry} } ) {
	my $title = $_->{title}->content;
	my $author = $_->{author}->{uri}->content . '/uploads?v=2';
	my $id = $_->{'media:group'}->{'yt:videoid'}->content;
	my @thumbs =
	map { $_->{url} }
	@{ $_->{'media:group'}->{'media:thumbnail'} }[ -3 .. -1 ]; #last three thumbs are start, middle and end of video
	my $duration = $_->{'media:group'}->{'yt:duration'}->{seconds};
	next if $videos{$id} \|\| $seen{$id};
	push( @$ret, [ $title, $author, $id, $duration, \@thumbs ] );
	}
	if ( $x->{feed}->{link}( 'rel', 'eq', 'next' ) ) {
	$url = $x->{feed}{link}( 'rel', 'eq', 'next' )->{href}->content;
	my @next = get_videos_from_feed($url);
	return @$ret, @next;
	}
	else {
	return @$ret;
	}

	#use Mojo::UserAgent;
	#state $ua = Mojo::UserAgent->new;
	#my $dom = $ua->get($url)->res->dom;
	#for ( $dom->find('media\:group')->each ) {
	# my $title = $_->at('media\:title')->text;
	# my $author = $_->at('yt\:uploaderId')->text;
	# $author =
	# sprintf 'http://gdata.youtube.com/feeds/api/users/%s/uploads?v=2',
	# $author;
	# my $id = $_->at('yt\:videoid')->text;
	# my @thumbs =
	# map { $_->{url} } ( $_->find('media\:thumbnail')->each )[ -3 .. -1 ];
	# my $duration = $_->at('yt\:duration')->{seconds};
	# next if $videos{$id} \|\| $seen{$id};
	# push( @$ret, [ $title, $author, $id, $duration, \@thumbs ] );
	#}
	#return @$ret unless ( $url = $dom->at('link[rel="next"]') );

	#my @next = get_videos_from_feed( $url->{href} );
	#return @$ret, @next;
	}

	sub process {
	my $vid = shift @to_process;

	my @ret;

	my @related_videos =
	get_videos_from_feed "http://gdata.youtube.com/feeds/api/videos/"
	. $vid->[2]
	. "/related?v=2";
	my @author_uploads =
	$authors{ $vid->[1] } ? () : get_videos_from_feed $vid->[1];

	for ( ( @related_videos, @author_uploads ) ) {
	next if $videos{ $_->[2] } \|\| $seen{ $_->[2] } \|\| $authors{ $_->[1] };
	$seen{ $_->[2] } = 1;
	push @ret, $_;
	}

	$authors{ $vid->[1] } = 1;
	return @ret;

	}

	my @to_be_selected; #contains all the videos derived from a single video, filtered in post
	get '/' => sub {
	my $self = shift;
	@to_be_selected = process;
	my $picked = scalar keys %videos;
	my $processed = scalar keys %seen;
	my $out = "<!html><head><title>Picked $picked out of $processed</title></head><body><form method='post'>";
	for (@to_be_selected) {
	my ( $title, $id, $duration ) = @{$_}[ 0, 2, 3 ];
	$out .=
	"<div><p>$title ($duration seconds)</p><input type='checkbox' name='$id'>";
	$out .= "<img src='$_'>" for @{ $_->[4] };
	$out .= '</div>';
	}
	$out .= '<input type="submit" value="Submit">';
	$self->render( text => $out );
	};

	post '/' => sub {
	my $self = shift;
	my %wanted = map { $_ => 1 } $self->param; #params contains the ids of the video that were selected to be saved
	@to_be_selected = grep { $wanted{ $_->[2] } } @to_be_selected;
	$videos{ $_->[2] } = $_ for @to_be_selected;
	push @to_process, @to_be_selected;
	@to_be_selected = ();
	$self->redirect_to('/');
	};

	get '/view' => sub {
	my $self = shift;
	my $picked = scalar keys %videos;
	my $processed = scalar keys %seen;
	my $out = "<!html><head><title>Picked $picked out of $processed</title></head><body>";
	for ( values %videos ) {
	my ( $title, $id, $duration ) = @{$_}[ 0, 2, 3 ];
	$out .= "<div><p><a href='https://www.youtube.com/watch?v=$id'>$title</a> ($duration seconds)</p>";
	$out .= '<img src="' . $_ . '">' for @{ $_->[4] };
	$out .= '</div>';
	}
	$out .= "http://www.youtube.com/watch?v=$_<br>" for keys %videos;
	$out .= '<hr><h1>Seen</h1>';
	$out .= "http://www.youtube.com/watch?v=$_<br>" for keys %seen;
	$out .= '<hr><h1>Authors</h1>';
	$out .= "$_<br>" for keys %authors;
	$self->render( text => $out );
	};

	app->start;