unweb/youtube.com/search.pl

#!/usr/bin/perl
# Copyright (c) 2021 smpl <smpl@slamkode.ml>
# SPDX-License-Identifier: Zlib

use Encode;
use JSON::XS;
use HTML::TreeBuilder 5 -weak;
use LWP;
use LWP::UserAgent;
use MIME::Base64;
use Getopt::Long;
use strict;

sub print_help {
	print "Usage: " . $0 . " [--sort][--uploaded][--type][--duration][--features] QUERY\n" .
		"--sort upload|views|rating|relevance\n" .
		"--uploaded hour|day|week|month|year\n" .
		"--type video|channel|playlist|movie\n" .
		"--duration short|long\n" .
		"--features live|4k|hd|sub|cc|360|vr180|3d|hdr|location|purchased\n\n";
}

sub parse_cli {
	my %options = (
		"sort" => {
			"upload"	=> "\x08\x02", 
			"views"		=> "\x08\x03", 
			"rating"	=> "\x08\x01", 
			"relevance" => "\x08\x00"
		},
		"uploaded" => {
			"hour"	=> "\x08\x01",
			"day"	=> "\x08\x02",
			"week"	=> "\x08\x03",
			"month"	=> "\x08\x04",
			"year"	=> "\x08\x05"
		},
		"type" => {
			"video"		=> "\x10\x01",
			"channel"	=> "\x10\x02",
			"playlist"	=> "\x10\x03",
			"movie"		=> "\x10\x04"
		},
		"duration" => {
			"short"	=> "\x18\x01",
			"long"	=> "\x18\x02"
		},
		"features" => {
			"live"		=> "\x40\x01",
			"4k"		=> "\x70\x01",
			"hd"		=> "\x20\x01",
			"sub"		=> "\x28\x01",
			"cc"		=> "\x30\x01",
			"360"		=> "\x78\x01",
			"vr180"		=> "\xd0\x01\x01",
			"3d"		=> "\x38\x01",
			"hdr"		=> "\xc8\x01\x01",
			"location"	=> "\xb8\x01\x01",
			"purchased"	=> "\x48\x01"
		}
	);
	
	my ($sp, $sp_raw);
	my ($sort, $uploaded, $type, $duration, $features);
	my $filter_bytes = 0;
	my $filter = "";
	
	GetOptions(
		"sort:s" => \$sort,
		"uploaded:s" => \$uploaded,
		"type:s" => \$type,
		"duration:s" => \$duration,
		"features:s" => \$features
	) or die("Error in command line arguments\n");
	
	if($sort && defined($options{"sort"}{$sort})) {
		$sp_raw = $options{"sort"}{$sort};
	}
	
	if($type ne "video" && $type ne "movie") {
		if($duration && defined($options{"duration"}{$duration})) {
			print "\n--duration can only be combined with type video or movie.\n\n";
			print_help;
			exit 1;
		}
		if($features && defined($options{"features"}{$features})) {
			print "\n--features can only be combined with type video or movie.\n\n";
			print_help;
			exit 1;
		}
		if($uploaded && defined($options{"uploaded"}{$uploaded})) {
			print "\n--uploaded can only be combined with type video or movie.\n\n";
			print_help;
			exit 1;
		}
	}
	
	if($uploaded && defined($options{"uploaded"}{$uploaded})) {
		$filter_bytes += bytes::length($options{"uploaded"}{$uploaded});
		$filter .= $options{"uploaded"}{$uploaded};
	}
	if($type && defined($options{"type"}{$type})) {
		$filter_bytes += bytes::length($options{"type"}{$type});
		$filter .= $options{"type"}{$type};
	}
	if($duration && defined($options{"duration"}{$duration})) {
		$filter_bytes += bytes::length($options{"duration"}{$duration});
		$filter .= $options{"duration"}{$duration};
	}
	if($features && defined($options{"features"}{$features})) {
		$filter_bytes += bytes::length($options{"features"}{$features});
		$filter .= $options{"features"}{$features};
	}
	
	if($filter_bytes > 0) {
		$sp = encode_base64($sp_raw . "\x12" . chr($filter_bytes) . $filter);
	} else {
		$sp = encode_base64($sp_raw);
	}
	
	$sp =~ s/([+\/=])/sprintf("%%25%02X", ord($1))/eg;
	
	return $sp;
}

my $sp = "";

if($#ARGV != 0) {
	if($#ARGV < 0) {
		print_help;
		exit;
	}
	
	$sp = parse_cli;
	
	if($#ARGV != 0) {
		print_help;
		exit;
	}
}

my $params = $ARGV[0];
if(bytes::length($sp) > 0) {
	$params .= "&sp=".$sp;
}

my $ua = LWP::UserAgent->new;
$ua->agent("Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:76.0) Gecko/20100101 Firefox/76.0");
$ua->cookie_jar( {} );
$ua->cookie_jar()->set_cookie(0, 'CONSENT', 'YES+cb.20210530-19-p0.en+FX+027', '/', '.youtube.com');
my $response = $ua->get('https://www.youtube.com/results?search_query='.$params);

my $jsonobject;
my $coder = JSON::XS->new->pretty->allow_nonref;
my $tree = HTML::TreeBuilder->new;

$tree->parse_content($response->content);

my $body = $tree->find_by_tag_name('body');

my $ytInitialData_tag = $body->look_down(
	_tag => "script",
	sub { $_[0]->as_HTML =~ /ytInitialData/ }
);

$ytInitialData_tag->as_HTML =~ /.*ytInitialData[^=]*= (.*});.*/;
my $ytInitialData = $1;

$jsonobject = $coder->decode($ytInitialData);
my $utf8_encoded_json_text = $coder->encode($jsonobject);

my @results = @{$jsonobject->{"contents"}->{'twoColumnSearchResultsRenderer'}->{'primaryContents'}->{'sectionListRenderer'}->{'contents'}->[0]->{'itemSectionRenderer'}->{'contents'}};

sub get_textarray
{
	my $aref = shift @_;
	my $text = "";
	foreach my $line (@{$aref}) {
		$text .= $line->{'text'};
	}
	return $text;
}

sub print_video
{
	my $video = shift @_;
	print "Title: " . $video->{'videoRenderer'}->{'title'}->{'runs'}->[0]->{'text'} . "\n";
	print "Length: " . $video->{'videoRenderer'}->{'lengthText'}->{'simpleText'} . "\n";
	print "Published: " . $video->{'videoRenderer'}->{'publishedTimeText'}->{'simpleText'} . "\n";
	print "Views: " . $video->{'videoRenderer'}->{'viewCountText'}->{'simpleText'} . "\n";
	print "By: " . $video->{'videoRenderer'}->{'ownerText'}->{'runs'}->[0]->{'text'} . "\n";
	print "Feed: ";
	my $feeduri = $video->{'videoRenderer'}->{'ownerText'}->{'runs'}->[0]->{'navigationEndpoint'}->{'commandMetadata'}->{'webCommandMetadata'}->{'url'};
	if($feeduri =~ m{^/user/}) {
		print 'https://www.youtube.com/feeds/videos.xml?user=' . $feeduri =~ s!^/user/!!r;
	} else {
		print 'http://www.youtube.com/feeds/videos.xml?channel_id=' . $feeduri =~ s!^/channel/!!r;
	}
	print "\n";
	print "Url: https://www.youtube.com/watch?v=" . $video->{'videoRenderer'}->{'videoId'} . "\n\n";
}

sub print_playlist
{
	my $playlist = shift @_;
	print "Title: " . $playlist->{'playlistRenderer'}->{'title'}->{'simpleText'} . "\n";
	print "Videos: " . $playlist->{'playlistRenderer'}->{'videoCount'} . "\n";
	print "Url: https://www.youtube.com/playlist?list=" . $playlist->{'playlistRenderer'}->{'playlistId'} . "\n";
	print "Feed: https://www.youtube.com/feeds/videos.xml?playlist_id=" . $playlist->{'playlistRenderer'}->{'playlistId'} . "\n\n";
}

sub print_channel
{
	my $channel = shift @_;
	print "Title: " . $channel->{'channelRenderer'}->{'title'}->{'simpleText'} . "\n";
	if($channel->{'channelRenderer'}->{'descriptionSnippet'}) {
		print "Description: " . get_textarray($channel->{'channelRenderer'}->{'descriptionSnippet'}->{'runs'}) . "\n";
	}
	if($channel->{'channelRenderer'}->{'subscriberCountText'}) {
		print "Subscribers: " . $channel->{'channelRenderer'}->{'subscriberCountText'}->{'simpleText'} . "\n";
	}
	print "Videos: " . get_textarray($channel->{'channelRenderer'}->{'videoCountText'}->{'runs'}) . "\n";
	print "Url: https://www.youtube.com/channel/" . $channel->{'channelRenderer'}->{'channelId'} . "\n";
	print "Feed: http://www.youtube.com/feeds/videos.xml?channel_id=" . $channel->{'channelRenderer'}->{'channelId'} . "\n\n";
}

foreach my $item (@results) {
	my ($key, $title, $length, $date, $views, $id);
	foreach my $k (keys %{$item}) {
		$key = $k;
	}

	if($key eq 'videoRenderer') {
		print "[video]\n";
		print_video($item);
	} elsif($key eq 'playlistRenderer') {
		print "[playlist]\n";
		print_playlist($item);
	} elsif($key eq 'channelRenderer') {
		print "[channel]\n";
		print_channel($item);
	}
}