use LWP::UserAgent;
use HTTP::Cookies;
use HTML::TokeParser;
+use HTML::Entities qw(decode_entities);
+use Encode;
use Data::Dumper;
+use videosite::JSArrayParser;
use strict;
my $self = $class->SUPER::new();
$self->{'NAME'} = 'youtube';
- $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*youtube.(?:com|de|co.uk)/watch\?(?:.+=.+&)*v=([-a-zA-Z0-9_]+))',
- '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))'];
+ $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\?(?:.+=.+&)*v=([-a-zA-Z0-9_]+))',
+ '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\#\!v=([-a-zA-Z0-9_]+))',
+ '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))',
+ '(http://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/([-a-zA-Z0-9_]+))',
+ '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/\w+\?.*/([-a-zA-Z0-9_]+))'];
$self->{'_PARAMS'} = {
'QUALITY' => ['normal', 'Quality of the video to download.', {
'normal' => 'standard resolution flash video',
my $videourl;
my $quality = $self->_getval('QUALITY');
my %preflist = (
- 'hd' => [22, 35, 18, 34, 6, 5],
- 'h264' => [18, 34, 22, 35, 6, 5],
- 'high' => [34, 35, 18, 22, 6, 5],
- 'normal' => [6, 5, 34, 35, 18, 22]);
+ 'hd' => [37, 22, 35, 18, 34, 6, 5],
+ 'h264' => [18, 34, 37, 22, 35, 6, 5],
+ 'high' => [34, 35, 18, 37, 22, 6, 5],
+ 'normal' => [6, 5, 34, 35, 18, 22, 37]);
my $preflist;
+ my $jsp;
$url =~ m|$pattern|;
$url = $1;
$metadata->{'TITLE'} = undef;
$metadata->{'DLURL'} = undef;
+ $self->debug("Matched id %s from pattern %s", $2, $pattern);
+
$preflist = $preflist{$quality};
$self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
$p = HTML::TokeParser->new(\$content);
- while ($tag = $p->get_tag('div', 'meta', 'script')) {
+ SWF_ARGS: while ($tag = $p->get_tag('div', 'meta', 'script')) {
if ('meta' eq $tag->[0]) {
if ('title' eq $tag->[1]->{'name'}) {
$metadata->{'TITLE'} = $tag->[1]->{'content'};
+ # Convert HTML entities in the title. This is a bit convoluted.
+ $metadata->{'TITLE'} = encode("utf8",
+ decode_entities(
+ decode("utf8", $metadata->{'TITLE'})));
+
$self->debug('Title found: %s', $metadata->{'TITLE'});
}
} elsif ('script' eq $tag->[0]) {
+ my %urls;
+
$e = $p->get_text();
$self->debug("Found script: %s", $e);
- if ($e =~ m|\x22fmt_url_map\x22\s*:\s*\x22([^\x22]+)\x22|) {
- $self->debug("Found fmt_url_map: %s", $1);
+
+ if ($e =~ m|\x27SWF_ARGS\x27:\s+(.+),|) {
+ my $args = $1;
+
+ $self->debug("Found SWF_ARGS: %s", $args);
+ $jsp = videosite::JSArrayParser->new();
+ $self->debug("Using %s to parse", ref($jsp));
+ $r = $jsp->parse($args);
+
+ unless(defined($r)) {
+ $self->error("Found information hash, but could not parse");
+ return undef;
+ }
+
+ if (exists($r->{'fmt_url_map'}) and ($r->{'fmt_url_map'} ne '')) {
+ my $urls = $r->{'fmt_url_map'};
+
+ $self->debug("Video has fmt_url_map: %s", $urls);
+
+ $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
+ %urls = split(/[\|,]/, $urls);
+ $self->debug("Pagetype: old (SWF_ARGS), fmt_url_map");
+
+ } elsif (exists($r->{'t'}) and ($r->{'t'} ne '')) {
+ my $thash = $r->{'t'};
+
+ if (exists($r->{'fmt_map'}) && ($r->{'fmt_map'} ne '')) {
+ my $fmt = $r->{'fmt_map'};
+ my @fmt;
+
+ $self->debug('Video has fmt_map');
+ $fmt =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
+ @fmt = split(/,/, $fmt);
+ foreach (@fmt) {
+ split(/\//);
+ $urls{$_[0]} = sprintf('http://www.youtube.com/get_video?video_id=%s&fmt=%d&t=%s',
+ $metadata->{'ID'},
+ $_[0],
+ $thash);
+ }
+ $self->debug("Pagetype: 2009 (SWF_ARGS), t with fmt_map");
+
+ } else {
+ $urls{5} = sprintf('http://www.youtube.com/get_video?video_id=%s&t=%s',
+ $metadata->{'ID'},
+ $thash);
+ $self->debug("Pagetype: 2009 (SWF_ARGS), t without fmt_map");
+ }
+ } else {
+ $self->error('Neither fmt_url_map nor t found in video information hash');
+ return undef;
+ }
+ } elsif ($e =~ m|var swfHTML = .*fmt_url_map=([^\&]+)\&|) {
my $urls = $1;
- my %urls;
+ $self->debug("Video has fmt_url_map: %s", $urls);
$urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
%urls = split(/[\|,]/, $urls);
- $self->debug("Found quality levels [%s]", join(", ", keys(%urls)));
+ $self->debug("Pagetype: 2010 (swfHTML), fmt_url_map");
+ }
+ if (%urls) {
foreach (keys(%urls)) {
if ($_ == 35) {
$self->debug('Found flv,h264,large: %s', $urls{$_});
} elsif ($_ == 34) {
$self->debug('Found flv,h264: %s', $urls{$_});
} elsif ($_ == 22) {
- $self->debug('Found mp4,h264,large: %s', $urls{$_});
+ $self->debug('Found mp4,h264,720p: %s', $urls{$_});
+ } elsif ($_ == 37) {
+ $self->debug('Found mp4,h264,1080p: %s', $urls{$_});
} elsif ($_ == 18) {
$self->debug('Found mp4,h264: %s', $urls{$_});
} elsif ($_ == 5) {
}
$self->debug('URL found: %s', $metadata->{'DLURL'});
+ last SWF_ARGS;
}
} elsif ('div' eq $tag->[0]) {
- if (exists($tag->[1]->{'class'}) and ('errorBox' eq $tag->[1]->{'class'})) {
+ if (exists($tag->[1]->{'class'}) and ('yt-alert-content' eq $tag->[1]->{'class'})) {
$self->error("Could not get video data for youtube %s: %s",
$metadata->{'ID'}, $p->get_trimmed_text());
return undef;
return undef;
}
+ $self->debug("Got a cookie");
+
$r = $ua->get($videourl);
if ($r->base->as_string() =~ m,/verify_age,) {
$self->debug("Looking for session token...");