From 0aba4b4b872dc90259236362fbee3ec19af5164b Mon Sep 17 00:00:00 2001 From: Ralf Ertzinger Date: Fri, 19 Mar 2010 21:50:40 +0100 Subject: [PATCH] Add support for the new YouTube page layout --- videosite/YouTubeGrabber.pm | 19 +++++++++++++++++-- 1 file changed, 17 insertions(+), 2 deletions(-) diff --git a/videosite/YouTubeGrabber.pm b/videosite/YouTubeGrabber.pm index 769970d..9bc32d9 100644 --- a/videosite/YouTubeGrabber.pm +++ b/videosite/YouTubeGrabber.pm @@ -115,10 +115,12 @@ sub _parse { $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|\x27SWF_ARGS\x27:\s+(.+),|) { - my %urls; my $args = $1; $self->debug("Found SWF_ARGS: %s", $args); @@ -138,6 +140,8 @@ sub _parse { $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'}; @@ -155,17 +159,28 @@ sub _parse { $_[0], $thash); } + $self->debug("Pagetype: old (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: old (SWF_ARGS), t without fmt_map"); } } else { $self->error('Neither fmt_url_map nor t found in video information hash'); return undef; } - $self->debug("Found quality levels [%s]", join(", ", keys(%urls))); + } elsif ($e =~ m|var swfHTML = .*fmt_url_map=([^\&]+)\&|) { + my $urls = $1; + $self->debug("Video has fmt_url_map: %s", $urls); + + $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge; + %urls = split(/[\|,]/, $urls); + $self->debug("Pagetype: new (swfHTML), fmt_url_map"); + } + if (%urls) { foreach (keys(%urls)) { if ($_ == 35) { $self->debug('Found flv,h264,large: %s', $urls{$_}); -- 1.8.3.1