fix quoting in AsyncWgetFileGetter again

[videosite.git] / videosite / GoogleGrabber.pm

diff --git a/videosite/GoogleGrabber.pm b/videosite/GoogleGrabber.pm
index c1fe9eb..51d0e2e 100644 (file)
--- a/videosite/GoogleGrabber.pm
+++ b/videosite/GoogleGrabber.pm
@@ -3,28 +3,28 @@
 #
 # Grabber for video.google.com
 
-package GoogleGrabber;
+package videosite::GoogleGrabber;
 
-use GrabberBase;
-@ISA = qw(GrabberBase);
+use videosite::GrabberBase;
+@ISA = qw(videosite::GrabberBase);
 
-use LWP::Simple qw(!get);
-use HTML::Parser;
+use HTML::TokeParser;
 use Data::Dumper;
 
 use strict;
 
 sub new {
     my $class = shift;
-    my $self = $class->SUPER::new();
-
-    $self->{'NAME'} = 'google';
-    $self->{'PATTERNS'} = ['(http://video\.google\.com/videoplay\?docid=([-\d]+))'];
-
-    bless($self, $class);
-    $self->_prepare_parameters();
-
-    return $self;
+    my $self = $class->SUPER::new(
+        NAME => 'google',
+        PATTERNS => ['(http://video\.google\.com/videoplay\?docid=([-\d]+))'],
+        _PARAMS => {
+            QUALITY => ['normal', 'Quality of the video to download. normal = standard resolution flash video, h264 = high resolution MPEG4 video']
+        },
+        @_,
+    );
+
+    return bless($self, $class);
 }
 
 sub _parse {
@@ -33,10 +33,9 @@ sub _parse {
     my $pattern = shift;
     my $content;
     my $metadata = {};
-    my $p = HTML::Parser->new(api_version => 3);
-    my @accum;
-    my @text;
+    my $p;
     my $e;
+    my $quality = $self->_getval('QUALITY');
 
     $url =~ m|$pattern|;
     $url = $1;
@@ -48,29 +47,33 @@ sub _parse {
     $metadata->{'TITLE'} = undef;
     $metadata->{'DLURL'} = undef;
 
-    unless(defined($content = LWP::Simple::get(sprintf('http://video.google.com/videoplay?docid=%s', $2)))) {
+    unless(defined($content = $self->simple_get(sprintf('http://video.google.com/videohosted?docid=%s', $2)))) {
         $self->error('Could not download %s', $url);
         return undef;
     }
 
-    $p->handler(start => \@accum, "tagname, attr");
-    $p->handler(text => \@text, "text");
-    $p->report_tags(qw(embed div));
-    $p->utf8_mode(1);
-    $p->parse($content);
-
-    # Look for the title in the div tags
-    foreach $e (@accum) {
-        if ('div' eq $e->[0]) {
-            if ((exists($e->[1]->{'class'})) and ('title' eq $e->[1]->{'class'})) {
-                $metadata->{'TITLE'} = $e->[1]->{'title'};
+    $p = HTML::TokeParser->new(\$content);
+
+    # Look for the title
+    if ($p->get_tag('title')) {
+        $metadata->{'TITLE'} = $p->get_text();
+        $metadata->{'TITLE'} =~ s/\s?- Google Video$//s;
+    }
+
+    if ($quality eq 'h264') {
+        while ($e = $p->get_tag('a')) {
+            if ((exists($e->[1]{'id'})) and ('ipoddownloadlink' eq $e->[1]{'id'})) {
+                $metadata->{'DLURL'} = $e->[1]{'href'};
+                last;
             }
         }
-
-        if ('embed' eq $e->[0]) {
-            if ((exists($e->[1]->{'src'})) and ($e->[1]->{'src'} =~ m|^/googleplayer.swf\?\&videoUrl=([^&]+)\&|)) {
-                $metadata->{'DLURL'} = $1;
-                $metadata->{'DLURL'} =~ s/%(..)/chr(hex($1))/ge;
+    } else {
+        while ($e = $p->get_tag('script')) {
+            if ($p->get_text() =~ m|googleplayer\.swf\?\\46videoUrl\\75(.+?)\\46|s) {
+                my $u = $1;
+                $u =~ s/%(..)/chr(hex($1))/ge;
+                $metadata->{'DLURL'} = $u;
+                last;
             }
         }
     }