#!/usr/bin/perl
use strict;
use warnings;
use Gtk2 -init;
use Gtk2::WebKit;
use File::Copy;

my $url = 'http://www.youtube.com/watch?v=EAtBki0PsC0';  #'9uDgJ9_H0gg
+';
my $pwd = `pwd`;
chomp $pwd;

#where the flash will be cached temporarily by the browser
my $tmpdir =  Glib::get_tmp_dir(); 
print "$tmpdir\n";

# stat to monitor buffering
my $old_size = 0 ;

#  find names like FlashGRsDw9
opendir(my $dir, $tmpdir) or die("ack: $!");
my @files = grep /^Flash[A-Za-z0-9]{6}/, readdir $dir;
closedir $dir;
print "initial files in /tmp-> ", @files, "\n"; 

my $view = Gtk2::WebKit::WebView->new;

$view->signal_connect( 'notify::progress' => \&notify_progress, undef 
+);
$view->signal_connect( 'load_finished' => \&load_finished, undef );
$view->signal_connect( 'notify::load-status' => \&notify_load_status, 
+undef );

my $sw = Gtk2::ScrolledWindow->new;
$sw->add($view);

my $win = Gtk2::Window->new;
$win->set_default_size(800, 600);
$win->signal_connect(destroy => sub { Gtk2->main_quit });
$win->add($sw);
$win->show_all;

$view->open($url ); 

Gtk2->main;

sub notify_progress{
   my $load_progress = $view->get('progress');
    print "$load_progress\n";
}

sub load_finished{
      print "load complete\n";
      find_save();
return 0;
}

sub notify_load_status{
   print  $view->get('load_status'),"\n";
}


sub find_save{
   # this is just to detect the cache filename
   my $timer = Glib::Timeout->add (1000, sub { 
            # check for file every second
            # to allow flash loader to load file    
           print "test for file\n";
           my @files;
           opendir(my $dir, $tmpdir) or die("ack: $!");
           @files = grep /^Flash[A-Za-z0-9]{6}/, readdir $dir;
           closedir $dir;
           print "files-> ", @files, "\n"; 
       print scalar @files, "\n";
    
       if( scalar @files >= 1 ){     
                print "found file success\n";
                watch_save( $files[0] );  
            return 0; # end timer
        
          }else{ return 1 } # keep timer going
   
           });
}


sub watch_save{
   my $file = shift;
   my $watch = "$tmpdir/$file";
   print "watching file-> $watch\n";

# this is the problemsome point, if you are on a slow
# network, greater than 3 second delays may occur and trip a false EOF
# but it works on my slow dsl
   my $timer = Glib::Timeout->add (3000, sub{

      my $new_size = (stat $watch )[7]; 
      $new_size ||= 0;
      print "new size-> $new_size\n";      
      #see perldoc -f stat                                            
      #print lstat $file,"\n";;
      
       if(  $old_size != $new_size ){
             print "1 still buffering $old_size $new_size \n", ;
             $old_size = $new_size;
             return 1; #keep timer going 
         }else{
             #done unless there is a network hang
           print "flash done writing $pwd/$0-$file".'.flv',"\n";
           copy("$watch" ,"$pwd/$0-$file".'.flv') 
           or warn "Can not copy $!\n";
           return 0; #end timer    
         }
    });
}
[download]

What if you check the size of the file every 1.0 seconds or so, and grab it once it stops growing?

Or alternatively; scan in a tighter loop, but only grab a copy if the new file is larger than the last copy you made.

Yeah, stat'ing the file in a tight loop until no-growth is attained, is the way I would do it. But to be honest, I wanted to leave the code purposely ineffective so as not to teach people how to do DoS attacks. :-)

The node probably should be renamed "A way to save YouTube videos to disk".

I suspect the engineers at YouTube are probably already working on a way to stream to multiple cache files for a single video, so as to render this method obsolete. But, for the time being, it does offer a way to locally save youtube videos to disk, which is useful for archiving and offline viewing. I will post an example with a save button later, after I test how the file buffering goes.

I'm sure someone smarter than me out there, would know how to detect the progress of the on_load javascript request, as well as the cws file they send as a pre-loader, to the actual video. There seems to be alot of signals which one can tap into.

---

I'm not really a human, but I play one on earth.
Old Perl Programmer Haiku

For me, there may be no tommorrow. :-)

---

I'm not really a human, but I play one on earth.
Old Perl Programmer Haiku