#!/usr/bin/perl
use strict;
#

my $VERSION = "0.3 beta";

# 2004-08-25 - Fixing some Getopts things. Trying to get multiple-dld to work
# 2004-08-24 - Push URL, when they seem to be clips (Clips can also have large/medium/small links)
#              Added Getopts - get help with "--help"
# 2004-08-23 - Started putting appletrailer and appletrailers together
# 2004-08-18 - Fix for Alfie
#              http://www.apple.com/trailers/paramount/alfie/
#              TWO Embed Tags (Full Screen Button and Trailer)
#              _fs.mov (not fs1.mov)
# 2004-08-12 - Started Modification to get all Trailers from
#              http://www.apple.com/trailers
# -- Version 0.1 --
# 2004-03-16 - First Version
#              Takes an URL as parameter ( like the .../large.html 
#              or .../moviename/) and finds the trailer to download

# This Script gets the latest Apple Trailers

# Created by Leonard Tulipan (micattack@gmx.net)
# Homepage: http://downlets.sourceforge.net/
# Released under the GPL - This is free software

# Needs #
# HTML::TreeBuilder
# Bundle::LWP
# Term::ProgressBar
# File::Basename
# To install you probably just need to execute:
# perl -e "use CPAN; install HTML::TreeBuilder; install Bundle::LWP;"
# perl -e "use CPAN; install Term::ProgressBar; install File::Basename;"
# Answer NO if cpan asks if you want to configure manually. This doesn't
# select your local mirror, but it also doesn't bother you with a lot of
# questions


# Todo #
# Use getopts and e.g.:
# -a ... All Trailers (Now Default)
# -m ... Movie Name (partial regex matching)
# -l ... List (a)ll (default), (e)xclusive, (S)tudio
# -k ... Ask before downloading
# -p ... Prefered Version (small - medium - large) - currently always the largest
# -s ... Download from that Studio
 

# Set to 1 to see debug output
# Can now be set with --debug command line switch
my $debug=0;

# For immediate output of progress bar
$|++;

# HTML Treebuilder is needed for HTML File Syntax checking
use HTML::TreeBuilder;

# LWP UserAgent is used for downloading the HTML Files from the web
# make sure we have the modules we need, else die peacefully.
# perl -e "use CPAN; install Bundle::LWP;"
eval("use LWP 5.6.9;"); die "[err] LWP 5.6.9 or greater required.\n" if $@;

# Progress Bar
eval("use Term::ProgressBar;"); # prevent word-wrapping.
die "[err] Term::ProgressBar not installed.\n" if $@;

# For Command Line Options
use Getopt::Long;

# For dirname
use File::Basename;

my $version; my $help; my @url; my $ask; my $nobig;
GetOptions( 'version' => \$version, 'debug' => \$debug,
            'help' => \$help, 'url=s' => \@url, 
            'ask|k' => \$ask, 'nobig' => \$nobig);
 @url  = split(/,/,join(',',@url));


if($version || $help)
 {
  print " Appletrailers Version $VERSION\n";
  print " Created by Leonard Tulipan (micattack\@gmx.net)\n";
  print " Homepage: http://downlets.sourceforge.net/\n";
  print " Released under the GPL - This is free software\n";
  if($help)
  {
   print <<EndHelp;

 Help:
  --version       ... Print Version and exit (see it above, btw)
  --help/-h       ... This Help
  --url/-u <URL>  ... Download Trailer found on this URL (Specify the HTML not the MOV File
                      e.g: -f http://www.apple.com/trailers/paramount/team_america/
                      You can specify more than one url with a comma separator
                      or simply specify multiple -u on the command line
  --ask/-k        ... Ask before Downloading a Trailer
  --debug/-d      ... Turn Debugging On
EndHelp
  } 
  exit;
 } 

print " Debugging is now on\n" if $debug;

# Start a new instance from UserAgent
my $ua = LWP::UserAgent->new;
# Set User Agent to iTunes
$ua->agent('User-Agent: iTunes/4.6 (Macintosh; U; PPC Mac OS X 10.3)');

# We save the trailerpages ../large.html or the like in this array
my @trailerpages;


if(!@url)
{ # Get All Trailers, if no url is specified on the command line

 # Get the Trailer HTML Page 
 my $request = HTTP::Request->new('GET', "http://www.apple.com/trailers/");
 # The Itunes xml URL:
 #my $request = HTTP::Request->new('GET', "www.apple.com/moviesxml/h/index.xml");

 # Save the HTML in $response
 my $response = $ua->request($request);

 # Start a new instance of TreeBuilder
 my $tree = HTML::TreeBuilder->new();

 # Parse the content of $response (Trailerpage)
 $tree->parse($response->content);

 # Print out the content, if there seems to be something wrong (new html code, etc)
 #print $response->content;

 # Get the Trailer-Rows - Look for <div class=rowtext> and put each of those as a value of an array
 my @trailer_rows = $tree->look_down("_tag", "div", sub{ $_[0]->attr('class') eq "rowtext"});
 # Look for the exclusive Trailers inside the first <div class=rowtext>
 # Look for all <div class=link> Tags
 my @exclusive = $trailer_rows[0]->look_down("_tag", "div", sub{ $_[0]->attr('class') eq "link"});

 # Get all <a href> Tags and save those links in trailers array
 my $link;
 my @trailers;
 for(my $cnt=0; $cnt < $#exclusive; $cnt++) {
  $link = $exclusive[$cnt]->look_down("_tag", "a");
  push @trailers, $link->attr('href');
 }

 # Get the Studio Trailer List
 # This is the second <div class=rowtext> element
 # Inside look for all <li> Tags
 # And save the contained <a href> links into the trailers array
 my @studio = $trailer_rows[1]->look_down("_tag", "li");
 for(my $cnt=0; $cnt < $#studio; $cnt++) {
  $link = $studio[$cnt]->look_down("_tag", "a");
  push @trailers, $link->attr('href');
 }


 # Only get the Uniqe occurences (Trailes can be listed as exclusive AND studio trailers
 # Thank you groups.google.com for that :-)
 my @uniq;
 my %uniq;
 @uniq{@trailers} = ();
 @trailers= sort keys %uniq;

 # This would print out all Trailers, but that's too much debuging output
 #foreach (@trailers) {
 #   print "Trailer: ". $_ ."\n" if $debug;
 #}
 print "Checking ". ($#trailers + 1) ." Trailers\n" if $debug;


 foreach(@trailers) 
 { # Check on all trailerpages
   # Make absolute paths, so e.g:
   # "sony/spiderman2" will be made into
   # "/trailers/sony/spiderman2"
   # But leave paths which already start with "/" intact
  if(! /^\//)
  { $_ = "/trailers/" . $_; } 
  my $current = $_;
  # Kill trailing Slash
  my $currentdir = substr($current,0, length($current)-1);
  if(!($current =~ /\/$/))
  { # Only use dirname if URL does not end with "/"
    # This seems to be a bug in dirname, so we first fill it with the above value
    # without trailing slash, and only use dirname, if the trailer does not have a
    # trailing slash to begin with
    $currentdir = dirname($current);
  }
  print "Looking at: ". $currentdir ."\n" if $debug;

  # Get the Trailer HTML Page
  my $request = HTTP::Request->new('GET', "http://www.apple.com". $_);
 
  my $tr_resp = $ua->request($request); 

  # Start a new instance of TreeBuilder
  my $tr_tree = HTML::TreeBuilder->new();
 
  # Parse the content of $response (Trailerpage)
  $tr_tree->parse($tr_resp->content);
 
  # Sometimes we get the 404 Page with the title-tag, then we'd better not analyse that page
  if($tr_tree->look_down("_tag", "title", sub{ $_[0]->as_text eq "Apple - Page Not Found"}))
   { print "404 Page\n" if $debug }
  else { 

   # Get all links
   my @links = $tr_tree->look_down("_tag", "a"); 
 
   my $found = 0;
   for(my $cnt=0;$cnt <= $#links;$cnt++) {
    # Trailer links have a rollover, so onmouseout is set most of the time
    if($links[$cnt]->attr('onmouseout')) {
     # lg     ... e.g: Alien Vs Predator
     # large  ... Incredibles
     # bigger ...
     # 480    ... Maria Full Of Grace
     if($links[$cnt]->attr('href') =~ /(lg|large|high|bigger|480)/i && !($links[$cnt]->attr('href') =~ /^http/)  ) {
      $found = 1;
      print "Pushing URL: ". $currentdir  . "/"  . $links[$cnt]->attr('href') ."\n" if $debug;
      push @trailerpages, $currentdir . "/"  . $links[$cnt]->attr('href'); 
     }
    }
    # If no rollover is specified
    if($links[$cnt]->attr('href') =~ /(lg|large|high|bigger|480)/i && !($links[$cnt]->attr('href') =~ /^http/) ) {
     $found = 1;
     print "Pushing URL: ". $currentdir  . "/"  . $links[$cnt]->attr('href') ."\n" if $debug;
     push @trailerpages, $currentdir  . "/"  . $links[$cnt]->attr('href'); 
    }

    if( $links[$cnt]->attr('href') =~ /index/i && $found == 0) {
     # Not a Trailer, don't do more, but still could be, see area test below
     $found = 2;
     #print "Probably not a Trailer\n" if $debug;
    }
    # For Text Featurettes and other offsite stuff (www.parentalguide.org - which matches "lg" test)
    if(! $links[$cnt]->attr('href') =~ /^http/i && ! ($links[$cnt]->attr('href') =~ /^index/i) ) {
     $found = 1;
     print "Pushing URL: ". $currentdir  . "/"  . $links[$cnt]->attr('href') ."\n" if $debug;
     push @trailerpages, $currentdir . "/"  . $links[$cnt]->attr('href');
    } 
 
  }
  # We have an embed tag on the page, could be the trailer
  if($tr_tree->look_down("_tag", "embed") && $found ne 1) {
    # Is Probably alread the Trailer Page
    print "Pushing URL: ". $currentdir  . "/\n"  if $debug;
    push @trailerpages, $currentdir  . "/";
    $found = 1;
  }

  if($found ne 1) {      
   # Check area Links (Image Map)
   my @area;
   @area = $tr_tree->look_down("_tag", "area");
       
   for(my $cnt=0;$cnt <= $#area;$cnt++) {
    my $href = $area[$cnt]->attr('href'); 
    if($href  =~ /(lg|large|high|bigger|480)/i ) {
     if( !($href  =~ /^http/) )
     {    
      $found = 1;
      print "Pushing URL: ". $currentdir . "/"  . $href . "\n" if $debug;
      push @trailerpages, $currentdir . "/"  . $href;
     }
    }
   }  
  }

  if($found eq 0) {      
   # print "------------------------------\nhttp://www.apple.com" . $current . "\n------------------------------\n" if $debug;
   # print "NONE FOUND: \n" if $debug;
    for(my $cnt=0;$cnt <= $#links;$cnt++) {
     #print "-- ". $links[$cnt]->attr('href') . " --\n" if (not ($links[$cnt]->attr('href') =~ /^http/)) && $debug;
     # Links without http could still be clips pages, so we add them to the trailers array to be tested in this loop we're currently in
     if (! ($links[$cnt]->attr('href') =~ /^http/)) {
      print "Adding to Trailers: " . $current . $links[$cnt]->attr('href') . "\n" if $debug;
      push @trailers, $current . $links[$cnt]->attr('href'); 
     }
    }
    
  }
 } 
 }
} # end if(!url)
 else {
  # Just one url, check that one
  print "Pushing ". ($#url+1) . " URLs\n" if $debug;
  push @trailerpages, @url;
}

print "Getting Trailers from ". ($#trailerpages + 1) ." Trailerpages\n" if $debug;

# Make trailerpages unique
my @uniq;
my %uniq;
@uniq{@trailerpages} = ();
@trailerpages= sort keys %uniq;


print "Getting Trailers from ". ($#trailerpages + 1) ." Trailerpages\n" if $debug;

foreach (@trailerpages)
{
 print "Getting: " if $debug;
 print if $debug;
 print "\n" if $debug;

 if(! /^http/) { $_ = "http://www.apple.com" . $_; }

 my $request;
 my $response;
 my $tree;
 my @mov;
 my $movuri;
 my $test;
 my $result;
 my $remote_headers;
 my $total_size;
 my $fullmovuri;
 my $progress;

# Get the Trailer HTML Page (large.html or the like)
$request = HTTP::Request->new('GET', $_);

# Save the HTML in $response
$response = $ua->request($request);

# Start a new instance of TreeBuilder
$tree = HTML::TreeBuilder->new();

# Parse the content of $response (Trailerpage)
$tree->parse($response->content);

#my $mov = $tree->look_down("_tag", "embed", sub{ $_[0]->attr('pluginspage') eq "http://www.apple.com/quicktime/download/indext.html"} , sub{ $_[0]->attr('type') eq 'video/quicktime'});
@mov = $tree->look_down("_tag", "embed", sub{ $_[0]->attr('pluginspage') eq "http://www.apple.com/quicktime/download/indext.html"});

if($#mov >= 0)
{
 $movuri = $mov[$#mov]->attr('src');
 $test = basename($movuri);
 if ($test =~ /^qt[\/\-\_\d\w]*\.mov$/ || $movuri =~ /\/image/)
   { $movuri = $mov[$#mov]->attr('href'); }
 #print $movuri . "\n";
 $tree->delete();  

 #print $movuri;

 $ua = LWP::UserAgent->new( );
 $result = $ua->head($movuri);
 $remote_headers = $result->headers;
 $total_size = $remote_headers->content_length;
 if($total_size > 1000000)
 { $fullmovuri = $movuri; }
 else
 {
   # Get the MOV File
   $request = HTTP::Request->new('GET', $movuri);
   # Save the HTML in $response
   $response = $ua->request($request);

   #if($response->content =~ /\x20\x00\x00\x00(.*0\.mov)\x00/)
   if($response->content =~ /\x20\x00\x00\x00.?([\/\-\_\d\w]*\.mov)\x00/)
   {  $fullmovuri = $1;
      # Just the Directory
      $movuri = dirname($movuri);
      # add a Slash if not with Filename
      $movuri .= "/" if (!($fullmovuri =~ /^\//));
      # make full uri with dirname / filename
      $fullmovuri = $movuri . $fullmovuri;
   }
   #print "full: -$fullmovuri- \n";
  }

if( $fullmovuri ne "")
{
  my $final_data_length = undef;
  my $progress; # progress bar object.
  my $next_update = 0; # reduce ProgressBar use.

  if(! $nobig)
  {
   if( $fullmovuri =~ /(.*)240\.mov$/)
   {
    my $bigtest = $1 . "480.mov";
    print "Testing: $bigtest\n" if $debug;
    my $ua = LWP::UserAgent->new( );
    my $result = $ua->head($bigtest);
    my $remote_headers = $result->headers;
    $total_size = $remote_headers->content_length;
    if($total_size > 1000000)
    { $fullmovuri = $bigtest }
   }
   # Itunes Full Screen
   if( $fullmovuri =~ /(.*)m480\.mov$/)
   {
    my $bigtest = $1 . "ifs.mov";
    print "Testing: $bigtest\n" if $debug;
    my $ua = LWP::UserAgent->new( );
    my $result = $ua->head($bigtest);
    my $remote_headers = $result->headers;
    $total_size = $remote_headers->content_length;
    if($total_size > 1000000)
     { $fullmovuri = $bigtest }
   }
   # Standard Full Screen
   if( $fullmovuri =~ /(.*)m480\.mov$/)
   {
    my $bigtest = $1 . "fs1.mov";
    print "Testing: $bigtest\n" if $debug;
    my $ua = LWP::UserAgent->new( );
    my $result = $ua->head($bigtest);
    my $remote_headers = $result->headers;
    $total_size = $remote_headers->content_length;
    if($total_size > 1000000)
     { $fullmovuri = $bigtest }
   }
   # Standard Full Screen
   if( $fullmovuri =~ /(.*)m480\.mov$/)
   {
    my $bigtest = $1 . "fs.mov";
    print "Testing: $bigtest\n" if $debug;
    my $ua = LWP::UserAgent->new( );
    my $result = $ua->head($bigtest);
    my $remote_headers = $result->headers;
    $total_size = $remote_headers->content_length;
    if($total_size > 1000000)
     { $fullmovuri = $bigtest }
   }
  } # end ! $nobig

  my $url = $fullmovuri;
  
  if( -e basename($url)) {
   print "Trailer ". substr(basename($url), 0, 40) ." already exists\n" if $debug;
  } else {
  print "Downloading $url ", substr(basename($url), 0, 40);
  my $answer = "";
  if($ask) {
   print " $url OK? (Y/n): ";
   $answer = <STDIN>;
   chomp($answer);
  }
  print "\nAnswer: \"$answer\"\n" if $debug;

  open (MOVFILE, "> " . basename($url)) || die $!;
  if(! ($answer =~ /^n/i) ) {  
   # create a new useragent and download the actual URL.
   # all the data gets thrown into $final_data, which
   # the callback subroutine appends to.
   #my $ua = LWP::UserAgent->new( );
   $result = $ua->head($url);
   $remote_headers = $result->headers;
   $total_size = $remote_headers->content_length;
   printf( " ( %.2fMB )... \n", $total_size/1024/1024 );
   # initialize our progress bar.
   $final_data_length = undef;
   $next_update = 0;
   print "\nf: " . $final_data_length . "\n";

   $progress = Term::ProgressBar->new({count => $total_size, ETA => 'linear'});
   $progress->minor(0); # turns off the floating asterisks.
   $progress->max_update_rate(1); # only relevant when ETA is used.
   $response = $ua->get($url, ':content_cb' => \&callback, );
   # top off the progress bar.
   $progress->update($total_size);
   print "\nf: " . $final_data_length . "\n";

     # per chunk.
     sub callback {
      my ($data, $response, $protocol) = @_;
      print MOVFILE $data;
      $final_data_length += length($data);
      #print "\n" . $response. " " . length($data) . "\n";
      # reduce usage, as per example 3 in POD.
      $next_update = $progress->update($final_data_length)
      if $final_data_length >= $next_update;
     }

   $progress = undef;

  } else {
   print "Empty File created\n";
  }
  close (MOVFILE);

 }# $mov != 0


}

}
}