#!/usr/bin/perl
use XML::RSS

#
#   Little script to parse the tabbed movie sections off of hollywoodvideo.com.
#   (Now Playing, Coming Soon, Top 10, MVP
#
#   Requires:
#       XML::RSS
#       wget
#
#   Output:
#       For each section listed above, this script will produce <sectionname>.xml
#       (nowplaying.xml, mvp.xml, top10.xml, comingsoon.xml)
#
#   Original author:  Cameron Mallory
#
#   Released under http://creativecommons.org/licenses/by/2.0/
#   You are free:
#       * to copy, distribute, display, and perform the work
#       * to make derivative works
#       * to make commercial use of the work
#
$wget = "/usr/local/bin/wget";
$CHANNEL_WEBMASTER = 'your@email.com';
$ROOT_DIR = "/full/path/to/where/you/want/your/rss/to/go/";

##
##  Shouldn't need to change anything below here
##
@data = `$wget -q -O - http://www.hollywoodvideo.com`;
%movies = getMovieSections( @data );
generateRSS( %movies );


#
#   Use some RE's to parse out the movie sections
#
sub getMovieSections {
    my (@data) = @_;
    #
    #   Don't parse the whole file, so stop when we reach a magic line
    #
    $isFinished = 0; 
    
    foreach $line (@data){
    	chomp( $line );
    	if ( $line =~ /InitWidget\('condensed','uMLC4_rpt__ctl0_mc_rtc_139543',3, 3, 0\)/ ){
    		$cs = "top10";
    	} elsif (  $line =~ /InitWidget\('condensed','uMLC2_rpt__ctl0_mc_rtc_139465',3, 3, 0\)/ ){
    		$cs = "comingsoon";
    	} elsif ( $line =~ /InitWidget\('condensed','uMLC1_rpt__ctl0_mc_rtc_139446',2\.75, 3, 0\)/ ){
    		$cs = "nowplaying";
    	} elsif ( $line =~ /InitWidget\('condensed','uMLC3_rpt__ctl0_mc_rtc_140389',2\.5, 3, 0\)/ ){
    		$cs = "mvp";
    	}
    	if ( $line =~/More Top 10 Movies/ ){
    		$isFinished = 1;
    	}
    	if ( $isFinished == 0 && $cs ne "" && $line =~ /href="\/movies\/movie.aspx(.*?)<\/a>/ ){
    		$match = "href=\"http://www.hollywoodvideo.com\/movies\/movie.aspx$1<\/a>::";
    		$movies{ $cs } .= $match;
    	}
    }
    return %movies;
}

#
#   Generate RSS files for our movie sections
#
sub generateRSS {
    my (%movies ) = @_;

    chomp( $currDate = `date +'%a, %d %b %G %R:%S %Z'`);
    foreach $section ( keys %movies ){
        open(F, ">$ROOT_DIR/$section.xml");
        my $rss = new XML::RSS (version => '2.0');
        $rss->channel( title          => 'Hollywood Video - ' . $section,
                       link           => 'http://hollywoodvideo.com',
                       language       => 'en',
                       description    => 'Hollywood Video : RSS feeds',
                       #pubDate        => $currDate,
                       lastBuildDate  => $currDate,
                       webMaster      => $CHANNEL_WEBMASTER,
                       );
    
    	@sectionMovies = split("::", $movies{ $section });
    	foreach $showData ( @sectionMovies ){
    		$showData =~ /href="(.*?)">(.*?)<\/a>/;
    		$link = $1;
    		$name = $2;
            $name =~ s/&/&amp;/g; $name =~ s/</&lt;/g;$name =~ s/>/&gt;/g;
            
            $rss->add_item( title => $name, link => $link, description => $name, pubDate => $currDate );
    	}
    	print F $rss->as_string;
    	close(F);
    }
}