while ( my $raw = <DATA> ) {
    next unless $raw;
    my $uri = URI->new($raw);
    print $uri->host, $/;
    my ( $bad_assumption_helper ) =
        $uri->host =~ /([^.]+)\.(?:\w\w\w|\w\w\.\w\w)\z/;
    print "\t", $bad_assumption_helper, "\n";
}

__DATA__
http://shopping.rediff.com/
http://sportsillustrated.cnn.com/?cnn=yes
http://edition.cnn.com/
http://en.wikipedia.org/
http://search.cpan.org/
http://news.bbc.co.uk/
[download]

You should heed the snark of the JavaFan though. This isn't a perfect solution and the domain name includes the TLD. A bomb-proof approach to stripping it would require quite a bit more code and probably some lookups. Anything wherein you are required to strip it probably has broken assumptions.