Here's an HTML::TokeParser solution. The output is kind of messy, but it works.

#!/usr/bin/perl -w

use strict;

use HTML::TokeParser;

my $filename = $ARGV[0] or die 'not enough arguments';
my $parser   = new HTML::TokeParser ($filename);

while (my $token = $parser->get_token()) 
{
        my ($type, $tag) = ($token->[0], $token->[1]);

        # We don't want <layer> or <iframe> tags
        next if $tag eq "layer" || $tag eq "iframe";

        # We can stop reading when we hit the nodelets section
        last if $type eq "C" && $tag eq "<!-- nodelets start here -->"
+;          

        # Print the token's text.  All the token types except T
        # have their text as their last element.  How annoying.
        if ($type eq "T")
        { 
                print $tag;
        } else {
                print $token->[$#{$token}];
        }
}

# Add a closing </table>.  Netscape won't display a table if the tags 
+aren't
# balanced.
print "</table>\n";

# EOF
[download]

-Matt