If you only want to remove certain parts you could take a to look at HTML::TreeBuilder and friends and use it to selectively pull out the elements you want from what's there. Alternately if your HTML is well formed enough you could use XML::Twig to do something similar.

Another source of inspiration might be to get the slashcode source and look at its comment filtering (seeing as that's what this sounds like you're trying to do).

#!/usr/bin/perl -w
use HTML::Scrubber;
use strict;

my $html = q[



    a   => link 
    br  => 

    b   =>  bold 
    u   =>  UNDERLINE 
];

# only allow the following tags
my $scrubber = HTML::Scrubber->new( allow => [ qw[ p b i u hr br ] ] );
print $scrubber->scrub($html);

__END__
Output:

    a   => link
    br  => 

    b   =>  bold 
    u   =>  UNDERLINE 

There's a module based on HTML::Parser for this on my pages (not released to CPAN) that allows you just that and more. It allows you to specify not just the list of tags to allow, but also the attributes. So not more unexpected onMouseOvers and onLoads ;-)

Just the module name is a bit silly ...