in reply to Script is accurate 80% of time

I'm not sure how HTML::LinkExtor works, but if it doesn't lowercase the scheme for you, you need an i on the end of that regex, as in:
$attr_value->scheme =~ /\b(ftp|https?|file)\b/i # <---
Otherwise, you'd miss anything like
<A HREF="HTTP://www.perlmonks.org">
where the protocol is in caps.

Then again, maybe HTML::LinkExtor does that for you, and in that case I have no idea what the problem is :)
--
Mike

Edit: D'oh! Completely misread the question; I blame the lack of caffeine in my system this early :)