TODO list for Perl module WWW::PDAScraper
The user-agent should really be part of the object, I guess. That would
be neater.
And it should actually use WWW::Robot instead of LWP so it doesn't
hammer servers.
And we could either add arbitrary numbers of regexes for fixing up the
pages of sites which don't have a print-friendly version of the page, or
add a second level of parsing to find the print-friendly link, for sites
which don't have a logical relationship between the regular link and the
print-friendly.