This is the regex I use to recognize URLs in plain text:
use Regexp::Common qw(URI);
my $re = qr/$RE{URI}{HTTP}(?:#[\w_%:-]+)?(?<![.,])/;
The negative look-ahead takes care that in http://example.com/something, the comma isn't treated as part of the URL (it is a valid part of the URL, but usually you don't want to include it nevertheless).
For adding the trailing / and extracting the host name you need a bit more logic, for which I'm too lazy right now to write. I think that Regexp::Common has an option to capture the domain name somehow, but I haven't investigated in that either.
I hope this is of interest nonetheless.