Thanks for your really quick reply, that solved stay,
and I got the idea with the one liner.

Tero:
> > - if a page links to a site fetched earlier, link
> > would automatically be converted to a link
> > to local copy of that site (even when they
> > belong to different projects.

Xavier:
> would be much complex to implement (all
> websites structure should be parsed and stored in memory
> for lookup

Would it really need that much parsing? If the site is stored as
~
  websites
     http
        www.foo.com
           dir
              bar.html
              index.html
        www.bar.com
           index.html
           somedir
              test.html
Reading the site structure would be 'find' and replacing special
characters. Or, one could read the list of websites
(www.foo.com and www.bar.com here) and compare a link 
against filesystem when either of these sites is mentioned.

But this would just solve linking to sites we allready have.
There would need to be an option to check links archived in
websites, so that a link from older site to newer site 
would work.

ps. Email addreses in the forum are currently not obfuscated.