Re: Domain based sorting

At 11:14 2011-08-18, LuKreme wrote:
>given Sean's CELAN_FROM and FROM_DOMAIN I tried:
>
>:0
>* FROM_DOMAIN ?? .*\/([^\.]+)

break it down:
         .*              match zero or more of anything
         \/              start match capture
         ([^\.]+)        match one or more of anything NOT a dot

that regexp is intended to grab the first domain token.

since the front isn't anchored, there's no real need for .* before 
the match operator (though if you drop them, owing to some parsing 
issues in procmail, you should have a () before the match 
trigger).  Now, if you KNOW there's a hostname (or, by your 
presumption that you won't be dealing with domain.co.uk style 
domains), you could count dots

Try:

* FROM_DOMAIN ?? ()\/[^\.]+\.[^\.]+$

That'll capture the last two nodes of a domain specification

         mail.domain.tld -> domain.tld
         domain.tld -> domain.tld
         foo.mail.domain.tld -> domain.tld
         host.demon.co.uk -> co.uk                       (!!!)

>which works if the from domain is "domain.tld" but fails if the 
>domain is "mail.domain.tld" (I get "mail")
>
>So, I tried to anchor it to the end:
>
>:0
>* FROM_DOMAIN ?? .*\/([^\.]+)\....?$
>{ ROOT_DOMAIN = $MATCH }
>
>but that always gives me "domain.tld" which confuses me because I 
>thought the () match gave that portion to $MATCH

No, EVERYTHING matching the regexp after \/ is stored in 
$MATCH.  Parens merely group (say for a zero or more, or a series of or's):
         (mail\.)?
         (foo|bar|baz)\.
(as examples of syntax, not as examples of anything you should be 
employing for your present task).

If you anchor to the end of the string, then you're assuring you'll 
match to the end (assuming the prior regexps match)

---
  Sean B. Straw / Professional Software Engineering

  Procmail disclaimer: <http://www.professional.org/procmail/disclaimer.html>
  Please DO NOT carbon me on list replies.  I'll get my copy from the list.