Re: Domain based sorting

At 11:31 2011-08-18, LuKreme wrote:
>LuKreme <[email protected]> squawked out on Thursday 18-Aug-2011@12:14:46
> > So, I started to think (dangerous, I know) and I searched and 
> found Sean's post from a few of years ago about dealing with 
> getting domains in domain.co.uk sorts of situations:
>
>And a few minutes later I found Dan's post in the same thread with 
>(trimmed down to just the part I want)
>
>TLDREGEX = ([cC][oO][.][^.][^.]|[^.]+)

Doesn't need to be case sensitive unless someone explicitly makes a 
recipe case sensitive by specifying the 'D' flag.  The following is 
more succinct, and accomplishes the same thing within the example recipe:

TLDREGEX = (co[.][^.][^.]|[^.]+)

Note that the [.] expression might more commonly be expressed as 
\.  but one would have to double-escape it to \\. for the slash to 
appear in the resulting regexp string, so character classing it is in 
fact clearer.

># Get the domain name
>:0
>* $ FQDN  ?? ()\/[^.]+[.]$TLDREGEX^^
>* MATCH ?? ^^\/[^.]+
>{ DOMPART = $MATCH }
>
>This works perfectly as far as I can tell.

Though the second condition line drops the TLD portion(s) -- this 
will grab the "domain" from "domain.tld", "mail.domain.tld", or 
"mail.domain.co.uk".  However, the TLDREGEX is a 'co.xx' specific 
expression -- it'll trip up on something such as 'k12.ca.us' (and 
there are many variations on that), but will get "ca" for a domain in 
the "k12.ca.us" heirarchy for example (California schools, 
http://www.ed-data.k12.ca.us/), and ca.us is used for municipalities 
within the state.  nv.us is nevada, and predictably, other states use 
the same syntax.  then there's ca.gov - with a host of subdomains 
including some municipalities, and agencies, cdfa.ca.gov, etc.

The UK has "org.uk" and "net.uk" as well.

Admittedly, you're not likely to be ordering anything from .ca.us and 
the like, but in the context of parsing out a domain, there are many 
issues raised.

Considering the ICANN decision to open up the TLD naming to pretty 
much anything, some thought needs to be put into how domains are 
parsed - there's sure to be a LOT logic that will break.

---
  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.