[Spambayes] pop3 culler

[Andrew Dalke]

Richie Hindle:
> My approach would be to use the SpamBayes POP3 proxy (sb_server.py) 
> as-is,
> and write a script that used Python's POP3 client module to connect,
> retrieve the emails, look at the X-Spambayes-Classification header, and
> delete the ones marked as spam.

That's a cute way to do it.  It feels like a hack, because
of the extra server->proxy->culler network step, but OTOH
the machine isn't doing anything else.

> If I weren't storing the messages, I'd set a high spam threshold.

Yeah, and whitelist known good email sources.

Bob Coe:
> If Andrew's PC has a full-time connection to the Internet, he might be 
> able to do it without even writing a script. [...]
> Then assuming a firewall penetrable by an authenticated client, Andrew 
> would connect his laptop to his PC, download what he wants, leave what 
> he prefers to read later, and delete the rest.

It's a DSL connection, but I could get one of those dyndns
entries.  There are a few problems.  I would need to open up
my firewall a smidgeon, which I'm not happy about.  It would
also mean I can't use the web interface to my mail (eg, from
a cybercafe) or ssh (my client's site doesn't allow connections
to web-based email because downloads don't go through their
virus detectors).  All in all, too complicated for my tastes.
Richie's seems much simpler.

Tony Meyer:
> The only thing I would add to Richie's comments is that if your 
> connection
> from the NT box is slow or expensive, then you could get it to do this 
> based
> on header analysis only

DSL.  Bandwidth isn't a problem for this,

My take-home message ... wait, I am at home! .. is that no
such tool already exists.  Okay, I'll play around with it
for a bit and post what results.

					Andrew
					dalke at dalkescientific.com