Feature request: Can spamd remove markup as well?

["C. Bensend" <be...@bennyvision.com>]

Hey folks,

   I have a server that needs to strip off SpamAssassin markup.  That's all
it needs to do - incoming email has already been tagged by another
mailserver as being spam.  It just needs to strip the tags, and deliver
the email to a maildir.

   Unfortunately, this server needs to start up spamassassin upon each
email delivery, to execute a 'spamassasin --remove-markup'.  This is VERY
expensive, resources-wise.  To deliver ~1500 messages, it took about two
and a half hours.  Not so much.  :(

   The front-end mailserver runs spamd as a daemon, and that speeds things
up _substantially_.  I'd love to do the same sort of thing to strip the
SpamAssassin markup out of the incoming messages, but I don't see any such
feature in spamd or spamc.  Does such a thing exist?

   And if not, could it be added?  :)  Thoughts?  Flames?  Cluebats?

Thanks much,

Benny


-- 
"I can't believe it's not carp!"                    -- MXC

Re: Feature request: Can spamd remove markup as well?

["C. Bensend" <be...@bennyvision.com>]

>    The front-end mailserver runs spamd as a daemon, and that speeds things
> up _substantially_.  I'd love to do the same sort of thing to strip the
> SpamAssassin markup out of the incoming messages, but I don't see any such
> feature in spamd or spamc.  Does such a thing exist?

Sorry, forgot the details.  This is SpamAssassin 2.63 running on an
OpenBSD 3.5-STABLE machine.  It's a Celeron 533MHz, 256M RAM, etc.

Benny


-- 
"I can't believe it's not carp!"                    -- MXC

Re: Feature request: Can spamd remove markup as well?

[Theo Van Dinter <fe...@kluge.net>]

On Sun, May 02, 2004 at 11:07:59PM -0500, C. Bensend wrote:
> For example, on this particular system I'm working on (which is VERY VERY
> low volume), I can strip the spamassassin tags out of emails at the rate
> of about /1 email every eight seconds/.  I can parse each email and stuff
> them into the database at the rate of /13 emails per second/.  And yes,
> that's using perl.

You'll want to check out the 3.0 "spamassassin" script, it'll batch
process emails now, so you can do "spamassassin -d --mbox mbox |
load-to-db".

In the mean time, you can check out something like
http://www.kluge.net/~felicity/random/handlespam.txt

-- 
Randomly Generated Tagline:
Tactical?  TACTICAL!?!?  Hey, buddy, we went from kilotons to megatons
 several minutes ago.  We don't need no stinkin' tactical nukes.
 (By the way, do you have change for 10 million people?) --lwall

Re: Feature request: Can spamd remove markup as well?

["C. Bensend" <be...@bennyvision.com>]

> What's the MTA?

I'm sorry...  net-qmail (ie, qmail-1.03 with minimal patches).

> It might be simpler to do this outside of SA.
>
> If you're using postfix, e.g. you can use a content_filter
> to send all your mail through a script that does essentially
>
> grep -v "^X-Spam"
>
> and reinject back into your mailstream.

Personally, I'm kind of surprised that spamd/spamc don't have this
capability already - the startup costs for spamassassin are prohibitive
on many systems that process a lot of mail.  Granted, I'm not using
spamassassin like most folks would _on this particular host_.

For example, on this particular system I'm working on (which is VERY VERY
low volume), I can strip the spamassassin tags out of emails at the rate
of about /1 email every eight seconds/.  I can parse each email and stuff
them into the database at the rate of /13 emails per second/.  And yes,
that's using perl.

If any of you spamassassin developers are willing to work on this, I am
certainly willing to make sure you have pizza, beer, or whatever to help
you along.  :)

Benny


-- 
"I can't believe it's not carp!"                    -- MXC

Re: Feature request: Can spamd remove markup as well?

[Ari Jort <ar...@speakeasy.net>]

On Sat, May 01, 2004 at 11:49:21PM -0500, C. Bensend wrote:
> 
>    I have a server that needs to strip off SpamAssassin markup.  That's all
> it needs to do - incoming email has already been tagged by another
> mailserver as being spam.  It just needs to strip the tags, and deliver
> the email to a maildir.

What's the MTA?

It might be simpler to do this outside of SA.

If you're using postfix, e.g. you can use a content_filter
to send all your mail through a script that does essentially

grep -v "^X-Spam"

and reinject back into your mailstream.

ari

>    Unfortunately, this server needs to start up spamassassin upon each
> email delivery, to execute a 'spamassasin --remove-markup'.  This is VERY
> expensive, resources-wise.  To deliver ~1500 messages, it took about two
> and a half hours.  Not so much.  :(
> 
>    The front-end mailserver runs spamd as a daemon, and that speeds things
> up _substantially_.  I'd love to do the same sort of thing to strip the
> SpamAssassin markup out of the incoming messages, but I don't see any such
> feature in spamd or spamc.  Does such a thing exist?

-- 
Ari Jort
arijort@speakeasy.net