[clear] pyCSV

[Richard Kay]

I have implemented CSV processing on my
incoming email by importing 
from # csv-0.3x.py  David MacQuigg (c) 8/19/05 
into Tagspam version 0.4 

http://copsewood.net/tagspam/index.html

Feedback by sending me pairs of messages which should
pass and fail would be much appreciated.

The page includes a download which includes
some installation notes. Those wanting to use Tagspam 
will also need to read the Python source code.

Tagspam now implements CSV processing in addition 
to the DNSBL and DNSWL tests provided for some time 
by Tagspam. Feedback on how to make this
program more generic would be appreciated. This
would now also require a reliable means of obtaining the
HELO domain from received headers, or elsewhere.
The approach currently taken appears to work with 
my installation of Sendmail, but I have no idea
whether it would work elsewhere.

Initially I do not think CSV to be mature enough
for me to feel safe to use it at the MTA rejection stage. 
For experimental purposes I consider this technology
to be useful for filtering - if Tagspam detects a CSV fail
relevant messages can be sent to a spam folder. This
enables domain owners to publish CSV SRV records
enabling Tagspam users to suitably tag and filter
email spoofing their HELO domains. 

Clearly if and when CSV matures it will become possible to
confidently reject CSV failures at the MTA
stage. This would require a Milter
which handles CSV and installs readily using the package 
management (e.g. RPM or apt facility) on a production server with 
security support using a stable software distribution.

For my purposes the advantage of filtering is that
this can be safely done using a more experimental
setup, making a greater range of anti-spam experiments
and testing possible.

Richard Kay.