gnus & spamassassin

gnus & spamassassin

[Graham Murray]

Are there any plans to incorporate an interface into SpamAssassin (or
other spam detectors) into gnus? 

Spamassassin will mark mail which it considers spam (which is, of
course, not infallible) and gnus can split this into a 'spam'
group. However it would be nice if gnus could do 2 additional things:-

1) Pass mail in Spam & Ham groups (identified by the configuration in
   spam.el) through the spamassassin bayes learner (sa-learn-spam,
   sa-learn-nonspam) on group exit - in the same way as spam.el will
   pass it through bogfilter and ifile.

2) Spamassassin (as from V2.50) 'mangles' spam by creating an email
   report with the original (suspect) mail as a MIME (message/rfc822)
   attachment. It would be useful if gnus had a feature to 'decode and
   move' (pipe the mail through 'spamassassin -d' then action either
   Bm or Br) manually identified false positives to a non-spam group.

Re: gnus & spamassassin

[Kai Großjohann]

Graham Murray <graham@gmurray.org.uk> writes:

> Are there any plans to incorporate an interface into SpamAssassin (or
> other spam detectors) into gnus? 

Investigate the corresponding chapter in the docs.
-- 
Ambibibentists unite!

Re: gnus & spamassassin

[Ted Zlatanov]

On Sun, 26 Jan 2003, graham@gmurray.org.uk wrote:
> Are there any plans to incorporate an interface into SpamAssassin
> (or other spam detectors) into gnus?
> 
> Spamassassin will mark mail which it considers spam (which is, of
> course, not infallible) and gnus can split this into a 'spam'
> group. 

This can be done automatically if you write a spam-check-spamassassin
function (see spam-check-bogofilter for an example).  It would
probably be easier for the users to set a boolean than to manually
modify the split rules.

> However it would be nice if gnus could do 2 additional things:-
> 
> 1) Pass mail in Spam & Ham groups (identified by the configuration
>    in spam.el) through the spamassassin bayes learner
>    (sa-learn-spam, sa-learn-nonspam) on group exit - in the same way
>    as spam.el will pass it through bogfilter and ifile.

Can you or someone else provide the relevant functions?  See the
bogofilter spam/ham processing functions in spam.el, I'm sure
spamassassin won't be too different.  Basically,
spam-assassin-register-spam-routine and
spam-assassin-register-ham-routine are needed.

I don't mind doing it, but it would probably be better if someone who
uses spamassassin could do a sanity check on the parameters - they are
much more likely to get the functionality right.  Even pseudocode
would be helpful, I can do the rest.

> 2) Spamassassin (as from V2.50) 'mangles' spam by creating an email
>    report with the original (suspect) mail as a MIME
>    (message/rfc822) attachment. It would be useful if gnus had a
>    feature to 'decode and move' (pipe the mail through 'spamassassin
>    -d' then action either Bm or Br) manually identified false
>    positives to a non-spam group.

That's possible as an option, I think.  spam-ham-move-routine is the
function I need to modify; does spam-assassin-demangle-false-positives
sound good as an option name?

Ted