[Evolution-hackers] SpamBayes

[Rasmus Toftdahl Olesen <halfdan halfdans net>]

Hi There

I am the author of the SpamBayes using spam filtering plugin for
Evolution, so far i seem to have things working nicely (i based my code
on the Bogofilter plugin).

But there is one thing i am missing, i would like to be able to track
which messages have been marked as what - in SpamBayes you need to
untrain a wrongly tagged message before you can train on it tagged the
right way.

In other words, a message can have three states: not rated, junk or
not-junk. If the user wishes to train a message as e.g. junk i need to
know if it has previously been trained as not-junk (so i can untrain it)
before i train it as junk.

Rather than maintaining a database of "seen" messages on my own, i
thought it would be nice (as well as a good debugging feature) to be
able to add a custom header (X-Evolution-Spambayes-Plugin) to each
message as it is seen by the SpamBayes plugin, this would allow me to
see whether the message is new or not.

Is this possible?

I have tried using camel_medium_set_header and camel_medium_get_header
on the CamelMimeMessage passed to the "junk", "non-junk" hooks, but that
doesn't seem to work - but doesn't crash Evolution either.

Thanks for any insight you can provide.

My plugin is available form here:
http://halfdans.net/wiki.py/EvolutionSpamBayesPlugin

-- 
Rasmus Toftdahl Olesen <halfdan halfdans net>
halfdans.net