[Spambayes] Getting Evidence (Was: AssertionError: hamcount <= nham)

[Gerrit Holl]

Hi,

Skip Montanaro wrote:
>     Gerrit> I use it to check why a certain message has been classified as
>     Gerrit> spam or ham.  When I don't understand why a message is not
>     Gerrit> correctly classified, I launch sb_server and ask it to give the
>     Gerrit> full explanation (classify). Is there a script that simply does
>     Gerrit> this but uses stdout instead of a web server?
> 
> Sure.  Just spit the message back through sb_filter.py with the
> Headers.include_evidence option set to True.  If you're running from the CVS
> version of SpamBayes you run it like
> 
>     sb_filter.py -o Headers:include_evidence:True < msgfile

Yes.

But this is a 'short' version.

Is it possible to get a long version, like the webserver is giving, without
running the webserver? By long version, I mean at least two things: every
clue on a separate line, and every clue contains information on how many
spam messages contained this word and how many ham messages did. Of course,
if there is no script producing this output, it would be possible to write
one. I earlier succeeded in writing a script telling me how many spams
and how many hams I have. I did this by simply looking at the source code.
It is not as simple for what I want now. I have not found a simply
function/method producing an object containing all clues which can be
formatted to a table in a few lines of code. Is there such a function?

(is this still ontopic for this list or do questions on spambayes' classes
 belong in the spambayes-dev mailing list?)

good night to all europeans :P,
Gerrit.

-- 
122. If any one give another silver, gold, or anything else to keep, he
shall show everything to some witness, draw up a contract, and then hand
it over for safe keeping.
          -- 1780 BC, Hammurabi, Code of Law
-- 
Asperger Syndroom - een persoonlijke benadering:
	http://people.nl.linux.org/~gerrit/
Kom in verzet tegen dit kabinet:
	http://www.sp.nl/