Pitfall in Bogofilter

[David Relson]

On Sun, 17 Sep 2006 21:22:42 +0200
Rudi Lippert wrote:

> Hi.
> 
> I know this is not the usual way to submit problems, but I'm not
> going to sign up at sourceforge or join the mailing list just to
> state one problem. So, there it goes:
> Since Bogofilter has changed the X-Bogosity header from 'Yes' to
> 'Spam', it is no more possible for simple mail clients to filter
> using a rule like 'X-Bogosity contains spam', since the header always
> contains the word 'spam' inside the word 'spamicity'. So, if someone
> uses this kind of filter, they will end up with no mail at all - as
> happened to me lately after I set up the filter in KMail without
> looking carefully enough.
> 
> This problem could be worked around by moving the spamicity histogram
> to a separate header named 'X-BogoHistogram' or something like that.
> 
> Cheers,
> 
> Rudi

Hello Rudi,

Evidently you've been using bogofilter for a long while and it's been a
long while since you updated.  The change from Yes/No/Unsure to
Spam/Ham/Unsure was made over 1.5 years ago -- on 9 Jan 2005 and
released in version 0.93.5 on 22 Jan 2005.

The X-Bogosity looks like:

   X-Bogosity: Spam, tests=bogofilter, spamicity=0.9876543, ...

and there are lots of ways to distinguish spam from ham.  Here are some
methods:

- If your mail client is case-sensitive, test for "Spam" rather than
"spam".  

- Have you tried including the comma, i.e. testing for "Spam,"?  

- I use Sylpheed-Claws as my MUA and include enough context to
distinguish between "spam" and "spamicity" by testing for "X-Bogosity:
Spam".

- Bogofilter's configuration file, i.e. bogofilter.cf, includes the
ability to format the X-Bogosity line and use any set of tags you
want.  Rather than Spam/Ham/Unsure you could use Good/Bad/Uncertain or
you could use John/Paul/Ringo or any other three words!

I've CC'd my reply to the bogofilter mailing list so that this
response will be in the public record and can be found by googling.

HTH,

David