Bfproxy v0.3

[Tom Anderson]

I've released a new version of my training-by-email program, bfproxy. 
The notable addition is address-line options.  You may now send an email
of the form username+bfpID-command at yourdomain where "command" is one of
C, R, n, or s.  

C will use the x-bogosity header to determine the initial classification
and then unregister the original and reregister the opposite
classification (bogofilter -Ns,-Sn).  This is useful if you use the -u
flag to auto-classify emails and need to Correct that classification.

R is similar, except that it doesn't unregister.  So, if you don't use
-u, and you just put all of your false positives and false negatives in
a single "error" folder, you can use the R option to register all of
them as the opposite of their original classification (-s, -n).

The n and s flags are exactly as they are in bogofilter.  They will
register all attached emails as hams or spams respectively.  This way,
if you seperate out your unsures, false positives, and false negatives
into ham and spam folders, you can register each of them all at once.

Another change I made is that I've added some per-email information in
the output.  This is to verify which ones were or were not able to be
processed, and also provides a before and after spamicity value.  A
future addition may include the ability to train to exhaustion those
emails which are still not moved sufficiently to one side or the other. 
If anyone thinks that would be a good idea, please let me know.

Finally, I've added some benchmarking data at the end of the output.

You can download it here:
http://www.orderamidchaos.com/bogofilter/bfproxy

Feedback is appreciated.

Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://www.bogofilter.org/pipermail/bogofilter/attachments/20040207/ab52a69c/attachment.sig>