[SOLVED] Re: Accidentally misread a message into sa-learn. (Solution: Use sa-learn --forget)

[JKL <ju...@klunky.co.uk>]

On 03/19/2011 10:48 PM, JKL wrote:
> Good evening, and good weekend,
>
> I made a silly mistake in the dB.
>
> *** What happened
>  Whilst experimenting with some Logwatch options, I increased the
> granularity.  This meant the Email report to me contained content that
> correctly tripped SA.  ).  Sadly, this tripped me over & I though SA
> needed some training so I misguidedly fed the email into sa-learn.  This
> email was neither ham nor spam and should not have been categorised.
> Oh, the things one does on a Saturday night.
>
> *** The question
>  How can I remove the offending tokens from SA?    (If at all possible).
>
>
> *** Command used & result
> # sa-learn  --ham cur/
> Learned tokens from 1 message(s) (102 message(s) examined)
>
> *** The SA message headers
> X-Spam-Status: 
SNIP
> X-Spam-Report: * 
SNIP
> Best wishes, 
> S
>

Wonderful, there is a --forget option to sa-learn.  I did not read this,
but I read the man page again and it jumped out at me.

# sa-learn --forget cur/
Forgot tokens from 1 message(s) (1 message(s) examined)