Let's set up KMail now.
This rule is the first of 4 and does nothing more than piping the mail through DSPAM. DSPAM will scan the mail, and modify the mail (headers, subject tag) accordingly, and resubmits it to KMail.
Don't forget to uncheck the "Stop processing mail" box, as this is not the end of the filtering.
This rule needs to be executed right after rule 1 as this one will check the mail headers to see whether DSPAM has seen a spam or not. If it has, we mark it as spam and move the mail to our "quarantine" box.
To check whether it's spam, we look at the value of X-DSPAM-Result, which is either spam or ham. So the query is:
X-DSPAM-Result equals spam
The following two rules are not actual filters, but are used to retrain mails in case it gets classified the wrong way (false negative or false positive).
Do mind the checkboxes below, as we don't want to run the filter when mail comes in. For convenience, we also add it to the toolbar, allowing quick retraining if necessary.
For clarity: we do a Pipe through with the command below. We need to do this because we need to feed the mail again to DSPAM in order to retrain. We also mark it as ham and move it back to inbox.
dspam --stdout --deliver=innocent,spam --mode=teft --class=innocent --source=error
This rule is the same as rule 3 but instead retrains as spam, in case DSPAM missed it.
Pipe through command:
dspam --stdout --deliver=innocent,spam --mode=teft --class=spam --source=error
And we're nearly done.
