I couldn't find anyone who has done this already, so I did it myself -
Nice work!
How does this interact with the use/nonuse of report_safe? It seems to me that (with report_safe 1) you end up training bayes on the encapsulation, or (with report_safe 0) you end up training it on the reciprocal of the spamassassin-added headers.
To my knowledge, salearn removes/ignores any SpamAssassin headers, so it shouldn't skew your data.
Ryan Moore ---------- Perigee.net Corporation 704-849-8355 (sales) 704-849-8017 (tech) www.perigee.net
