Hi there. I'm trying to set up an IMAP based bayesian learner using the instructions in the SA wiki for RemoteIMAPFolder, etc.
I'm diverting messages to the IMAP mailstore from MIMEDefang, and I'm trying to set up MIMEDefang to replicate SA's "report_safe" encapsulation format so that sa-learn only learns the encapsulated message while ignoring the included SA report, etc. I appear to have done something wrong, however. Following the instructions in the wiki, I have fetchmail snagging messages from the appropriate IMAP folder and feeding them to sa-learn, but sa-learn doesn't appear to be properly detecting the message encapsulation. As far as I can tell from looking at the code, sa-learn does a check for the existence of the X-Spam-Checker-Version header to decide whether or not to call remove_spamassassin_markup(). Within that subroutine it checks for a Content-Type header matching a regexp which includes "multipart/mixed;" and some other things I don't quite follow. :-) As far as I can tell, though, the messages aren't being detected as encapsulated--I'm using the -D flag with sa-learn and "Removing Markup" never shows up in the dbg messages I expect from the code in remove_spamassassin_markup(), and the debug messages show URLs being parsed which are only present in the spamassassin report included in the body text, but not in the encapsulated message itself. Is there some other trick that I'm missing while generating a message that sa-learn will recognize as "report_safe" encapsulated? Thanks! Working with SA 3.10rc1, by the way. ---- Nels Lindquist <*> Information Systems Manager Morningstar Air Express Inc.