[Spambayes] updating x-spambayes-classification on already trained mail?
alex at chizang.net
Thu Apr 19 22:17:55 CEST 2007
* skip at pobox.com <skip at pobox.com>:
> Alex> For my test, I reran sb_filter.py on each message (maildir
> Alex> format, but not sure it matters) and discovered the following:
> Alex> a) sb_filter.py doesn't seem to change the header if it already
> Alex> exists in the message
> Hmmm... works for me. spam.multi is an mbox message dating from Oct
> 2006 that already has classification and evidence headers:
D'oh, pilot error. You're right -- sb_filter.py will rescore and
change the header, even if it already exists.
My misunderstanding was in how sb_filter interacts with procmail.
As the -h option says:
Output is always to standard output as a Unix-style mailbox.
Duh -- I'd assumed that it actually rewrote the file.
Looks like I simply need to modify my script to rewrite the file
with the output from sb_filter.
More information about the SpamBayes