Well, I got it working (started a thread like this one month ago or so)
thanks to some other users contributions.
What I did (following someone elses instructions) was insert these lines in
local.cf:
<code>
bayes_ignore_header ReSent-Date
bayes_ignore_header ReSent-From
bayes_ignore_header ReSent-Message-ID
bayes_ignore_header ReSent-Subject
bayes_ignore_header ReSent-To
all_spam_to [EMAIL PROTECTED]
</code>
and after that, set up the spam account, and generate a script to teach
spamassassin the forwarded messages. Maybe the bash programming could be
improved, but it's working for me like this:
<code>
#!/bin/bash
result_spam=$(ls /usr/local/virtual/[EMAIL PROTECTED]/new | wc -l)
if [ $result_spam -ne 0 ]
then
spamassassin -r -d --progress < /usr/local/virtual/spam@
mydomain.tld/new
echo 1 > /var/tmp/sa-state
rm -f /usr/local/virtual/[EMAIL PROTECTED]/new/*
fi
</code>
This script runs through cron once an hour along with some others which scan
my users IMAP folders (some of them only use a webmail, and some use MUAs,
such as Outlook Express), seach for spam and ham folders, checks if they're
not empty, and learn from their contents (that's why I put that "echo"
sentence. BTW, if anyone knows a better way to check for the existence of
spam in the precedent script, I will gladly accept their tips...
When I finish polishing this, I promess to make a nice package and share it.
Luis
2007/3/21, Wael Shaheen <[EMAIL PROTECTED]>:
Hello everyone,
am looking for a mechanism which allows my clients to report spam by
forwarding a message or attaching it to a single mailbox i.e
[EMAIL PROTECTED]
How can i do this, forwarded emails will have the sender information cut
off
before being fed to sa-learn
I would appreciate any hints-information in that direction and if any idea
would be better than what am thinking of
or have sa-learn run on the attached messages
Thank you
--
-------------------------------------------------
GNU-GPL: "May The Source Be With You...
-------------------------------------------------
