On Fri, 10 Oct 2003, Johann Spies wrote:
> I administer 3 email servers which use spamassassin. We are testing > the service with about 110 users whose email are scanned by SA. > > My arrangement with them is to send me either spam that scored too low > or false positives as attachments with either "spam" or "ham" in the > Subject-line. > > What I want to do now is to exctract those emails from the attachments > and put them in a maildir which can then be used by other processes to > feed them so sa-learn. > > Now my question: Are you aware of any tools that can do this? An > example of email that I receive (as Mutt shows the attachments) looks > like this: > > I 2 Apcn, all your relatives can get the sam [message/rfc822, 7bit, 4,2K] > I 3 ><no description> [text/plain, quoted, iso-8859-1, 2,9K] > I 4 dichotoaous rlomberg rbcvuijmbpqed [message/rfc822, 7bit, 3,3K] > I 5 ><no description> [text/plain, quoted, iso-8859-1, 2,0K] > I 6 Invite your friends to play live poker o [message/rfc822, 7bit, 1,7K] > I 7 ><no description> [text/plain, quoted, iso-8859-1, 0,4K] > I 8 Re: Vicodin 71uh [message/rfc822, 7bit, 1,4K] > I 9 ><no description> [text/plain, quoted, iso-8859-1, 0,3K] > I10 no more doctor to get your pharmys 8z [message/rfc822, 7bit, 1,6K] > I11 ><no description> [text/plain, quoted, iso-8859-1, 0,4K] > > I want to put items 2,4,6,8 and 10 in the maildir: ~/Mail/spam as > separate messages. <snip> Hi Johann, First I was confused by your phrase 'send me .. as attachments' beacuse there are 2 formats that various mailreader use: inline and attachment. So, you'd have to check for both types and extract them differently. Secondly, I was not sure if the user would 'reply' or 'forward' the mail to you. These 4 options can be sorted out with procmail. I also realized that the from: would be from your network and the to: would be you, also on your network, so that would differentiate it from other mails using procmail. Also, you can use a (maybe underused) mail filtering technique: the '+'. You can ask that the mail be send to : [EMAIL PROTECTED] or [EMAIL PROTECTED] and this is easy to filter. Or create a 2 new email addresses. These are suggestion to figure out which mails are for sa. Now once that is done, you need to determine what to do with the 4 mail types: inline forward, attachment forward, inline reply and attachment reply. procmail and/or perl may be need to do this simple check. Then you send it to the 'extract spam/ham and feed sa' script. HTH -Kev -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
