Home of the ZMI_GERMAN ruleset for spamassassin
This ruleset filters german language spam, and is very false positive proof, so
everybody can use it, as it will not disturb your mail flow.
How to get it:
- preferred method is via sa-update, channel sa.zmi.at, by using our gpg key 40F74481
first get the channel key from https://sa.zmi.at/sa-update-german/GPG.KEY
(I've uploaded the key for a double-check to wwwkeys.pgp.net, e-mail spam-german@zmi.at, key id 40F74481) then do
sa-update --import GPG.KEY
once, and now you can use
sa-update --gpgkey 40F74481 --channel sa.zmi.at
to get any updates. Do this around ONCE or TWICE A DAY, that should be enough.
- to use the channel without GPG, simply do
sa-update --nogpg --channel sa.zmi.at
If you ever have a problem with the ruleset, please check the spamassassin users list
users@spamassassin.apache.org first, for specific questions your
Maintainer: Michael Monnerie (spam-german@zmi.at) from protéger.at
PDF scanner for SpamAssassin
In order to scan PDFs, which are often used to hide spam, I searched and found an old
"PDFassassin" on the SpamAssassin Homepage.
As that is unmaintained, I copied it over, patched it a bit, and want to provide it here. I hope people
help me out with this, I'm not a programmer, but managed to make it run on our systems. With trivial effort, you can
grab it and install it in your SpamAssassin environment. Here's the HOWTO:
- get both pdf.cf and pdf.pm from here, and store it in your spamassassin config dir (where your other .cf files are, like /etc/mail/spamassassin)
- Install pdftotext, which is contained in the Xpdf package
- /usr/bin/pdftotext must be executable
- now restart spamd to re-read it's config
- Test with a message that has a PDF attachment:
cat message_with_pdf_attachment|spamassassin -D -L |less
and search for "pdf". This should at least bring a line with
plugin: loading PDF from /etc/mail/spamassassin/pdf.pm
,
and if the PDF itself contains text that looks like spam, bring a hit like this:
X-Spam-Status: Yes, hits=28.3 required=5.0 tests=AWL=4.040,BAYES_60=1,PDF=18.069,...
Note: the points are given by converting the pdf to text, and piping that through SpamAssassin again
Note that this is an early 0.1 version. It runs on my systems, and I hope others will help out to improve it.
The original version also called gocr (available via: http://jocr.sourceforge.net), but I don't want OCR so commented those lines out.
If other want/need this, we can activate it again as soon as someone sends a tested patch.