On May 28, 2010 John Hardin wrote: >> I am having the same issue, >> >>In fact, I was going to post the same exact question. >> >>We've been having a problem with emails that are completely blank >> except for an attached .rtf file which is where the spam message is. >> >> Is there a way of decoding that file and feeding it to bayes? > >It's becoming much more prevalent. I'm seeing lots of 419 and phishing spams as .txt and .html attachments, more rarely .doc or >.rtf. > >There's the ExtractText plugin, which we touched on in this context on the list a couple of months back; check teh archives...> > > http://whatever.frukt.org/spamassassin.text.shtml#ExtractText.pm
In my experience with ExtractText.pm it worked great with SA 3.2.5 I have had issues with using it with SA 3.3.1 The 3.3.1 server is a completely new server. Does anybody have SA 3.3.1 and ExtractText working together? Running spamassassin --lint gives this output: May 7 10:08:36.339 [7899] warn: plugin: eval failed: Insecure dependency in eval while running with -T switch at /etc/mail/spamassassin/ExtractText.pm line 578. I can run the unrtf, antiword, pdftohtml at the command line with expected results. I have verified that the paths are correct in .cf and .pm Running: OS CentOS 5.5 SA 3.3.1 unrtf 0.21 antiword 0.37 pdftohtml 0.36
