> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] 
> Sent: Tuesday, November 22, 2005 7:35 AM
> To: [email protected]
> Subject: rules and new line
> 
> I've got some spam mails that has this kind of pattern
> 
> A
> Q
> s
> G
> T
> Ue
> O
> 
> I've tried to make a rule but \n doesen't work with 
> spamassassin. Can anyone give me some advice or an example to 
> catch this kind of spam?
> 

Well, it depends on the content type of the message.  If the
content-type is text/html, you have to use a full test with the /s flag.

Not the most elegant (or efficient) test, but currently it's the only
way to go.  At least until there is a better mechanism for doing
multiline rawbody (see
http://issues.apache.org/SpamAssassin/show_bug.cgi?id=4691).  

full BROKEN_TEXT /A\nQ\ns\nG\nT\nUe\nO/s
score BROKEN_TEST 1.0

If the content type is text/plain, then you can use a body rule...
Because newlines get wrapped up by get_body() call, the resulting lines
look like this...

@4000000043832c1129eb21ec [7938] dbg: generic: A Q s G T Ue O

So your rule would look like this...

body BROKEN_TEXT_BODY /A Q s G T Ue O/
score BROKEN_TEST_BODY 1.0


I tested, and both rules work against your sample... Although the body
rule depends on Content-Type text/plain.

# cat message.txt | spamc
X-Spam-Tests:
AWL,BROKEN_TEXT,BROKEN_TEXT_BODY,MISSING_SUBJECT,NO_RECEIVED,
        NO_RELAYS,TO_CC_NONE


Thanks,
Dallas

Reply via email to