Justin Mason wrote:
Steve [Spamassasin] writes:
An ebay "watched item" email has been wrongly tagged as spam... with the
following rules:

--
 2.2 INVALID_DATE           Invalid Date: header (not RFC 2822)
 0.8 DATE_IN_PAST_06_12     Date: is 6 to 12 hours before Received: date
 0.1 TW_SJ                  BODY: Odd Letter Triples with SJ
 0.0 HTML_MESSAGE           BODY: HTML included in message
 3.0 BAYES_95               BODY: Bayesian spam probability is 95 to 99%
                            [score: 0.9887]
 0.2 HTML_TITLE_EMPTY       BODY: HTML title contains no text
-0.0 SARE_LEGIT_EBAY        Has signs it's from ebay, from, headers, uri
-1.1 AWL                    AWL: From: address is in the auto white-list
--


The (sanitised) headers read:


--
Subject:...
From:eBay <[EMAIL PROTECTED]>
Date:Wed, 22 Nov 2006 09:03:16 GMT-07:00

While I understand why this email may have triggered the Bayesian rule (where 
spammers have copied ebay's email style...) I am bemused by INVALID_DATE and 
DATE_IN_PAST_06_12.

The dates I see in the header look valid to me - and (if we allow for time 
international time differences) the message was sent two seconds before it was 
received.

Am I overlooking something here?  Why doesn't SpamAssassin like these dates?

they're malformed, missing spaces.  this is what an RFC-compliant
date looks like:

  Date: Wed, 22 Nov 2006 16:20:29 +0000

this is what the ebay.co.uk date looks like, according to yr mail:

  Date:Wed, 22 Nov 2006 09:03:16 GMT-07:00

note: missing spaces; extra ":" in the TZ offset; and the TZ name.  all
are non-rfc-compliant.

--j.


Technically the only thing wrong with the date is the TZ. Section 2.2 of RFC2822 states:

   Header fields are lines composed of a field name, followed by a colon
   (":"), followed by a field body, and terminated by CRLF.

No reference to a mandatory SP character starting the field body.

To the OP, since I highly doubt that you will get eBay to change their TZ format you should consider sa-learn'ing the messages as ham. On your SA setup these messages are hitting BAYES_95 which is adding 3 points to their score.

--
Craig

Reply via email to