Lately I've been noticing multi-part spam messages where the plain text part
of the message was some kind of news article and the HTML part was a totally
unrelated advertisement. (See the example below.) I assume this is another
spammer trick to try to corrupt Bayes, since most email clients won't
display the plain text section of the message.
So here's the question - for future releases of Spamassassin, would it be a
good/bad idea to set it up so that for a multi-part email message, Bayes
only learns from the HTML part of the message and ignores the plain text
section? In cases where both sections are an advertisement it should pick
up the appropriate tokens from the HTML section, and in other cases the
plain text section is probably just Bayes poison anyway.
Just a thought....
Sandy S.
Here's a sample of the body of one of these spams:
------=_NextPart_000_38C2B9_01C389FB.12DD8590
Content-Type: text/plain;
charset="Windows-1252"
Content-Transfer-Encoding: quoted-printable
February 17, 1987, just weeks before his death). Warhol originally
intended these daily records to be documentation of his minor "business"
expenses. He was just audited and felt the need to be extra careful. "In a
word it was a diary. But whatever its broader objective, its narrow one,
to satisfy tax auditors, was always on my mind" (Warhol xvi).
------=_NextPart_000_38C2B9_01C389FB.12DD8590
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<snipped out pages and pages of HTML code here, but it was an advertisement
for "Free Debt Reduction Analysis">
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
