Konstantin Kletschke wrote:
> I recently wondered why I get e-mails which seemed to not contain a
> body. Investigating this further I relaized that these are multipart
> mails which mostly contained only a html message, which mutt does
> display as an attachement, if ever.

This is off topic, but in mutt if you press 'v' it will view the mime
attachments.

But since the multiple parts are supposed to be equivalent seeing
plain text should be equivalent to seeing the html.  Of course we know
this is a favorite spammer trick.  In MS-Outlook the reverse policy is
in effect.  There the html is preferred over the text.  So most of the
unwashed masses, the targets for spammers, see the html and the plain
text is unseen.  In the plain text spammers are now commonly placing
bayes poison, random words or quotes from novels, trash, whatever.

> I wanted to adjust SA to recognize this as spam but it IS already
> spam, but not completely, hey, look yourself.

Huh?

> I use an debian unstable and its SA deb package.
> ii  spamassassin   2.61-2  Perl-based spam filter using text analysis

Good.  SpamAssassin 2.61 is current.  (BTW...  I use Debian too.)

> Jan  3 13:16:14 zappa spamd[12088]: info: setuid to konsti succeeded
> Jan  3 13:16:14 zappa spamd[12088]: processing message <[EMAIL PROTECTED]> for 
> konsti:1000.
> Jan  3 13:16:14 zappa spamd[12088]: clean message (4.4/6.5) for konsti:1000 in 0.2 
> seconds, 3122 bytes.

What is important about the above logfile snippet?  I am missing the
point.

> Same mail now in a debug session:
> 
> [EMAIL PROTECTED]:~$ spamassassin < spammail.txt

That is not debug.  If you had added the -t and -D options, then I
would call it debug.  Try this instead.

  spamassassin -tD < spammail.txt 2>&1 | pager

> X-Mail-Format-Warning: Bad RFC2822 header formatting in >From konsti  Sat Jan  3 
> 13:16:14 2004
> Return-path: <[EMAIL PROTECTED]>
> Envelope-to: [EMAIL PROTECTED]
> Delivery-date: Sat, 03 Jan 2004 13:16:14 +0100
> Received: from localhost.doom
> ...
> ...bla...
> ...
> Message-ID: <[EMAIL PROTECTED]>
> ...
> ...bla...
> ...
> X-Spam-Flag: YES
> X-Spam-Checker-Version: SpamAssassin 2.61 (1.212.2.1-2003-12-09-exp) on
>         zappa.doom
> X-Spam-Level: ******
> X-Spam-Status: Yes, hits=6.7 required=6.5 tests=BAYES_99,HTML_70_80,
>         HTML_IMAGE_ONLY_02,HTML_MESSAGE,RCVD_IN_BL_SPAMCOP_NET,RCVD_IN_DSBL,
>         RCVD_IN_SORBS autolearn=no version=2.61
> ...
> ...bla...
> ...
> 
> Where does the difference come from?

Difference from what?  What question are you asking?  Please state the
nature of the medical emergency.

> Konsti's user_prefs, if they are ignored in one session are
> 
> score MIME_HTML_ONLY            5
> score MIME_HTML_MOSTLY          5
> score HTML_90_100               5

None of those may have applied to this message.  Try this instead.

score HTML_MESSAGE 5

> score BAYES_99                  3

The result here depends upon your training set.

> auto_learn                      0 
> required_hits                   6.5
> 
> But in my understanding it already is considered from both...

Huh?

Bob

Attachment: pgp00000.pgp
Description: PGP signature

Reply via email to