procmail
[Top] [All Lists]

RE: Spammish?

2003-02-20 07:59:02
fleet(_at_)teachout(_dot_)org wrote:


On Mon, 17 Feb 2003 dman(_at_)nomotek(_dot_)com wrote:

formail -ds procmail ~/spamtest/test.rc < ~/Mail/backup.1

I suspect the -d is what's messing you up.

BINGO.  Removing the -d flag got rid of all the 0 hits 

Good.

                    How many people who use Outlook (and not
also Exchange, which would show up in the headers as well if it were
there) have direct pipelines to incoming SMTP servers at 
your upstream provider?

And just how do you get procmail to determine that it is significant
that the X-Mailer is Outlook without Exchange, and that the 
message was inserted into the system without benefit of an ISP?

(And, since I don't send my mail through an ISP, my messages 
probably get looked at with a jaundiced eye!) :)


Jesus, but you want all my crown-jewel secrets, don't you.  :)

 :0  # 030211 () RCVD_COUNT shouldn't be low for home-style mail clients
  * $                     ^X-Mailer:(.*\<)?$HOME_MUA
  * $                     LOCALSOURCE  ??  $FALSE
  * $             -1^0    MYTAG        ??  $TRUE
  *                1^0  ! domKEY       ??  ^^(ccc|panx)^^
  *                1^0    domKEY       ??  ^^bofh^^
  * $  $RC_THRESHOLD^0
  * $   -$RCVD_COUNT^0
  { RX = "${RX:+$RX, }UBE.XM.NONBULK+PIPELINED" }

Catches between 15% and 25% of all my spam with
no false pozzes.  The $domKEY thing essentially
adjusts the length of the threshold for a couple
of odd domains that are forwarded to my main host
via different means (meaning Received count 
differs from my norm).  Currently: 22 of last 100 
spam messages were caught by the above.

 3:30pm [~/Mail] 417[1]> distro | sed -n '1,/PIPE/p'
Finding distribution for "^X-Recipe-ID: " within 
selected file(s) (default: [*])
  51 UBE.ID.FAKE
  41 UBE.TRUST<LOWEST
  41 UBE.VH.!HOTHOO
  35 UBE.RC.LOW_COUNT+TO.!ME+TRUST<HIGH
  33 UBE.SJ.END+(SPACEY|NUMS|NOVOWELS)
  32 UBE.OH.RETROFIT-MUA
  28 UBE.DT.BOGUS
  23 UBE.ID.MYUPSTREAM
  22 UBE.XM.NONBULK+PIPELINED

What's in $HOME_MUA are just a half-dozen obviously home-style
mailer names culled from my recent mail.  Putting that together
took about five or ten minutes.  I could add lots more, but
don't bother.  I go with what the spammers forge most often.

$RC_THRESHOLD is the number 3 at present.  If my mail headers
change with software or system upgrades or changes, I can
change the number.  (I did that once recently.)

Oh, and:

  TRUE         = .                   # normalize Boolean nomenclature
  FALSE        = ^^^^                # ditto


You maintain a list of X-Mailers? Categorized?

Occasionally I run

        sed -n '1,/^$/{ /^X-Mailer:/p; }'

on my archived recent spam and good mail and save the results to
different respective files.  Then "sort -u" on the result.  Then,
from the two files,

        comm -23 baddies goodies

shows me the ones only the spammers used recently in my mail.

I'm not manic about this.  I do it when I'm bored and think my
list needs to be updated.  The files currently are 46 and 37
lines long, respectively.  I could put this all in a script,
but haven't even bothered to do that yet.


This is an area I need to work on.  Your example recipes will 
be helpful!

I ain't gonna spill 'em all at once.  Buy the book!  :)  :-)

Dallman



_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail

<Prev in Thread] Current Thread [Next in Thread>