On 17 Dec 1997 jari(_dot_)aalto(_at_)poboxes(_dot_)com wrote:
I'm trying to count big letter words in the message body, but
I'm unable to contruct the score recipe right. Say, that
I tolerate 3 big letter words, and if there is more, then
I consider it UBE. The regexp should ignore some words like:
SMTP, AM, IP, base64-decoded-lines.
I've found that often people use capital letters more than three times,
so I started filtering by lines with no lowercase. It's just a simple
recipe, but works quite well. From junkfilter:
# if there is no lowercase in too much of the message...
:0BD
* -5^0
* 1^1 ![a-z]
{ SPAMMER="used too much capitalization" }
Of course, the initial 5 lines of all-caps can be modified to whatever
you feel is appropriate.
HTH.
GReg
--
Gregory S. Sutter "How do I read this file?"
mailto:gsutter(_at_)pobox(_dot_)com "You uudecode it."
http://www.pobox.com/~gsutter/ "I I I decode it?"