> -----Original Message-----
> I think you might need to uncomment
>=20
> loadplugin Mail::SpamAssassin::Plugin::TextCat
>=20
> in /etc/mail/spamassassin/v310.pre (or your equivalent).
>=20
> As an aside, I'm afraid I simply reject some of them out of hand if

they
> contain a particular MIME charset. Using an Exim ACL, I do:
>=20
> deny message =3D Sorry, no one speaks this language here.
> condition =3D ${if eq{lc:$mime_charset}{gb2312}{1}{0}}
>=20
> That one's simplified Chinese, but I reject other languages as well,
> including Russian.
>=20
> You may be able to do something similar with your MTA. And it's one

less
> message for SA to scan
>=20


Does anyone know how to do this with Qmail, or is it best left to
Spamassassin?

I've written some custom rules in my local.cf file to tag the Russian
spam that seems to be on the rise. I think they're getting around some
of my regular expressions, however, and I'd really like a better way to
do it. We do have clients of Asian origin, so it would be kind of nice
to be able to tag spam for the character sets (koi8-r & windows-1255)
that the Russian spam uses, leaving other charsets alone.

Currently I'm just regexp'ing the subject and body of the message for
those character sets and scoring accordingly.=20

A better way would be to match phrases from the spam messages
themselves, although this will cause more load on SA, it would be a
"more world-friendly" method.=20
This involves translating the spam messages into English, then matching
up a spam-like phrase to tag, then finding that phrase in the actual
email file on the server(no codepage conversions necessary) then
grabbing the section of non-sensical UTF-8 characters to put in my
local.cf file for scoring.=20

Anyone have better rules/methods?

Cheers,
Mike