I'm sure I don't need to explain the amount of CJK spam which gets
flung at folks who can't even read it, let alone having any possible
interest in its contents.
Since we as postmasters can't read CJK either, there's not very much
that we can do in a content-related fashion, consequently we're only
too glad that it gets booted out by SA on the basis of some
combination of charset, "excessive" quoted-printable, lots of
(apparently) uppercase letters, chance matches with the Nigerian scam,
and so on. So far, so good.
But, we do have just a few Chinese and Japanese recipients, who seem
to be doomed to have some of their bona fide mails blocked by the same
mechanism, if we don't take some action to help them. We could
exclude them from the SA tests - but then they'd be getting lots of
other spam that we would otherwise protect them from. Conversely we
could water-down the specific SA tests that are liable to block CJK
content, but then we'd be back to accepting more CJK spam for all of
our users.
Now, when SA is used in this context, it's understandable that the
per-user preferences (user_prefs) machinery of SA isn't usable (see
e.g
http://lists.merlins.org/archives/sa-exim/2002-July/000089.html )
At the moment the only thing I can think of is to recognise these
recipient(s) at RCPT TO time (i.e in exim 4 ACL) and cut an extra
header; then somehow use the presence of the header at SA invocation
time to relax the CJK tests.
Or has anyone got a useful recipe already, please?
