[vps-mail] How can procmail identify report_safe-wrapped spam?

[Lang Zerner <lang@xxxxxxxxxxxx>]

Some of my VPS v2 customers complain that spam stays in their inboxes, 
even when they're configured to have spam delivered to another mailbox, 
or to /dev/null.

It appears that the SpamAssassin default is now to have report_safe on.  
That means that when SpamAssassin identifies a message as spam, it adds 
an "X-Spam-Status: Yes"header as usual, but *then* it wraps the spammy 
message as a MIME attachment in an automatically-generated "report" message:

Spam detection software, running on the system 
"mail-server.mydomain.com", has
identified this incoming email as possible spam. The original message
has been attached to this so you can view it (if it isn't spam) or block
similar future email. If you have any questions, see
the administrator of that system for details.

Unfortunately, the report message (which SpamAssassin creates to safely 
wrap the spam and thus prevent inadvertent viewing of potentially 
malicious scripts) is not itself spam, so its X-Spam-Status header says 
it is *not* spam.  That means procmail no longer identifies the message 
as spam, so users who want to redirect spam to /dev/null or another 
mailbox no longer get the behavior they used to.

Okay, that's the problem.  Now, I'm trying to devise a solution.  One is 
to turn off report_safe globally, so procmail can see the 
"X-Spam-Status: Yes" header again.  This isn't ideal, for all the 
reasons that report_safe is a good idea.  But without doing this, I 
don't see any way for procmail to identify one of these report messages 
other than scanning the body of the message, which isn't really reliable 
because the body text may change in future versions of SA.

How can I get procmail to reliably identify report_safe messages as 
spam?  Is there a way to make SpamAssassin put an identifying header 
onto messages it wraps in report_safe messages?

Thanks,
--Lang