http://bugzilla.spamassassin.org/show_bug.cgi?id=3269





------- Additional Comments From [EMAIL PROTECTED]  2004-04-15 23:18 -------
Subject: Re:  MIME parser issues vs. 2.63

On Thu, Apr 15, 2004 at 10:39:06PM -0700, [EMAIL PROTECTED] wrote:
> Hmmm... is there a logical way we could also catch multiple closes.

it does.  =1 when it finds an open, -- when it finds a close, alert on
!= 0.  so 1 open and 2 closes makes -1, which != 0, so trigger.

> Ideally, we'd test complete correctness to catch incorrectly nested
> stuff and other types of errors:
> 
> --boundary_a
> 
> --boundary_b
> 
> --boundary_a--
> 
> --boundary_b--

the way the parser works, that'll get caught if the message is boundary_a
and part 1 specifies boundary_b.  anything after the boundary_a close
is ignored.

so the rule triggers due to no close for boundary_b.

> I think the current test also would miss an extra close (before an open
> and close which would leave the state at zero).

yes, which is (iirc) perfectly valid according to the rfc.  any before
the open is "preamble" and is discarded.





------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to