Problems with UTF8 and docx (???)

[Mark Martinec]

Jakob,

> We run "amavisd-new-2.7.0-pre14" on a fedora system defaulting to UTF-8.
> Since some time we have users complaining about docx documents that
> cannot be sent (but not all docx documents have this problem). When
> looking into the amavisd log, I see
> 
> Mar 29 17:17:27 xxx amavis[21041]: (21041-02) Checking: 9oiHPxyAyzpd
> [80.153.159.39] <xy at df.com> -> <er at df.de>
> Mar 29 17:17:28 xxx amavis[21041]: (21041-02) (!!)TROUBLE in check_mail:
> check-banned FAILED: Malformed UTF-8 character (fatal) at
> /usr/local/sbin/amavisd line 4820.
> Mar 29 17:17:28 xxx amavis[21041]: (21041-02) (!)PRESERVING EVIDENCE in
> /var/amavis/tmp/amavis-20110329T171654-21041-6eGAXmsw
> 
> When looking at the eMail I cannot see real problems.
[...]
> ... but still the only way to avoid the documents ending up as unsent
> mail is to disable banned checking altogether.
> perl, v5.8.8 built for x86_64-linux-thread-multi

I would like to see either a log at level 5, or the preserved sample message
(off list), if that is possible. What I'd be looking for is where does the
message part name causing trouble come from. It would be better to
fix this at the source of trouble, rather than relaxing the error handling
in lookup_re.

  Mark