[Mimedefang] RE: Subject query (Rob Dege)

[David F. Skoll]

On Wed, 3 Apr 2002 no-reply at aftermagic.com wrote:

> Here's the function, perhaps something similar to this can makes its
> way into future revisions of MimeDefang?!?

I'm reluctant to include functions like this on both technical and
philosophical grounds.  I will include it in a contrib/ directory,
though.

The philosophical grounds:  I think scanning content for specific "dirty"
words goes beyond the intent of MIMEDefang, which is to stop dangerous
content, and detect Spam.  But that's just my opinion; what you use it for
is up to you.

The technical grounds:  The function has many problems; let me suggest
some improvements.

The dirty-word list is loaded for each message.  You might as well do it
once when the filter is loaded.  You could write a function called
init_dirty_list, and call it directly from /etc/mail/mimedefang-filter:

# Outside of filter_begin, filter_end, etc...
init_dirty_list();

>   if($mode eq "body") {
>     unless (open(MESG, "<INPUTMSG")) { # get the message to scan

INPUTMSG contains the raw message, including possibly encoded parts.  The
match will fail on Base64-encoded parts.  Instead, you want to scan each
part in filter().  You can get at the decoded content by opening
$entity->bodyhandle->path (if it is defined.)

You really want three functions:

dirty_subject()       which uses the $Subject global variable
dirty_headers()       which opens "./HEADERS", and
dirty_entity($entity) which opens $entity->bodyhandle->path

Finally, a more ambitious improvement might be to compile all the
dirty phrases into one moster Perl regexp, and let Perl optimize the
matching instead of rolling your own loop. :-)

(Btw, to the original poster:  Please use your real e-mail address when
posting to the list.  This list is a closed list -- you have to subscribe
to post.  I was kind and let the message through, but don't like doing
that.)

--
David.