Newbie query: working on already delivered messages - batch mode?

[Chris Wilkes]

On Thu, Jun 09, 2005 at 04:49:39PM -0700, D Ruggiero wrote:
> So, Bogofilter's happily installed on my ancient RH 7.1 mail server (don't 
> ask) and is so much easier, more efficient, and accurate to the client-side 
> filter I had been using that it's laughable. So far, so good - thanks for 
> the great work.

What version of BerkeleyDB did you use?  I'm just wondering how you went
about upgrading that as it is usually a pain.

> However: due to my bizarro setup here, I'm only able to filter messages 
> after they're already delivered (ie, in my maildir as separate files, one 
> per message). Again, don't ask. :)  What I really want to do is simply have 
> Bogofilter look at each message and insert its X-Bogosity header "in-situ", 
> without otherwise changing the message or its filename. Right now I'm doing 
> this with a bash script:

That's probably not a good idea to do it to the exact same file name.
What happens when your bogofilter script runs and the message is popped
off the server at that moment?

> #!/bin/bash
> cd [my maildir]
> find . -type f -newer bogotime > /tmp/bogofiles
> touch bogotime
> 
> xargs -i -r bash -c "/bin/mv {} bogotmp; bogofilter -peI bogotmp -O {}" \; 
> < /tmp/bogofiles
> 
> rm -f bogotmp /tmp/bogofiles
> exit 0

If you have control to which folder these emails are sent to have them
go to a folder like
  ./Maildir/.unfiltered/
and then go through those and move them into the
  ./Maildir/new
directory using the "safecat" program:
  http://jeenyus.net/~budney/linux/software/safecat.html

> Two problems with this: first, it seems like an ugly kluge, and second, it 
> requires a file move and a separate invocation of bogofilter for every 
> message. Surely there must be a better way? Is there method to get 
> Bogofilter to operate on all the found "new" messages all at once, adding 
> its header to each, and keeping them in the same file names, all in one 
> fell swoop?

What you're doing sounds like its requires processing on each individual
email beyond a simple "score this email" -- you're actually writing
information to the file.

> (I have a feeling that "batch" mode is what I want, but I'm not sure how to 
> make that work on separate messages rather than one big input stream. I'm 
> also very hazy on the difference between "-b" and "-B"....)

I use "-B *" to do an operation on all the emails in my current
directory.  But I only use it to print out the spam score or register
them as spam/nospam.

> Any advice or insight is appreciated.

If you have IMAP access to the server I would make up a "spam" directory
and instead of putting the spam score in the header move the spam to
that folder.  However that doesn't work out so well with POP.

Chris