> > :0
> > * ^To: [0-9]
> > {
> > :0
> > |/usr/local/bin/junclient
> > }
> > Since the "^To: [0-9]" condition failed, then its recipe action
> > is not performed. The next recipe is then examined. If there
> > are no more recipes, then procmail tries to deliver the e-mail
> > to the default mailbox for the current user. If there is no
> > current mailbox for the current user (an e-mail list), then
> > procmail *bounces* the e-mail.
>
> That is what I thought procmail was supposed to do, but I couldn't
> seem to get it to work (that is it would not write to an existing
> user's mailbox). The problem, I discovered, actually had nothing
> to do with procmail, it had to do with my FX/etc/passwd %[^:] line
> in my sendmail.cf and the rule that I had added to deal with $~X
> users - the problem was that the currently installed sendmail had
> not compiled in the SCANF option when it was created. :(
It sounds like your rule is not setting the userid before having
procmail invoked.
> > By the way, the recipe condition, as it stands, is not very
> > flexible. What about the case where there is more (or less)
> > than exactly one space between the color and the first number?
> > Ie:
> >
> > To: 1234(_at_)foo(_dot_)bar(_dot_)com
> > Subject: Problems
> >
>
> Is there a way to make the recipie handle such an issue? I cannot
> see any way to make it so that it would handle an arbitrary number
> of spaces... do you?
You should read up on regexps. Especially with the extended regexps
used in procmail, it is very easy match variable length patterns. To
take care of the case above, use a recipe like:
PATH=$PATH:/usr/local/bin
:0
* ^To:[ ]*[0-9]
| junclient
The '[ ]*' means an arbitrary number of spaces or tabs (including none).
If you wanted to allow for arbitrary comment text in front of the
address, such as this example:
To: "The Defender Of the Truth" 12345(_dot_)4567(_at_)aohell(_dot_)com
then you might consider a recipe condition like this:
* ^To:(.*[^0-9a-zA-Z])?\/[0-9.]+
after a successful match, $MATCH will be the user-portion of the address
(assuming that the user addresses are all numeric or periods, like
A-O-HELL accounts).
*Please* read up on regexps: see the man page on egrep(1).
G'luck
_____________________________________________________________________
Alan Stebbens <stebbens(_at_)sgi(_dot_)com> http://reality.sgi.com/stebbens