spam addrs

[David Relson]

On Tue, 15 Jun 2004 11:11:03 -0400
Tom Anderson wrote:

> From: "David Relson" <relson at osagesoftware.com>
> > Bogofilter has many optional features.  I doubt that anyone uses
> > them all.  Additionally there are frameworks included, such as
> > support of different charsets, that are awaiting users to be
> > completed.
> 
> True enough.  I'm not an expert in much of that stuff, but seeing as
> I've done some research into this particular functionality, I just
> want to ensure that you're aware of the possible shortfalls of what
> you're trying to implement.  Feel free to ignore my comments as you
> please.

Tom,

I appreciate the info on variety of MTA generated Receive: statements.
My approach was radically affected by the info.  The original idea was
to use the lexer grammer in lexer_v3.l, but that quickly showed itself
to be unmanageable.  Modifying token.c turned out to be a much better
solution since it already knew when a Received: statement was being
processed and it knows when an IPADDR is found.  Adding a bit of code
allowed ignoring Received statements without ip addresses and allowed
keeping the last ip address of the first Received statement having one. 
All together, the feature adds about 25 lines of code, including
declarations, processing of '%I' in formats, etc.

I've also included "Not guaranteed to be the originating address of the
message." in two places.

Regards,

David