new spam.el (alpha code)

[Ted Zlatanov <tzz@lifelogs.com>]

[-- Attachment #1: Type: text/plain, Size: 2089 bytes --]

Hello,

I am attaching the latest spam.el code.  90% of the work was done by
François Pinard, who deserves the credit for not only adding an
interface to bogofilter, but also implementing a lot of new and
necessary features.  There's also summary exit functionality to
process articles marked as spam.  There's lots of FIXME comments, and
instructions on installing bogofilter and a procmail recipe too.  The
goal is to add more spam checks as time goes on.

(bogofilter is a program that implements statistical analysis in an
external process from Emacs.  See the comments inside spam.el)

I also added a BBDB check, which uses the BBDB like a whitelist.
Credit goes to Alexander Kotelnikov.  Alexander, please let me know if
it's OK to use your code :)  The code assumes bbdb.el is loaded -
that's probably a bad idea, so I marked it with a FIXME.  I'm also not
sure about the whitelist/blacklist logic, that may be a little bogus.

I have not committed to CVS yet, because all the code is in alpha.
The CVS spam.el is fairly stable (well, it's broken, and that's a sort
of stability, isn't it? :) so I'd rather let people test the code and
fix bugs outside CVS for the first few days.  The spam-split function,
especially, has completely changed, and the new spam.el may break code
that uses it.

The work currently in spam.el can use RBL, whitelist/blacklist/BBDB,
and bogofilter.  We can add other Bayesian analyzers; several external
ones were discussed already.  The spam-stats.el Lisp statistical
analysis could be included as well.  I think 95% of the work for
spam-stats.el is done by the author of the library already, only the
final glue in spam.el needs to be put together.

The documentation for the spam.el library is in the Lisp comments
right now.  I will write real docs as soon as we have a stable
user-side interface.  I know they are necessary.

I will be gone from tomorrow until Sept. 25 on vacation, so feel free
to kick spam.el around until then.  Someone else can commit to CVS if
I don't respond quickly.

Thanks
Ted


[-- Attachment #2: spam.el --]
[-- Type: application/emacs-lisp, Size: 20981 bytes --]