Re: how convert my huge nnmail-split-methods into nnmail-split-fancy ?

Re: how convert my huge nnmail-split-methods into nnmail-split-fancy ?

[Hubert Canon]

Ted Zlatanov <tzz@lifelogs.com> schrieb :

> On Tue, 24 Jun 2003, hcanon@alussinan.org wrote:
>> I am trying to use spam-split but I discovered that I need to use
>> nnmail-split-fancy. The problem is that I already have a
>> nnmail-split-methods which has grown out of proportions (3 pages).
>> 
>> Is there a way to use spam-split as is, or should I convert my
>> nnmail-split-methods into a nnmail-split-fancy ? Is there a way
>> (some macro or function) to do it automatically ?
>
> There isn't one currently, but it shouldn't be too hard to write one
> (spam-split-methods maybe).  The key is to cache the return of
> (spam-split) because the group checking function seems to be invoked
> for every mailbox in the nnmail-split-methods list.  I don't use
> nnmail-split-methods so I can't test this very well, but the general
> algorithm should be:

I tried your proposed method, and it seems to work : I have some spams
going into the spam group. But I don't get many spam anymore, so I am
not totally sure that I don't loose some of them.


> 1. initialize a hashtable spam-split-methods-cache outside
>    spam-split-methods
>
> (setq spam-split-methods-cache 
>   (make-hash-table
>     :test 'equal))
>
> 2. spam-split-methods
>
> a) get the message-id
>
> b) if no entry in spam-split-methods-cache for that message ID, make
> entry with the result of (spam-split)
>
> c) get the spam-split-methods-cache entry for the current message-ID,
> and compare it to the group name passed to us from the splitting
> functionality
>
> (defun spam-split-methods (groupname)
> "Call spam-split for consumption by nnmail-split-methods"
>  (let ((message-id (message-fetch-field "Message-ID"))
>        (val (gethash message-id spam-split-methods-cache)))
>   (when (and (stringp groupname) (stringp message-id))
>    (unless val
>     (setq val (spam-split))
>     (puthash message-id val spam-split-methods-cache))
>    (equal groupname val))))

I corrected your `let' with the obvious `let*'

> This is untested, let me know by e-mail if there are problems and the
> result will probably make it to spam.el eventually.

For the moment, it seems to work. I will use it some more time to
convince me that it really works. I would be more confident if other
volunteer to try it.


-- 
Hubert Canon
Please excuse my bad English