option to not count header tokens

[Matthias Andree]

Am 29.04.2009, 13:44 Uhr, schrieb Dmitry <vdb at mail.ru>:

> David Relson wrote:
>> On Tue, 28 Apr 2009 23:42:52 +0400
>> Dmitry wrote:
>>
>>> Hello,
>>>
>>> My ignorelist.db grows every day, still I find more and more useless
>>> header tokens hiding obvious spam. Spammers are smart. They mimic
>>> normal MUA headers. When the message body is short (3-5 words),
>>> headers often cause spams to be classified as ham in such way, that
>>> even marking this message as spam doesn't help.
>>>
>>> My question is: Is it possible to make an option in the config to not
>>> count any invisible header tokens?
>>
>> Dmitry,
>>
>> It sounds like you're asking for spamitarium.  If it does what you
>> want, use it.
>>
>> Given that a script already seems to exist to do what you want, there's
>> no obvious reason to add code to bogofilter.
>
> David,
>
> My main concern is better filtering. Bogofilter does addionnal job when
> parsing headers. I just found that without header tokens its filtering
> capability is more accurate. So, why another tool? In one shot you can
> make bogofilter faster and better... It is not that I am asking to add
> some code. I want to get rid of existing unwanted functionality.

Dmitry,

your aim is not quite clear to me.

Do you suggest:

(1) That bogofilter does not look at invisible headers at all? Then  
Spamitarium does indeed just that: pre-filter what bogofilter sees.

OR

(2) That bogofilter should not mark tokens specially that were obtained  
 from headers? For instance, do you want that bogofilter stops using head:  
and subj: and similar prefixes and just uses these tokens as though they  
were found in the message body?

Seems you want the former.

HTH

-- 
Matthias Andree