274133 - Analyze headers in junk filter

Status: RESOLVED WORKSFORME

(Thunderbird :: General, enhancement)

Description

[Jerry Quinn]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.3) Gecko/20041007 Galeon/1.3.18 (Debian package 1.3.18-1.1)
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.3) Gecko/20041007 Galeon/1.3.18 (Debian package 1.3.18-1.1)

It seems like thunderbird doesn't look at headers when learning what is spam. 
The filter is pretty good, but I already run a pop3 spam proxy (POPfile) that
appends a header to emails.  I actively train that one, and if thunderbird would
look at the header, it would probably be able to be more accurate.

Reproducible: Always
Steps to Reproduce:
1.
2.
3.

Comment 1

[Frank Wein [:mcsmurf]]

You can create a new message filter, which checks for this specific header and
then marks the message as spam. Invalid (i dont think we can check for every
header a anti-spam program creates, there are too many of them)? 

Comment 2

[Jestre]

WFM
A quick look in my training.dat file shows From: headers, To: headers and
x-header-noreverseip:ip.name.lookup.failed[222.77.23.102] as well as hundreds if
not thousands of other headers listed.  TB bayesian junk filter does filter the
headers too.
If you train TB on the mail already marked by your other filter it should learn
fast to mark anything with that header as junk, however if some of the
legitimate mail gets marked by the other filter it may mess up the weighting
especially if you retrain it in TB.

Also try setting up a 'Message Filter' for the other spam filters appended
header(s) to filter them out and select "Set Junk status to: Junk'.
This will also help auto train TB based on the other filters results.

Also you may want to look at the thread
http://forums.mozillazine.org/viewtopic.php?t=167758

Comment 3

[Scott MacGregor]

we add tokens for each header