filter evasion

David Relson relson at osagesoftware.com
Fri Nov 7 19:26:05 CET 2003


On Fri, 7 Nov 2003 12:57:17 -0500 (EST)
Stefan Mashkevich <mash at mashke.org> wrote:

> On Fri, 7 Nov 2003, David Relson wrote:
> 
> > > Now what do we do about the font=white words?
> > 
> > John,
> > 
> > I don't have a good answer for that, at present.  Since bogofilter
> > is scoring the innards of <font> tags, it has _some_ info on the
> > ruse.
> 
> An all but blind shot -- but, given that we seem to be likely to
> encounter more witty experiments with tags in the future, could it
> make sense to treat them (and possibly attributes) specially? Say,
> <font color=white> would yield something like
> 
> tag:font
> tag:font:white
> 
> The latter should expose the criminal intent clearly enough, without
> resorting to rendering the message with a graphical engine and OCR'ing
> it back :-)
> 
>                                                        Stefan

Bogofilter currently scores the innards of a, img, and font tags.  
Attached is a patch that will add a:, img:, and font: prefixes to those
tokens.  Let me know how well it works!
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch.tag.html.1107.txt
URL: <http://www.bogofilter.org/pipermail/bogofilter/attachments/20031107/a650db2d/attachment.txt>


More information about the Bogofilter mailing list