Re: How to combine spam-split with Subject filtering.

Re: How to combine spam-split with Subject filtering.

[Ted Zlatanov]

On Wed, 12 Jan 2005, oub@mat.ucm.es wrote:

> Finally I got my BBDB and spam-stat based spam filtering to work for
> my imap server. Now my university provides finally its own spam
> filter, which add to possible spam at the beginning of the Subject the
> phrase:
> [SPAM POSIBLE:]
> 
> So I would like to take care of this in the splitting process, and
> split every mail with this subject into a different imap folder so I
> thought the following setting should work:
> (setq
>  nnimap-split-rule  'nnimap-split-fancy
>  nnimap-split-inbox "INBOX"
>  nnimap-split-fancy
>  '(|
>    ("Subject" "[POSIBLE SPAM]:" "SPAM.POSS")
>    (: setq spam-use-BBDB-exclusive nil)
>    (: spam-split 'spam-use-BBDB 'spam-use-stat "SPAM.REAL")
>    (: setq spam-use-BBDB-exclusive t fake-variable nil)
>    (: spam-split 'spam-use-BBDB "SPAM.HAM")
>    (: setq spam-use-BBDB-exclusive nil)
>    "MAILBOX")) 
> 
> it did not however. What do I miss.

You could use spam-use-regex-headers.  Add an entry to the
spam-regex-headers-spam list to match your subject.  The check will be
done against the headers as a whole using re-search-forward so just
escape the brackets appropriately.

It would be nice if in addition to "Subject" we had "Subject+raw" as
an alternative in case the user wants an exact match without regular
expressions (ditto for all the other special headers that can be used,
e.g. "From+raw").  Then "[POSSIBLE SPAM]" would just match, without
the confusion about regex escaping and word boundaries.

Ted

Re: How to combine spam-split with Subject filtering.

[Ted Zlatanov]

On Thu, 10 Feb 2005, oub@mat.ucm.es wrote:
> You mean this:
> 
> `spam-regex-headers-spam' is a variable declared in Lisp.
>   -- loaded from "spam"
> 
> Value: ("^X-Spam-Flag: YES" "^Subject: POSIBLE SPAM SPAM.POSS")

I don't think this is what you want.  Maybe more like

("^X-Spam-Flag: YES" "^Subject: POSIBLE SPAM")

because SPAM.POSS is the mailbox, right?

To get the mailbox "SPAM.POSS", call 

(: spam-split 'spam-use-regex-headers "SPAM.POSS")

in your split-fancy rules.

Ted

Re: How to combine spam-split with Subject filtering.

[Uwe Brauer]

>>>>> "Ted" == Ted Zlatanov <tzz@lifelogs.com> writes:

    Ted> On Wed, 12 Jan 2005, oub@mat.ucm.es wrote:

[snip]

    Ted>  You could use  spam-use-regex-headers.  Add  an entry to the
    Ted> spam-regex-headers-spam  list  to match  your   subject.  The
    Ted>  check  will be done  against  the headers  as  a whole using
    Ted> re-search-forward so just escape the brackets appropriately.

You mean this:

`spam-regex-headers-spam' is a variable declared in Lisp.
  -- loaded from "spam"

Value: ("^X-Spam-Flag: YES" "^Subject: POSIBLE SPAM SPAM.POSS")


    Ted> It would be nice if in addition to "Subject" we had "Subject+raw" as
    Ted> an alternative in case the user wants an exact match without regular
    Ted> expressions (ditto for all the other special headers that can be used,
    Ted> e.g. "From+raw").  Then "[POSSIBLE SPAM]" would just match, without
    Ted> the confusion about regex escaping and word boundaries.
Very true.


Uwe 

Re: How to combine spam-split with Subject filtering.

[Uwe Brauer]

>>>>> "Aidan" == Aidan Kehoe <kehoea@parhasard.net> writes:

    Aidan>  Ar an triú lá déag de mí Eanair, scríobh Uwe Brauer: 

    >> So does isearch-forward-regexp behaves differently from
    >> putting this string in a function which uses a regexp search???

    Aidan> Yes;  if you think about  it,  when you put  a  string in a
    Aidan> function, you  need  to do  string escaping for  characters
    Aidan> like \ and ",  whereas for isearch-forward-regexp there’s
    Aidan> no need for the string escaping,  because no string literal
    Aidan> is involved.

    Aidan> So typing M-C-s  \[POSIBLE SPAM\] RET  does the same regexp
    Aidan> match  as does M-:  (re-search-forward "\\[POSIBLE SPAM\\]"
    Aidan> nil t)  RET . Some variation  on this latter  is better for
    Aidan> building regexps you’re going to use in  a program, in my
    Aidan> experience.
Ok, now doing
M-:  (re-search-forward "\\[POSIBLE SPAM\\]" nil t)
in a spam group which contains a message with this subject, finds this
message. However putting

   ("Subject" "\\[POSIBLE SPAM\\]" "SPAM.POSS")
into nnimap-split-fancy function, does not find this message that is
it is passed to the next entry the spam-split function.

I tried to run edebug on nnimap-split-fancy, but nothing useful came
out.

I am really puzzled, is there somewhere a bug?

Uwe