First, specify bbdb/gnus-split-method as the split method to use, and specify nnmail-split-fancy as the fallback if the BBDB doesn’t specify a group.
(setq nnmail-crosspost nil nnmail-split-methods 'bbdb/gnus-split-method bbdb/gnus-split-nomatch-function 'nnmail-split-fancy bbdb/gnus-split-myaddr-regexp gnus-ignored-from-addresses nnmail-split-fancy `(| ("Subject" "color-theme" "mail.emacs") (any ,gnus-ignored-from-addresses "mail.misc") (any ".*@emacswiki.org" "mail.emacs.wiki") "mail.spam"))
Now use the gnus-private and gnus-public fields in your BBDB. The first time you add such a field to an entry using C-o in the BBDB buffer, the new field will be defined for you. After that, these two fields are just like any other notes field.
The gnus-private field is used to identify friends, people, companies – the mails you get will be From these addresses. The value of the field is the folder to file any mails from that entry in.
Alex Schroeder net: firstname.lastname@example.org gnus-private: mail.friends
The gnus-public field is used to identify mailing lists – the mails you get will be To these addresses. The value specifies both the folder and an additional regexp which must match the sender.
emacs-devel net: email@example.com gnus-public: mail.emacs.devel .*
I actually always use .* as the regexp in the value of gnus-public fields. Does anybody know under which circumstances these would make sense? – AlexSchroeder
Not satisfied yet? The problem with the approach above is that the BBDB is checked first, and then your FancySplitMail rules are considered. But that means that spam which is sent to mailing lists cannot be filtered! Thus, what we want is to use FancySplitMail rules and call the BBDB splitting somewhere in the middle. Here is an example, which assumes that any mails which have only a text/html part are probably spam.
(setq nnmail-crosspost nil nnmail-split-methods 'nnmail-split-fancy bbdb/gnus-split-default-group nil bbdb/gnus-split-nomatch-function nil bbdb/gnus-split-myaddr-regexp gnus-ignored-from-addresses gnus-total-expirable-newsgroups (regexp-opt '("mail.rohrpost" "mail.guile" "mail.gnutella" "mail.emacs.devel" "mail.emacs.ilisp")) nnmail-split-fancy `(| ("Content-Type" "text/html" "mail.spam") (: (lambda () (car (bbdb/gnus-split-method)))) ("Subject" "color-theme" "mail.emacs") ;; mail from and to me (any ,gnus-ignored-from-addresses "mail.misc") (any ".*@emacswiki.org" "mail.emacs.wiki") ;; anonymous mail, probably spam "mail.spam"))
Another possibility would be to restrict this to mail in weird coding systems which you do not expect. Example for Korean spam:
("Content-Type" "text/html; charset=\"ks_c_5601-1987\"" "mail.spam")
I prefer a solution using add-hook techniques :
(add-hook 'nnmail-split-abbrev-alist (cons 'content-spam "text/html\\|big5\\|gb2312\\|ks_c_.*\\|euc-kr"))
then we can check for it in the split rules with
("Content-Type" content-spam "mail.spam")
You will probably have to add more of these entries as time goes by. The generic solution is problematic if you have friends which do not understand how to switch off HTML mail sending…
If you are using SpamAssassin, your spam mails will contain headers created by the assassin. You can use those headers to split them. One such possibility would be to add a rule such as the following to your splitting rules (
("x-spam-flag" "YES" "mail.spam")
If you want to use GroupParameters to tell which messages belong, you can set these as well. One such possibility would be to set
And, as we have seen above, you can use the BBDB as a whitelist by creating a BBDB entry for people and mailing lists you know.
In order to combine these, however, you have to use something like this (original example by KirkStrauser):
(setq nnmail-split-methods 'bbdb/gnus-split-method bbdb/gnus-split-nomatch-function 'nnmail-split-fancy nnmail-split-fancy `(| ("x-spam-flag" "YES" "mail.spam") (: gnus-group-split-fancy nil t nil) "mail.misc"))
Mail is split as follows: