Synch to No Gnus 200405141948.
[elisp/gnus.git-] / lisp / spam.el
index 22c1dcf..39a6748 100644 (file)
 
 ;;; Code:
 
-(require 'path-util)
-
 (eval-when-compile (require 'cl))
 
+(require 'path-util)
 (require 'gnus-sum)
-
 (require 'gnus-uu)                     ; because of key prefix issues
 ;;; for the definitions of group content classification and spam processors
 (require 'gnus)
@@ -75,7 +73,7 @@
 (defgroup spam nil
   "Spam configuration.")
 
-(defcustom spam-directory "~/News/spam/"
+(defcustom spam-directory (nnheader-concat gnus-directory "spam/")
   "Directory for spam whitelists and blacklists."
   :type 'directory
   :group 'spam)
@@ -164,6 +162,11 @@ The regular expression is matched against the address."
   :type 'boolean
   :group 'spam)
 
+(defcustom spam-use-gmane-xref nil
+  "Whether the Gmane spam xref should be used by `spam-split'."
+  :type 'boolean
+  :group 'spam)
+
 (defcustom spam-use-blacklist nil
   "Whether the blacklist should be used by `spam-split'."
   :type 'boolean
@@ -220,6 +223,18 @@ Enable this if you want Gnus to invoke Bogofilter on new messages."
   :type 'boolean
   :group 'spam)
 
+(defcustom spam-use-bsfilter-headers nil
+  "Whether bsfilter headers should be used by `spam-split'.
+Enable this if you pre-process messages with Bsfilter BEFORE Gnus sees them."
+  :type 'boolean
+  :group 'spam)
+
+(defcustom spam-use-bsfilter nil
+  "Whether bsfilter should be invoked by `spam-split'.
+Enable this if you want Gnus to invoke Bsfilter on new messages."
+  :type 'boolean
+  :group 'spam)
+
 (defcustom spam-use-BBDB nil
   "Whether BBDB should be used by `spam-split'."
   :type 'boolean
@@ -247,8 +262,22 @@ considered spam."
   :type 'boolean
   :group 'spam)
 
+(defcustom spam-use-spamassassin nil
+  "Whether spamassassin should be invoked by `spam-split'.
+Enable this if you want Gnus to invoke SpamAssassin on new messages."
+  :type 'boolean
+  :group 'spam)
+
+(defcustom spam-use-spamassassin-headers nil
+  "Whether spamassassin headers should be checked by `spam-split'.
+Enable this if you pre-process messages with SpamAssassin BEFORE Gnus sees
+them."
+  :type 'boolean
+  :group 'spam)
+
 (defcustom spam-install-hooks (or
                               spam-use-dig
+                              spam-use-gmane-xref
                               spam-use-blacklist
                               spam-use-whitelist
                               spam-use-whitelist-exclusive
@@ -256,8 +285,12 @@ considered spam."
                               spam-use-hashcash
                               spam-use-regex-headers
                               spam-use-regex-body
-                              spam-use-bogofilter-headers
                               spam-use-bogofilter
+                              spam-use-bogofilter-headers
+                              spam-use-spamassassin
+                              spam-use-spamassassin-headers
+                              spam-use-bsfilter
+                              spam-use-bsfilter-headers
                               spam-use-BBDB
                               spam-use-BBDB-exclusive
                               spam-use-ifile
@@ -283,14 +316,23 @@ All unmarked article in such group receive the spam mark on group entry."
   :type '(repeat (string :tag "Group"))
   :group 'spam)
 
+
+(defcustom spam-gmane-xref-spam-group "gmane.spam.detected"
+  "The group where spam xrefs can be found on Gmane.
+Only meaningful if you enable `spam-use-gmane-xref'."
+  :type 'string
+  :group 'spam)
+
 (defcustom spam-blackhole-servers '("bl.spamcop.net" "relays.ordb.org"
                                    "dev.null.dk" "relays.visi.com")
-  "List of blackhole servers."
+  "List of blackhole servers.
+Only meaningful if you enable `spam-use-blackholes'."
   :type '(repeat (string :tag "Server"))
   :group 'spam)
 
 (defcustom spam-blackhole-good-server-regex nil
-  "String matching IP addresses that should not be checked in the blackholes."
+  "String matching IP addresses that should not be checked in the blackholes.
+Only meaningful if you enable `spam-use-blackholes'."
   :type '(radio (const nil)
                (regexp :format "%t: %v\n" :size 0))
   :group 'spam)
@@ -301,22 +343,26 @@ All unmarked article in such group receive the spam mark on group entry."
   :group 'spam)
 
 (defcustom spam-regex-headers-spam '("^X-Spam-Flag: YES")
-  "Regular expression for positive header spam matches."
+  "Regular expression for positive header spam matches.
+Only meaningful if you enable `spam-use-regex-headers'."
   :type '(repeat (regexp :tag "Regular expression to match spam header"))
   :group 'spam)
 
 (defcustom spam-regex-headers-ham '("^X-Spam-Flag: NO")
-  "Regular expression for positive header ham matches."
+  "Regular expression for positive header ham matches.
+Only meaningful if you enable `spam-use-regex-headers'."
   :type '(repeat (regexp :tag "Regular expression to match ham header"))
   :group 'spam)
 
 (defcustom spam-regex-body-spam '()
-  "Regular expression for positive body spam matches."
+  "Regular expression for positive body spam matches.
+Only meaningful if you enable `spam-use-regex-body'."
   :type '(repeat (regexp :tag "Regular expression to match spam body"))
   :group 'spam)
 
 (defcustom spam-regex-body-ham '()
-  "Regular expression for positive body ham matches."
+  "Regular expression for positive body ham matches.
+Only meaningful if you enable `spam-use-regex-body'."
   :type '(repeat (regexp :tag "Regular expression to match ham body"))
   :group 'spam)
 
@@ -402,6 +448,53 @@ your main source of newsgroup names."
                 (const :tag "Use the default"))
   :group 'spam-bogofilter)
 
+(defgroup spam-bsfilter nil
+  "Spam bsfilter configuration."
+  :group 'spam)
+
+(defcustom spam-bsfilter-path (executable-find "bsfilter")
+  "File path of the Bsfilter executable program."
+  :type '(choice (file :tag "Location of bsfilter")
+                (const :tag "Bsfilter is not installed"))
+  :group 'spam-bsfilter)
+
+(defcustom spam-bsfilter-header "X-Spam-Flag"
+  "The header inserted by Bsfilter to flag spam."
+  :type 'string
+  :group 'spam-bsfilter)
+
+(defcustom spam-bsfilter-probability-header "X-Spam-Probability"
+  "The header that Bsfilter inserts in messages."
+  :type 'string
+  :group 'spam-bsfilter)
+
+(defcustom spam-bsfilter-spam-switch "--add-spam"
+  "The switch that Bsfilter uses to register spam messages."
+  :type 'string
+  :group 'spam-bsfilter)
+
+(defcustom spam-bsfilter-ham-switch "--add-ham"
+  "The switch that Bsfilter uses to register ham messages."
+  :type 'string
+  :group 'spam-bsfilter)
+
+(defcustom spam-bsfilter-spam-strong-switch "--sub-spam"
+  "The switch that Bsfilter uses to unregister ham messages."
+  :type 'string
+  :group 'spam-bsfilter)
+
+(defcustom spam-bsfilter-ham-strong-switch "--sub-clean"
+  "The switch that Bsfilter uses to unregister spam messages."
+  :type 'string
+  :group 'spam-bsfilter)
+
+(defcustom spam-bsfilter-database-directory nil
+  "Directory path of the Bsfilter databases."
+  :type '(choice (directory
+                 :tag "Location of the Bsfilter database directory")
+                (const :tag "Use the default"))
+  :group 'spam-bsfilter)
+
 (defgroup spam-spamoracle nil
   "Spam spamoracle configuration."
   :group 'spam)
@@ -419,12 +512,76 @@ When nil, use the default spamoracle database."
                 (const :tag "Use the default"))
   :group 'spam-spamoracle)
 
+(defgroup spam-spamassassin nil
+  "Spam SpamAssassin configuration."
+  :group 'spam)
+
+(defcustom spam-spamassassin-path (executable-find "spamassassin")
+  "File path of the spamassassin executable program.
+Hint: set this to \"spamc\" if you have spamd running.  See the spamc and
+spamd man pages for more information on these programs."
+  :type '(choice (file :tag "Location of spamc")
+                (const :tag "spamassassin is not installed"))
+  :group 'spam-spamassassin)
+
+(defcustom spam-spamassassin-arguments ()
+  "Arguments to pass to the spamassassin executable.
+This must be a list.  For example, `(\"-C\" \"configfile\")'."
+  :type '(restricted-sexp :match-alternatives (listp))
+  :group 'spam-spamassassin)
+
+(defcustom spam-spamassassin-spam-flag-header "X-Spam-Flag"
+  "The header inserted by SpamAssassin to flag spam."
+  :type 'string
+  :group 'spam-spamassassin)
+
+(defcustom spam-spamassassin-positive-spam-flag-header "YES"
+  "The regex on `spam-spamassassin-spam-flag-header' for positive spam
+identification"
+  :type 'string
+  :group 'spam-spamassassin)
+
+(defcustom spam-spamassassin-spam-status-header "X-Spam-Status"
+  "The header inserted by SpamAssassin, giving extended scoring information"
+  :type 'string
+  :group 'spam-spamassassin)
+
+(defcustom spam-sa-learn-path (executable-find "sa-learn")
+  "File path of the sa-learn executable program."
+  :type '(choice (file :tag "Location of spamassassin")
+                (const :tag "spamassassin is not installed"))
+  :group 'spam-spamassassin)
+
+(defcustom spam-sa-learn-rebuild t
+  "Whether sa-learn should rebuild the database every time it is called
+Enable this if you want sa-learn to rebuild the database automatically.  Doing
+this will slightly increase the running time of the spam registration process.
+If you choose not to do this, you will have to run \"sa-learn --rebuild\" in
+order for SpamAssassin to recognize the new registered spam."
+  :type 'boolean
+  :group 'spam-spamassassin)
+
+(defcustom spam-sa-learn-spam-switch "--spam"
+  "The switch that sa-learn uses to register spam messages"
+  :type 'string
+  :group 'spam-spamassassin)
+
+(defcustom spam-sa-learn-ham-switch "--ham"
+  "The switch that sa-learn uses to register ham messages"
+  :type 'string
+  :group 'spam-spamassassin)
+
+(defcustom spam-sa-learn-unregister-switch "--forget"
+  "The switch that sa-learn uses to unregister messages messages"
+  :type 'string
+  :group 'spam-spamassassin)
+
 ;;; Key bindings for spam control.
 
 (gnus-define-keys gnus-summary-mode-map
-  "St" spam-bogofilter-score
+  "St" spam-generic-score
   "Sx" gnus-summary-mark-as-spam
-  "Mst" spam-bogofilter-score
+  "Mst" spam-generic-score
   "Msx" gnus-summary-mark-as-spam
   "\M-d" gnus-summary-mark-as-spam)
 
@@ -504,16 +661,20 @@ finds ham or spam.")
 (defvar spam-list-of-processors
   '((gnus-group-spam-exit-processor-report-gmane spam spam-use-gmane)
     (gnus-group-spam-exit-processor-bogofilter   spam spam-use-bogofilter)
+    (gnus-group-spam-exit-processor-bsfilter    spam spam-use-bsfilter)
     (gnus-group-spam-exit-processor-blacklist    spam spam-use-blacklist)
     (gnus-group-spam-exit-processor-ifile        spam spam-use-ifile)
     (gnus-group-spam-exit-processor-stat         spam spam-use-stat)
     (gnus-group-spam-exit-processor-spamoracle   spam spam-use-spamoracle)
+    (gnus-group-spam-exit-processor-spamassassin spam spam-use-spamassassin)
     (gnus-group-ham-exit-processor-ifile         ham spam-use-ifile)
     (gnus-group-ham-exit-processor-bogofilter    ham spam-use-bogofilter)
+    (gnus-group-ham-exit-processor-bsfilter      ham spam-use-bsfilter)
     (gnus-group-ham-exit-processor-stat          ham spam-use-stat)
     (gnus-group-ham-exit-processor-whitelist     ham spam-use-whitelist)
     (gnus-group-ham-exit-processor-BBDB          ham spam-use-BBDB)
     (gnus-group-ham-exit-processor-copy          ham spam-use-ham-copy)
+    (gnus-group-ham-exit-processor-spamassassin  ham spam-use-spamassassin)
     (gnus-group-ham-exit-processor-spamoracle    ham spam-use-spamoracle))
   "The `spam-list-of-processors' list.
 This list contains pairs associating a ham/spam exit processor
@@ -588,6 +749,66 @@ Respects the process/prefix convention."
     (gnus-summary-remove-process-mark article)
     (spam-report-gmane article)))
 
+(defun spam-necessary-extra-headers ()
+  "Return the extra headers spam.el thinks are necessary."
+  (let (list)
+    (when (or spam-use-spamassassin
+             spam-use-spamassassin-headers
+             spam-use-regex-headers)
+      (push 'X-Spam-Status list))
+    list))
+
+(defun spam-user-format-function-S (headers)
+  (when headers
+    (spam-summary-score headers)))
+
+(defun spam-article-sort-by-spam-status (h1 h2)
+  "Sort articles by score."
+  (let (result)
+    (dolist (header (spam-necessary-extra-headers))
+      (let ((s1 (spam-summary-score h1 header))
+           (s2 (spam-summary-score h2 header)))
+      (unless (= s1 s2)
+       (setq result (< s1 s2))
+       (return))))
+    result))
+
+(defun spam-extra-header-to-number (header headers)
+  "Transform an extra header to a number."
+  (if (gnus-extra-header header headers)
+      (cond
+       ((eq header 'X-Spam-Status)
+       (string-to-number (gnus-replace-in-string
+                          (gnus-extra-header header headers)
+                          ".*hits=" "")))
+       (t nil))
+    nil))
+
+(defun spam-summary-score (headers &optional specific-header)
+  "Score an article for the summary buffer, as fast as possible.
+With SPECIFIC-HEADER, returns only that header's score.
+Will not return a nil score."
+  (let (score)
+    (dolist (header 
+            (if specific-header
+                (list specific-header)
+              (spam-necessary-extra-headers)))
+      (setq score 
+           (spam-extra-header-to-number header headers))
+      (when score 
+       (return)))
+    (or score 0)))
+
+(defun spam-generic-score (&optional recheck)
+  "Invoke whatever scoring method we can."
+  (interactive "P")
+  (cond
+   ((or spam-use-spamassassin spam-use-spamassassin-headers)
+    (spam-spamassassin-score recheck))
+   ((or spam-use-bsfilter spam-use-bsfilter-headers)
+    (spam-bsfilter-score recheck))
+   (t (spam-bogofilter-score recheck))))
+
 ;;; Summary entry and exit processing.
 
 (defun spam-summary-prepare ()
@@ -611,7 +832,7 @@ Respects the process/prefix convention."
             (new-articles (spam-list-articles
                            gnus-newsgroup-articles
                            classification))
-            (changed-articles (gnus-set-difference old-articles new-articles)))
+            (changed-articles (spam-set-difference new-articles old-articles)))
        ;; now that we have the changed articles, we go through the processors
        (dolist (processor-param spam-list-of-processors)
          (let ((processor (nth 0 processor-param))
@@ -626,7 +847,8 @@ Respects the process/prefix convention."
            ;; call spam-register-routine with specific articles to unregister,
            ;; when there are articles to unregister and the check is enabled
            (when (and unregister-list (symbol-value check))
-             (spam-register-routine classification check t unregister-list))))))
+             (spam-register-routine 
+              classification check t unregister-list))))))
 
     ;; find all the spam processors applicable to this group
     (dolist (processor-param spam-list-of-processors)
@@ -637,19 +859,21 @@ Respects the process/prefix convention."
                   (spam-group-processor-p gnus-newsgroup-name processor))
          (spam-register-routine classification check))))
 
-    (if spam-move-spam-nonspam-groups-only
-       (when (not (spam-group-spam-contents-p gnus-newsgroup-name))
-         (spam-mark-spam-as-expired-and-move-routine
-          (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
-      (gnus-message 5 "Marking spam as expired and moving it to %s"
-                   gnus-newsgroup-name)
+    (unless (and spam-move-spam-nonspam-groups-only
+                (spam-group-spam-contents-p gnus-newsgroup-name))
+      (gnus-message 6 "Marking spam as expired and moving it to %s"
+                   (gnus-parameter-spam-process-destination 
+                    gnus-newsgroup-name))
       (spam-mark-spam-as-expired-and-move-routine
        (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
 
     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
     ;; expire spam, in case the above did not expire them
-    (gnus-message 5 "Marking spam as expired without moving it")
-    (spam-mark-spam-as-expired-and-move-routine nil)
+    (when (< 0 (spam-list-articles
+               gnus-newsgroup-articles
+               'spam))
+      (gnus-message 6 "Marking spam as expired without moving it")
+      (spam-mark-spam-as-expired-and-move-routine nil))
 
     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
              (and (spam-group-spam-contents-p gnus-newsgroup-name)
@@ -665,24 +889,38 @@ Respects the process/prefix convention."
            (spam-register-routine classification check)))))
 
     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
-      (gnus-message 5 "Copying ham")
+      (gnus-message 6 "Copying ham")
       (spam-ham-copy-routine
        (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
 
     ;; now move all ham articles out of spam groups
     (when (spam-group-spam-contents-p gnus-newsgroup-name)
-      (gnus-message 5 "Moving ham messages from spam group")
+      (gnus-message 6 "Moving ham messages from spam group")
       (spam-ham-move-routine
        (gnus-parameter-ham-process-destination gnus-newsgroup-name))))
 
   (setq spam-old-ham-articles nil)
   (setq spam-old-spam-articles nil))
 
+(defun spam-set-difference (list1 list2)
+  "Return a set difference of LIST1 and LIST2.  
+When either list is nil, the other is returned."
+  (if (and list1 list2)
+      ;; we have two non-nil lists
+      (progn
+       (dolist (item (append list1 list2))
+         (when (and (memq item list1) (memq item list2))
+           (setq list1 (delq item list1))
+           (setq list2 (delq item list2))))
+       (append list1 list2))
+    ;; if either of the lists was nil, return the other one
+    (if list1 list1 list2)))
+
 (defun spam-mark-junk-as-spam-routine ()
   ;; check the global list of group names spam-junk-mailgroups and the
   ;; group parameters
   (when (spam-group-spam-contents-p gnus-newsgroup-name)
-    (gnus-message 5 "Marking %s articles as spam"
+    (gnus-message 6 "Marking %s articles as spam"
                  (if spam-mark-only-unseen-as-spam
                      "unseen"
                    "unread"))
@@ -788,22 +1026,12 @@ Respects the process/prefix convention."
     (spam-ham-copy-or-move-routine nil groups)))
 
 (defun spam-get-article-as-string (article)
-  (let ((article-buffer (spam-get-article-as-buffer article))
-       article-string)
-    (when article-buffer
-      (save-window-excursion
-       (set-buffer article-buffer)
-       (setq article-string (buffer-string))))
-    article-string))
-
-(defun spam-get-article-as-buffer (article)
-  (let ((article-buffer))
-    (when (numberp article)
-      (save-window-excursion
-       (gnus-summary-goto-subject article)
-       (gnus-summary-show-article t)
-       (setq article-buffer (get-buffer gnus-article-buffer))))
-    article-buffer))
+  (when (numberp article)
+    (with-temp-buffer
+      (gnus-request-article-this-buffer
+       article
+       gnus-newsgroup-name)
+      (buffer-string))))
 
 ;; disabled for now
 ;; (defun spam-get-article-as-filename (article)
@@ -840,7 +1068,7 @@ Respects the process/prefix convention."
          (mail-header-extra data-header))
         (t
          nil))
-       (gnus-error 5 "Article %d has a nil data header" article)))))
+       (gnus-message 6 "Article %d has a nil data header" article)))))
 
 (defun spam-fetch-field-from-fast (article &optional prepared-data-header)
   (spam-fetch-field-fast article 'from prepared-data-header))
@@ -866,7 +1094,7 @@ Respects the process/prefix convention."
          (spam-fetch-field-fast article 'xref dh))
         (when (spam-fetch-field-fast article 'extra dh)
           (format "%s\n" (spam-fetch-field-fast article 'extra dh))))
-      (gnus-error
+      (gnus-message
        5
        "spam-generate-fake-headers: article %d didn't have a valid header"
        article))))
@@ -874,24 +1102,31 @@ Respects the process/prefix convention."
 (defun spam-fetch-article-header (article)
   (save-excursion
     (set-buffer gnus-summary-buffer)
+    (gnus-read-header article)
     (nth 3 (assq article gnus-newsgroup-data))))
 
 \f
 ;;;; Spam determination.
 
 (defvar spam-list-of-checks
-  '((spam-use-blacklist         . spam-check-blacklist)
-    (spam-use-regex-headers     . spam-check-regex-headers)
-    (spam-use-regex-body        . spam-check-regex-body)
-    (spam-use-whitelist         . spam-check-whitelist)
-    (spam-use-BBDB              . spam-check-BBDB)
-    (spam-use-ifile             . spam-check-ifile)
-    (spam-use-spamoracle         . spam-check-spamoracle)
-    (spam-use-stat              . spam-check-stat)
-    (spam-use-blackholes        . spam-check-blackholes)
-    (spam-use-hashcash          . spam-check-hashcash)
-    (spam-use-bogofilter-headers . spam-check-bogofilter-headers)
-    (spam-use-bogofilter        . spam-check-bogofilter))
+  '((spam-use-blacklist                .       spam-check-blacklist)
+    (spam-use-regex-headers            .       spam-check-regex-headers)
+    (spam-use-gmane-xref               .       spam-check-gmane-xref)
+    (spam-use-regex-body               .       spam-check-regex-body)
+    (spam-use-whitelist                .       spam-check-whitelist)
+    (spam-use-BBDB                     .       spam-check-BBDB)
+    (spam-use-BBDB-exclusive           .       spam-check-BBDB)
+    (spam-use-ifile                    .       spam-check-ifile)
+    (spam-use-spamoracle               .       spam-check-spamoracle)
+    (spam-use-stat                     .       spam-check-stat)
+    (spam-use-blackholes               .       spam-check-blackholes)
+    (spam-use-hashcash                 .       spam-check-hashcash)
+    (spam-use-spamassassin-headers     .       spam-check-spamassassin-headers)
+    (spam-use-spamassassin             .       spam-check-spamassassin)
+    (spam-use-bogofilter-headers       .       spam-check-bogofilter-headers)
+    (spam-use-bogofilter               .       spam-check-bogofilter)
+    (spam-use-bsfilter-headers         .       spam-check-bsfilter-headers)
+    (spam-use-bsfilter                 .       spam-check-bsfilter))
   "The spam-list-of-checks list contains pairs associating a
 parameter variable with a spam checking function.  If the
 parameter variable is true, then the checking function is called,
@@ -911,7 +1146,9 @@ definitely a spam.")
     spam-use-regex-body
     spam-use-stat
     spam-use-bogofilter
+    spam-use-bsfilter
     spam-use-blackholes
+    spam-use-spamassassin
     spam-use-spamoracle)
   "The spam-list-of-statistical-checks list contains all the mail
 splitters that need to have the full message body available.
@@ -956,7 +1193,7 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
                         (and specific-checks (memq (car pair) specific-checks))
                         ;; or, given no specific checks, spam-use-CHECK is set
                         (and (null specific-checks) (symbol-value (car pair))))
-                   (gnus-message 5 "spam-split: calling the %s function"
+                   (gnus-message 6 "spam-split: calling the %s function"
                                  (symbol-name (cdr pair)))
                    (setq decision (funcall (cdr pair)))
                    ;; if we got a decision at all, save the current check
@@ -1006,7 +1243,7 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
             registry-lookup)
         
         (unless id
-          (gnus-error 5 "Article %d has no message ID!" article))
+          (gnus-message 6 "Article %d has no message ID!" article))
         
         (when (and id spam-log-to-registry)
           (setq registry-lookup (spam-log-registration-type id 'incoming))
@@ -1060,6 +1297,10 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
                         nil
                         spam-whitelist-unregister-routine
                         nil)
+    (spam-use-ham-copy   nil
+                        nil
+                        nil
+                        nil)
     (spam-use-BBDB      spam-BBDB-register-routine
                         nil
                         spam-BBDB-unregister-routine
@@ -1082,10 +1323,18 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
                         ;; does Gmane support unregistration?
                         nil
                         nil)
+    (spam-use-spamassassin spam-spamassassin-register-ham-routine
+                          spam-spamassassin-register-spam-routine
+                          spam-spamassassin-unregister-ham-routine
+                          spam-spamassassin-unregister-spam-routine)
     (spam-use-bogofilter spam-bogofilter-register-ham-routine
                         spam-bogofilter-register-spam-routine
                         spam-bogofilter-unregister-ham-routine
-                        spam-bogofilter-unregister-spam-routine))
+                        spam-bogofilter-unregister-spam-routine)
+    (spam-use-bsfilter  spam-bsfilter-register-ham-routine
+                        spam-bsfilter-register-spam-routine
+                        spam-bsfilter-unregister-ham-routine
+                        spam-bsfilter-unregister-spam-routine))
   "The spam-registration-functions list contains pairs
 associating a parameter variable with the ham and spam
 registration functions, and the ham and spam unregistration
@@ -1121,21 +1370,19 @@ functions")
   (let ((mark-check (if (eq classification 'spam)
                        'spam-group-spam-mark-p
                      'spam-group-ham-mark-p))
-       list mark-cache-yes mark-cache-no)
+       alist mark-cache-yes mark-cache-no)
     (dolist (article articles)
       (let ((mark (gnus-summary-article-mark article)))
-       (unless (memq mark mark-cache-no)
-         (if (memq mark mark-cache-yes)
-             (push article list)
-           ;; else, we have to actually check the mark
-           (if (funcall mark-check
-                        gnus-newsgroup-name
-                        mark)
-               (progn
-                 (push article list)
-                 (push mark mark-cache-yes))
-             (push mark mark-cache-no))))))
-    list))
+       (unless (or (memq mark mark-cache-yes)
+                   (memq mark mark-cache-no))
+         (if (funcall mark-check
+                      gnus-newsgroup-name
+                      mark)
+             (push mark mark-cache-yes)
+           (push mark mark-cache-no)))
+       (when (memq mark mark-cache-yes)
+         (push article alist))))
+    alist))
 
 (defun spam-register-routine (classification
                              check
@@ -1162,7 +1409,7 @@ functions")
                            gnus-newsgroup-articles
                            classification)))
        ;; process them
-       (gnus-message 5 "%s %d %s articles with classification %s, check %s"
+       (gnus-message 5 "%s %d %s articles as %s using backend %s"
                      (if unregister "Unregistering" "Registering")
                      (length articles)
                      (if specific-articles "specific" "")
@@ -1194,8 +1441,10 @@ functions")
           type
           cell-list))
 
-      (gnus-error 5 (format "%s called with bad ID, type, classification, check, or group"
-                           "spam-log-processing-to-registry")))))
+      (gnus-message 
+       5 
+       (format "%s call with bad ID, type, classification, spam-check, or group"
+              "spam-log-processing-to-registry")))))
 
 ;;; check if a ham- or spam-processor registration has been done
 (defun spam-log-registered-p (id type)
@@ -1204,8 +1453,10 @@ functions")
             (spam-process-type-valid-p type))
        (cdr-safe (gnus-registry-fetch-extra id type))
       (progn
-       (gnus-error 5 (format "%s called with bad ID, type, classification, or check"
-                             "spam-log-registered-p"))
+       (gnus-message
+        5 
+        (format "%s called with bad ID, type, classification, or spam-check"
+                "spam-log-registered-p"))
        nil))))
 
 ;;; check what a ham- or spam-processor registration says
@@ -1224,6 +1475,7 @@ functions")
        nil
       decision)))
 
+
 ;;; check if a ham- or spam-processor registration needs to be undone
 (defun spam-log-unregistration-needed-p (id type classification check)
   (when spam-log-to-registry
@@ -1240,8 +1492,10 @@ functions")
                (setq found t))))
          found)
       (progn
-       (gnus-error 5 (format "%s called with bad ID, type, classification, or check"
-                             "spam-log-unregistration-needed-p"))
+       (gnus-message
+        5 
+        (format "%s called with bad ID, type, classification, or spam-check"
+                "spam-log-unregistration-needed-p"))
        nil))))
 
 
@@ -1264,8 +1518,8 @@ functions")
           type
           new-cell-list))
       (progn
-       (gnus-error 5 (format "%s called with bad ID, type, check, or group"
-                             "spam-log-undo-registration"))
+       (gnus-message 6 (format "%s call with bad ID, type, spam-check, or group"
+                               "spam-log-undo-registration"))
        nil))))
 
 ;;; set up IMAP widening if it's necessary
@@ -1275,6 +1529,20 @@ functions")
       (setq nnimap-split-download-body-default t))))
 
 \f
+;;;; Gmane xrefs
+(defun spam-check-gmane-xref ()
+  (let ((header (or
+                (message-fetch-field "Xref")
+                (message-fetch-field "Newsgroups")))
+       (spam-split-group (if spam-split-symbolic-return
+                             'spam
+                           spam-split-group)))
+    (when header                       ; return nil when no header
+      (when (string-match spam-gmane-xref-spam-group
+                         header)
+         spam-split-group))))
+
+\f
 ;;;; Regex body
 
 (defun spam-check-regex-body ()
@@ -1326,7 +1594,7 @@ functions")
       (with-temp-buffer
        (insert headers)
        (goto-char (point-min))
-       (gnus-message 5 "Checking headers for relay addresses")
+       (gnus-message 6 "Checking headers for relay addresses")
        (while (re-search-forward
                "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
          (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
@@ -1344,13 +1612,13 @@ functions")
                (if spam-use-dig
                    (let ((query-result (query-dig query-string)))
                      (when query-result
-                       (gnus-message 5 "(DIG): positive blackhole check '%s'"
+                       (gnus-message 6 "(DIG): positive blackhole check '%s'"
                                      query-result)
                        (push (list ip server query-result)
                              matches)))
                  ;; else, if not using dig.el
                  (when (query-dns query-string)
-                   (gnus-message 5 "positive blackhole check")
+                   (gnus-message 6 "positive blackhole check")
                    (push (list ip server (query-dns query-string 'TXT))
                          matches)))))))))
     (when matches
@@ -1401,7 +1669,7 @@ functions")
                   (record (and net-address
                                (bbdb-search-simple nil net-address))))
              (when net-address
-               (gnus-message 5 "%s address %s %s BBDB"
+               (gnus-message 6 "%s address %s %s BBDB"
                              (if remove "Deleting" "Adding")
                              from
                              (if remove "from" "to"))
@@ -1731,7 +1999,7 @@ REMOVE not nil, remove the ADDRESSES."
         (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
        (remove-function
         (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
-       from addresses unregister-list)
+       from addresses unregister-list article-unregister-list)
     (dolist (article articles)
       (let ((from (spam-fetch-field-from-fast article))
            (id (spam-fetch-field-message-id-fast article))
@@ -1747,6 +2015,7 @@ REMOVE not nil, remove the ADDRESSES."
                 (null unregister)
                 (spam-log-unregistration-needed-p
                  id 'process declassification de-symbol))
+           (push article article-unregister-list)
            (push from unregister-list))
          (unless sender-ignored
            (push from addresses)))))
@@ -1755,7 +2024,7 @@ REMOVE not nil, remove the ADDRESSES."
        (funcall enter-function addresses t) ; unregister all these addresses
       ;; else, register normally and unregister what we need to
       (funcall remove-function unregister-list t)
-      (dolist (article unregister-list)
+      (dolist (article article-unregister-list)
        (spam-log-undo-registration
         (spam-fetch-field-message-id-fast article)
         'process
@@ -1799,17 +2068,18 @@ REMOVE not nil, remove the ADDRESSES."
          spam-split-group)))))
 
 ;; return something sensible if the score can't be determined
-(defun spam-bogofilter-score ()
+(defun spam-bogofilter-score (&optional recheck)
   "Get the Bogofilter spamicity score"
-  (interactive)
+  (interactive "P")
   (save-window-excursion
     (gnus-summary-show-article t)
     (set-buffer gnus-article-buffer)
-    (let ((score (or (spam-check-bogofilter-headers t)
+    (let ((score (or (unless recheck
+                      (spam-check-bogofilter-headers t))
                     (spam-check-bogofilter t))))
+      (gnus-summary-show-article)
       (message "Spamicity score %s" score)
-      (or score "0"))
-    (gnus-summary-show-article)))
+      (or score "0"))))
 
 (defun spam-check-bogofilter (&optional score)
   "Check the Bogofilter backend for the classification of this message"
@@ -1891,7 +2161,7 @@ REMOVE not nil, remove the ADDRESSES."
                  (goto-char (point-min))
                  (when (re-search-forward "^X-Spam: yes;" nil t)
                    spam-split-group))
-             (error "Error running spamoracle" status))))))))
+             (error "Error running spamoracle: %s" status))))))))
 
 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
   "Run spamoracle in training mode."
@@ -1913,8 +2183,8 @@ REMOVE not nil, remove the ADDRESSES."
                           `("-f" ,spam-spamoracle-database
                             "add" ,arg)
                         `("add" ,arg)))))
-         (when (not (eq 0 status))
-           (error "Error running spamoracle" status)))))))
+         (unless (eq 0 status)
+           (error "Error running spamoracle: %s" status)))))))
 
 (defun spam-spamoracle-learn-ham (articles &optional unregister)
   (spam-spamoracle-learn articles nil unregister))
@@ -1929,12 +2199,202 @@ REMOVE not nil, remove the ADDRESSES."
   (spam-spamoracle-learn-spam articles t))
 
 \f
+;;;; SpamAssassin
+;;; based mostly on the bogofilter code
+(defun spam-check-spamassassin-headers (&optional score)
+  "Check the SpamAssassin headers for the classification of this message."
+  (if score                            ; scoring mode
+      (let ((header (message-fetch-field spam-spamassassin-spam-status-header)))
+       (when header
+         (if (string-match "hits=\\(-?[0-9.]+\\)" header)
+             (match-string 1 header)
+           "0")))
+    ;; spam detection mode
+    (let ((header (message-fetch-field spam-spamassassin-spam-flag-header))
+         (spam-split-group (if spam-split-symbolic-return
+                                'spam
+                              spam-split-group)))
+         (when header                  ; return nil when no header
+           (when (string-match spam-spamassassin-positive-spam-flag-header
+                               header)
+             spam-split-group)))))
+
+(defun spam-check-spamassassin (&optional score)
+  "Check the SpamAssassin backend for the classification of this message."
+  (let ((article-buffer-name (buffer-name)))
+    (with-temp-buffer
+      (let ((temp-buffer-name (buffer-name)))
+       (save-excursion
+         (set-buffer article-buffer-name)
+         (apply 'call-process-region
+                (point-min) (point-max) spam-spamassassin-path
+                nil temp-buffer-name nil spam-spamassassin-arguments))
+       ;; check the return now (we're back in the temp buffer)
+       (goto-char (point-min))
+       (spam-check-spamassassin-headers score)))))
+
+;; return something sensible if the score can't be determined
+(defun spam-spamassassin-score (&optional recheck)
+  "Get the SpamAssassin score"
+  (interactive "P")
+  (save-window-excursion
+    (gnus-summary-show-article t)
+    (set-buffer gnus-article-buffer)
+    (let ((score (or (unless recheck
+                      (spam-check-spamassassin-headers t))
+                    (spam-check-spamassassin t))))
+      (gnus-summary-show-article)
+      (message "SpamAssassin score %s" score)
+      (or score "0"))))
+
+(defun spam-spamassassin-register-with-sa-learn (articles spam
+                                                &optional unregister)
+  "Register articles with spamassassin's sa-learn as spam or non-spam."
+  (if articles
+      (let ((action (if unregister spam-sa-learn-unregister-switch
+                     (if spam spam-sa-learn-spam-switch
+                       spam-sa-learn-ham-switch)))
+           (summary-buffer-name (buffer-name)))
+       (with-temp-buffer
+         ;; group the articles into mbox format
+         (dolist (article articles)
+           (let (article-string)
+             (save-excursion
+               (set-buffer summary-buffer-name)
+               (setq article-string (spam-get-article-as-string article)))
+             (when (stringp article-string)
+               (insert "From \n") ; mbox separator (sa-learn only checks the
+                                  ; first five chars, so we can get away with
+                                  ; a bogus line))
+               (insert article-string)
+               (insert "\n"))))
+         ;; call sa-learn on all messages at the same time
+         (apply 'call-process-region
+                (point-min) (point-max)
+                spam-sa-learn-path
+                nil nil nil "--mbox"
+                (if spam-sa-learn-rebuild
+                    (list action)
+                  `("--no-rebuild" ,action)))))))
+
+(defun spam-spamassassin-register-spam-routine (articles &optional unregister)
+  (spam-spamassassin-register-with-sa-learn articles t unregister))
+
+(defun spam-spamassassin-register-ham-routine (articles &optional unregister)
+  (spam-spamassassin-register-with-sa-learn articles nil unregister))
+
+(defun spam-spamassassin-unregister-spam-routine (articles)
+  (spam-spamassassin-register-with-sa-learn articles t t))
+
+(defun spam-spamassassin-unregister-ham-routine (articles)
+  (spam-spamassassin-register-with-sa-learn articles nil t))
+
+\f
+;;;; Bsfilter
+;;; based mostly on the bogofilter code
+(defun spam-check-bsfilter-headers (&optional score)
+  (if score
+      (or (nnmail-fetch-field spam-bsfilter-probability-header)
+         "0")
+    (let ((header (nnmail-fetch-field spam-bsfilter-header))
+         (spam-split-group (if spam-split-symbolic-return
+                               'spam
+                             spam-split-group)))
+      (when header ; return nil when no header
+       (when (string-match "YES" header)
+         spam-split-group)))))
+
+;; return something sensible if the score can't be determined
+(defun spam-bsfilter-score (&optional recheck)
+  "Get the Bsfilter spamicity score"
+  (interactive "P")
+  (save-window-excursion
+    (gnus-summary-show-article t)
+    (set-buffer gnus-article-buffer)
+    (let ((score (or (unless recheck
+                      (spam-check-bsfilter-headers t))
+                    (spam-check-bsfilter t))))
+      (gnus-summary-show-article)
+      (message "Spamicity score %s" score)
+      (or score "0"))))
+
+(defun spam-check-bsfilter (&optional score)
+  "Check the Bsfilter backend for the classification of this message"
+  (let ((article-buffer-name (buffer-name))
+       (dir spam-bsfilter-database-directory)
+       return)
+    (with-temp-buffer
+      (let ((temp-buffer-name (buffer-name)))
+       (save-excursion
+         (set-buffer article-buffer-name)
+         (apply 'call-process-region
+                (point-min) (point-max)
+                spam-bsfilter-path
+                nil temp-buffer-name nil
+                "--pipe"
+                "--insert-flag"
+                "--insert-probability"
+                (when dir
+                  (list "--homedir" dir))))
+       (setq return (spam-check-bsfilter-headers score))))
+    return))
+
+(defun spam-bsfilter-register-with-bsfilter (articles
+                                            spam
+                                            &optional unregister)
+  "Register an article, given as a string, as spam or non-spam."
+  (dolist (article articles)
+    (let ((article-string (spam-get-article-as-string article))
+         (switch (if unregister
+                     (if spam
+                         spam-bsfilter-spam-strong-switch
+                       spam-bsfilter-ham-strong-switch)
+                   (if spam
+                       spam-bsfilter-spam-switch
+                     spam-bsfilter-ham-switch))))
+      (when (stringp article-string)
+       (with-temp-buffer
+         (insert article-string)
+         (apply 'call-process-region
+                (point-min) (point-max)
+                spam-bsfilter-path
+                nil nil nil switch
+                "--update"
+                (when spam-bsfilter-database-directory
+                  (list "--homedir"
+                        spam-bsfilter-database-directory))))))))
+
+(defun spam-bsfilter-register-spam-routine (articles &optional unregister)
+  (spam-bsfilter-register-with-bsfilter articles t unregister))
+
+(defun spam-bsfilter-unregister-spam-routine (articles)
+  (spam-bsfilter-register-spam-routine articles t))
+
+(defun spam-bsfilter-register-ham-routine (articles &optional unregister)
+  (spam-bsfilter-register-with-bsfilter articles nil unregister))
+
+(defun spam-bsfilter-unregister-ham-routine (articles)
+  (spam-bsfilter-register-ham-routine articles t))
+
+\f
 ;;;; Hooks
 
 ;;;###autoload
-(defun spam-initialize ()
-  "Install the spam.el hooks and do other initialization"
+(defun spam-initialize (&rest symbols)
+  "Install the spam.el hooks and do other initialization.
+When SYMBOLS is given, set those variables to t.  This is so you
+can call spam-initialize before you set spam-use-* variables on
+explicitly, and matters only if you need the extra headers
+installed through spam-necessary-extra-headers."
   (interactive)
+
+  (dolist (var symbols)
+    (set var t))
+
+  (dolist (header (spam-necessary-extra-headers))
+    (add-to-list 'nnmail-extra-headers header)
+    (add-to-list 'gnus-extra-headers header))
+
   (setq spam-install-hooks t)
   ;; TODO: How do we redo this every time spam-face is customized?
   (push '((eq mark gnus-spam-mark) . spam-face)