Synch to No Gnus 200402092051.
[elisp/gnus.git-] / lisp / spam.el
index 235409b..2dec99b 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)
@@ -132,9 +130,9 @@ Competition."
   :group 'spam)
 
 (defcustom spam-disable-spam-split-during-ham-respool nil
-  "Whether `spam-split' should be ignored while resplitting ham in a process
-destination.  This is useful to prevent ham from ending up in the same spam
-group after the resplit.  Don't set this to t if you have spam-split as the
+  "Whether `spam-split' should be ignored while resplitting ham.
+This is useful to prevent ham from ending up in the same spam
+group after the resplit.  Don't set this to t if you have `spam-split' as the
 last rule in your split configuration."
   :type 'boolean
   :group 'spam)
@@ -247,6 +245,19 @@ 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-blacklist
@@ -256,8 +267,10 @@ 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-BBDB
                               spam-use-BBDB-exclusive
                               spam-use-ifile
@@ -407,8 +420,8 @@ your main source of newsgroup names."
   :group 'spam)
 
 (defcustom spam-spamoracle-database nil
-  "Location of spamoracle database file. When nil, use the default
-spamoracle database."
+  "Location of spamoracle database file.
+When nil, use the default spamoracle database."
   :type '(choice (directory :tag "Location of spamoracle database file.")
                 (const :tag "Use the default"))
   :group 'spam-spamoracle)
@@ -419,15 +432,87 @@ 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)
 
+(defvar spam-cache-lookups t
+  "Whether spam.el will try to cache lookups using `spam-caches'.")
+
+(defvar spam-caches (make-hash-table
+                    :size 10
+                    :test 'equal)
+  "Cache of spam detection entries.")
+
 (defvar spam-old-ham-articles nil
   "List of old ham articles, generated when a group is entered.")
 
@@ -438,15 +523,21 @@ spamoracle database."
   "If non-nil, `spam-split' is disabled, and always returns nil.")
 
 (defvar spam-split-last-successful-check nil
-  "`spam-split' will set this to nil or a spam-use-XYZ check if it
-  finds ham or spam.")
+  "Internal variable.
+`spam-split' will set this to nil or a spam-use-XYZ check if it
+finds ham or spam.")
 
 ;; convenience functions
+(defun spam-clear-cache (symbol)
+  "Clear the spam-caches entry for a check."
+  (remhash symbol spam-caches))
+
 (defun spam-xor (a b)
-  "Logical exclusive `or'."
+  "Logical A xor B."
   (and (or a b) (not (and a b))))
 
 (defun spam-group-ham-mark-p (group mark &optional spam)
+  "Checks if MARK is considered a ham mark in GROUP."
   (when (stringp group)
     (let* ((marks (spam-group-ham-marks group spam))
           (marks (if (symbolp mark)
@@ -455,9 +546,11 @@ spamoracle database."
       (memq mark marks))))
 
 (defun spam-group-spam-mark-p (group mark)
+  "Checks if MARK is considered a spam mark in GROUP."
   (spam-group-ham-mark-p group mark t))
 
 (defun spam-group-ham-marks (group &optional spam)
+  "In GROUP, get all the ham marks."
   (when (stringp group)
     (let* ((marks (if spam
                      (gnus-parameter-spam-marks group)
@@ -467,9 +560,11 @@ spamoracle database."
       marks)))
 
 (defun spam-group-spam-marks (group)
+  "In GROUP, get all the spam marks."
   (spam-group-ham-marks group t))
 
 (defun spam-group-spam-contents-p (group)
+  "Is GROUP a spam group?"
   (if (stringp group)
       (or (member group spam-junk-mailgroups)
          (memq 'gnus-group-spam-classification-spam
@@ -477,6 +572,7 @@ spamoracle database."
     nil))
 
 (defun spam-group-ham-contents-p (group)
+  "Is GROUP a ham group?"
   (if (stringp group)
       (memq 'gnus-group-spam-classification-ham
            (gnus-parameter-spam-contents group))
@@ -489,16 +585,18 @@ spamoracle database."
     (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-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 contains pairs associating a
-ham/spam exit processor variable with a classification and a
-spam-use-* variable.")
+  "The `spam-list-of-processors' list.
+This list contains pairs associating a ham/spam exit processor
+variable with a classification and a spam-use-* variable.")
 
 (defun spam-group-processor-p (group processor)
   (if (and (stringp group)
@@ -561,6 +659,23 @@ spam-use-* variable.")
 (defun spam-group-ham-processor-spamoracle-p (group)
   (spam-group-processor-p group 'gnus-group-ham-exit-processor-spamoracle))
 
+(defun spam-report-articles-gmane (n)
+  "Report the current message as spam.
+Respects the process/prefix convention."
+  (interactive "P")
+  (dolist (article (gnus-summary-work-articles n))
+    (gnus-summary-remove-process-mark article)
+    (spam-report-gmane article)))
+
+(defun spam-generic-score ()
+  (interactive)
+  "Invoke whatever scoring method we can."
+  (if (or
+       spam-use-spamassassin
+       spam-use-spamassassin-headers)
+      (spam-spamassassin-score)
+    (spam-bogofilter-score)))
+
 ;;; Summary entry and exit processing.
 
 (defun spam-summary-prepare ()
@@ -761,22 +876,12 @@ spam-use-* variable.")
     (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)
@@ -790,48 +895,84 @@ spam-use-* variable.")
 ;;     article-filename
 ;;       nil)))
 
-(defun spam-fetch-field-from-fast (article)
-  "Fetch the `from' field quickly, using the internal gnus-data-list function"
-  (if (and (numberp article)
-          (assoc article (gnus-data-list nil)))
-      (mail-header-from
-       (gnus-data-header (assoc article (gnus-data-list nil))))
-    nil))
-
-(defun spam-fetch-field-subject-fast (article)
-  "Fetch the `subject' field quickly, using the internal
-  gnus-data-list function"
-  (if (and (numberp article)
-          (assoc article (gnus-data-list nil)))
-      (mail-header-subject
-       (gnus-data-header (assoc article (gnus-data-list nil))))
-    nil))
-
-(defun spam-fetch-field-message-id-fast (article)
-  "Fetch the `Message-ID' field quickly, using the internal
-  gnus-data-list function"
-  (if (and (numberp article)
-          (assoc article (gnus-data-list nil)))
-      (mail-header-message-id
-       (gnus-data-header (assoc article (gnus-data-list nil))))
-    nil))
+(defun spam-fetch-field-fast (article field &optional prepared-data-header)
+  "Fetch a field quickly, using the internal gnus-data-list function"
+  (when (numberp article)
+    (let* ((data-header (or prepared-data-header
+                           (spam-fetch-article-header article))))
+      (if (arrayp data-header)
+       (cond
+        ((equal field 'from)
+         (mail-header-from data-header))
+        ((equal field 'message-id)
+         (mail-header-message-id data-header))
+        ((equal field 'subject)
+         (mail-header-subject data-header))
+        ((equal field 'references)
+         (mail-header-references data-header))
+        ((equal field 'date)
+         (mail-header-date data-header))
+        ((equal field 'xref)
+         (mail-header-xref data-header))
+        ((equal field 'extra)
+         (mail-header-extra data-header))
+        (t
+         nil))
+       (gnus-error 5 "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))
+
+(defun spam-fetch-field-subject-fast (article &optional prepared-data-header)
+  (spam-fetch-field-fast article 'subject prepared-data-header))
+
+(defun spam-fetch-field-message-id-fast (article &optional prepared-data-header)
+  (spam-fetch-field-fast article 'message-id prepared-data-header))
+
+(defun spam-generate-fake-headers (article)
+  (let ((dh (spam-fetch-article-header article)))
+    (if dh
+       (concat
+        (format 
+         (concat "From: %s\nSubject: %s\nMessage-ID: %s\n"
+                 "Date: %s\nReferences: %s\nXref: %s\n")
+         (spam-fetch-field-fast article 'from dh)
+         (spam-fetch-field-fast article 'subject dh)
+         (spam-fetch-field-fast article 'message-id dh)
+         (spam-fetch-field-fast article 'date dh)
+         (spam-fetch-field-fast article 'references dh)
+         (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
+       5
+       "spam-generate-fake-headers: article %d didn't have a valid header"
+       article))))
+
+(defun spam-fetch-article-header (article)
+  (save-excursion
+    (set-buffer gnus-summary-buffer)
+    (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-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))
   "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,
@@ -851,11 +992,14 @@ definitely a spam.")
     spam-use-regex-body
     spam-use-stat
     spam-use-bogofilter
+    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.")
+splitters that need to have the full message body available.
+Note that you should fetch extra headers if you don't like this,
+e.g. fetch the 'Received' header for spam-use-blackholes.")
 
-;;;TODO: modify to invoke self with each check if invoked without specifics
 (defun spam-split (&rest specific-checks)
   "Split this message into the `spam' group if it is spam.
 This function can be used as an entry in the variable `nnmail-split-fancy',
@@ -877,7 +1021,9 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
        (save-excursion
          (save-restriction
            (dolist (check spam-list-of-statistical-checks)
-             (when (and (symbolp check) (symbol-value check))
+             (when (and (symbolp check)
+                        (or (symbol-value check)
+                            (memq check specific-checks)))
                (widen)
                (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
                              (symbol-name check))
@@ -887,9 +1033,11 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
                  decision)
              (while (and list-of-checks (not decision))
                (let ((pair (pop list-of-checks)))
-                 (when (and (symbol-value (car pair))
-                            (or (null specific-checks)
-                                (memq (car pair) specific-checks)))
+                 (when (or
+                        ;; either, given specific checks, this is one of them
+                        (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"
                                  (symbol-name (cdr pair)))
                    (setq decision (funcall (cdr pair)))
@@ -898,8 +1046,7 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
                      (setq spam-split-last-successful-check (car pair)))
 
                    (when (eq decision 'spam)
-                     (if spam-split-symbolic-return
-                         (setq decision spam-split-group)
+                     (unless spam-split-symbolic-return
                        (gnus-error
                         5
                         (format "spam-split got %s but %s is nil"
@@ -916,44 +1063,73 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
   (let* ((group gnus-newsgroup-name)
         (autodetect (gnus-parameter-spam-autodetect group))
         (methods (gnus-parameter-spam-autodetect-methods group))
-        (first-method (nth 0 methods)))
-  (when (and autodetect
-            (not (equal first-method 'none)))
+        (first-method (nth 0 methods))
+        (articles (if spam-autodetect-recheck-messages
+                      gnus-newsgroup-articles
+                    gnus-newsgroup-unseen))
+        article-cannot-be-faked)
+
+    (dolist (check spam-list-of-statistical-checks)
+      (when (and (symbolp check)
+                (memq check methods))
+       (setq article-cannot-be-faked t)
+       (return)))
+
+    (when (memq 'default methods)
+      (setq article-cannot-be-faked t))
+
+    (when (and autodetect
+              (not (equal first-method 'none)))
     (mapcar
      (lambda (article)
        (let ((id (spam-fetch-field-message-id-fast article))
             (subject (spam-fetch-field-subject-fast article))
-            (sender (spam-fetch-field-from-fast article)))
-        (unless (and spam-log-to-registry
-                     (spam-log-registered-p id 'incoming))
-          (let* ((spam-split-symbolic-return t)
-                 (spam-split-symbolic-return-positive t)
-                 (split-return
-                  (with-temp-buffer
-                    (gnus-request-article-this-buffer
-                     article
-                     group)
-                    (if (or (null first-method)
-                            (equal first-method 'default))
-                        (spam-split)
-                      (apply 'spam-split methods)))))
-            (if (equal split-return 'spam)
-                (gnus-summary-mark-article article gnus-spam-mark))
-
-            (when (and split-return spam-log-to-registry)
-              (when (zerop (gnus-registry-group-count id))
-                (gnus-registry-add-group
-                 id group subject sender))
-
+            (sender (spam-fetch-field-from-fast article))
+            registry-lookup)
+        
+        (unless id
+          (gnus-error 5 "Article %d has no message ID!" article))
+        
+        (when (and id spam-log-to-registry)
+          (setq registry-lookup (spam-log-registration-type id 'incoming))
+          (when registry-lookup
+            (gnus-message
+             9
+             "spam-find-spam: message %s was already registered incoming"
+             id)))
+
+        (let* ((spam-split-symbolic-return t)
+               (spam-split-symbolic-return-positive t)
+               (fake-headers (spam-generate-fake-headers article))
+               (split-return
+                (or registry-lookup
+                    (with-temp-buffer
+                      (if article-cannot-be-faked
+                          (gnus-request-article-this-buffer
+                           article
+                           group)
+                        ;; else, we fake the article
+                        (when fake-headers (insert fake-headers)))
+                      (if (or (null first-method)
+                              (equal first-method 'default))
+                          (spam-split)
+                        (apply 'spam-split methods))))))
+          (if (equal split-return 'spam)
+              (gnus-summary-mark-article article gnus-spam-mark))
+          
+          (when (and id split-return spam-log-to-registry)
+            (when (zerop (gnus-registry-group-count id))
+              (gnus-registry-add-group
+               id group subject sender))
+              
+            (unless registry-lookup
               (spam-log-processing-to-registry
                id
                'incoming
                split-return
                spam-split-last-successful-check
                group))))))
-     (if spam-autodetect-recheck-messages
-        gnus-newsgroup-articles
-       gnus-newsgroup-unseen)))))
+    articles))))
 
 (defvar spam-registration-functions
   ;; first the ham register, second the spam register function
@@ -988,6 +1164,10 @@ 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
@@ -1068,7 +1248,7 @@ 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" "")
@@ -1100,8 +1280,10 @@ functions")
           type
           cell-list))
 
-      (gnus-message 5 (format "%s called with bad ID, type, classification, check, or group"
-                             "spam-log-processing-to-registry")))))
+      (gnus-error 
+       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)
@@ -1110,10 +1292,28 @@ functions")
             (spam-process-type-valid-p type))
        (cdr-safe (gnus-registry-fetch-extra id type))
       (progn
-       (gnus-message 5 (format "%s called with bad ID, type, classification, or check"
-                               "spam-log-registered-p"))
+       (gnus-error 
+        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
+;;; returns nil if conflicting registrations are found
+(defun spam-log-registration-type (id type)
+  (let ((count 0)
+       decision)
+    (dolist (reg (spam-log-registered-p id type))
+      (let ((classification (nth 0 reg)))
+       (when (spam-classification-valid-p classification)
+         (when (and decision
+                    (not (eq classification decision)))
+           (setq count (+ 1 count)))
+         (setq decision classification))))
+    (if (< 0 count)
+       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
@@ -1130,8 +1330,10 @@ functions")
                (setq found t))))
          found)
       (progn
-       (gnus-message 5 (format "%s called with bad ID, type, classification, or check"
-                               "spam-log-unregistration-needed-p"))
+       (gnus-error 
+        5 
+        (format "%s called with bad ID, type, classification, or spam-check"
+                "spam-log-unregistration-needed-p"))
        nil))))
 
 
@@ -1154,8 +1356,8 @@ functions")
           type
           new-cell-list))
       (progn
-       (gnus-message 5 (format "%s called with bad ID, type, check, or group"
-                               "spam-log-undo-registration"))
+       (gnus-error 5 (format "%s call with bad ID, type, spam-check, or group"
+                             "spam-log-undo-registration"))
        nil))))
 
 ;;; set up IMAP widening if it's necessary
@@ -1207,7 +1409,7 @@ functions")
 
 (defun spam-check-blackholes ()
   "Check the Received headers for blackholed relays."
-  (let ((headers (nnmail-fetch-field "received"))
+  (let ((headers (message-fetch-field "received"))
        (spam-split-group (if spam-split-symbolic-return
                              'spam
                            spam-split-group))
@@ -1272,6 +1474,12 @@ functions")
       (require 'bbdb)
       (require 'bbdb-com)
 
+      ;; when the BBDB changes, we want to clear out our cache
+      (defun spam-clear-cache-BBDB (&rest immaterial)
+       (spam-clear-cache 'spam-use-BBDB))
+
+      (add-hook 'bbdb-change-hook 'spam-clear-cache-BBDB)
+
       (defun spam-enter-ham-BBDB (addresses &optional remove)
        "Enter an address into the BBDB; implies ham (non-spam) sender"
        (dolist (from addresses)
@@ -1307,13 +1515,30 @@ functions")
 
       (defun spam-check-BBDB ()
        "Mail from people in the BBDB is classified as ham or non-spam"
-       (let ((who (nnmail-fetch-field "from"))
+       (let ((who (message-fetch-field "from"))
              (spam-split-group (if spam-split-symbolic-return
                                    'spam
-                                 spam-split-group)))
+                                 spam-split-group))
+             bbdb-cache bbdb-hashtable)
+         (when spam-cache-lookups
+           (setq bbdb-cache (gethash 'spam-use-BBDB spam-caches))
+           (unless bbdb-cache
+             (setq bbdb-cache
+                   ;; this is the expanded (bbdb-hashtable) macro
+                   ;; without the debugging support
+                   (with-current-buffer (bbdb-buffer)
+                     (save-excursion
+                       (save-window-excursion
+                         (bbdb-records nil t)
+                         bbdb-hashtable))))
+             (puthash 'spam-use-BBDB bbdb-cache spam-caches)))
          (when who
            (setq who (nth 1 (gnus-extract-address-components who)))
-           (if (bbdb-search-simple nil who)
+           (if
+               (if spam-cache-lookups
+                   (symbol-value
+                    (intern-soft who bbdb-cache))
+                 (bbdb-search-simple nil who))
                t
              (if spam-use-BBDB-exclusive
                  spam-split-group
@@ -1321,6 +1546,8 @@ functions")
 
   (file-error (progn
                (defalias 'bbdb-search-simple 'ignore)
+               (defalias 'bbdb-records 'ignore)
+               (defalias 'bbdb-buffer 'ignore)
                (defalias 'spam-check-BBDB 'ignore)
                (defalias 'spam-BBDB-register-routine 'ignore)
                (defalias 'spam-enter-ham-BBDB 'ignore)
@@ -1481,7 +1708,8 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
 With a non-nil REMOVE, remove them."
   (interactive "sAddress: ")
   (spam-enter-list address spam-whitelist remove)
-  (setq spam-whitelist-cache nil))
+  (setq spam-whitelist-cache nil)
+  (spam-clear-cache 'spam-use-whitelist))
 
 ;;; address can be a list, too
 (defun spam-enter-blacklist (address &optional remove)
@@ -1489,7 +1717,8 @@ With a non-nil REMOVE, remove them."
 With a non-nil REMOVE, remove them."
   (interactive "sAddress: ")
   (spam-enter-list address spam-blacklist remove)
-  (setq spam-blacklist-cache nil))
+  (setq spam-blacklist-cache nil)
+  (spam-clear-cache 'spam-use-whitelist))
 
 (defun spam-enter-list (addresses file &optional remove)
   "Enter ADDRESSES into the given FILE.
@@ -1518,6 +1747,32 @@ REMOVE not nil, remove the ADDRESSES."
              (insert a "\n")))))
       (save-buffer))))
 
+(defun spam-filelist-build-cache (type)
+  (let ((cache (if (eq type 'spam-use-blacklist)
+                  spam-blacklist-cache
+                spam-whitelist-cache))
+       parsed-cache)
+    (unless (gethash type spam-caches)
+      (while cache
+       (let ((address (pop cache)))
+         (unless (zerop (length address)) ; 0 for a nil address too
+           (setq address (regexp-quote address))
+           ;; fix regexp-quote's treatment of user-intended regexes
+           (while (string-match "\\\\\\*" address)
+             (setq address (replace-match ".*" t t address))))
+         (push address parsed-cache)))
+      (puthash type parsed-cache spam-caches))))
+
+(defun spam-filelist-check-cache (type from)
+  (when (stringp from)
+    (spam-filelist-build-cache type)
+    (let (found)
+      (dolist (address (gethash type spam-caches))
+       (when (and address (string-match address from))
+         (setq found t)
+         (return)))
+      found)))
+
 ;;; returns t if the sender is in the whitelist, nil or
 ;;; spam-split-group otherwise
 (defun spam-check-whitelist ()
@@ -1527,7 +1782,7 @@ REMOVE not nil, remove the ADDRESSES."
                            spam-split-group)))
     (unless spam-whitelist-cache
       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
-    (if (spam-from-listed-p spam-whitelist-cache)
+    (if (spam-from-listed-p 'spam-use-whitelist)
        t
       (if spam-use-whitelist-exclusive
          spam-split-group
@@ -1540,7 +1795,7 @@ REMOVE not nil, remove the ADDRESSES."
                            spam-split-group)))
     (unless spam-blacklist-cache
       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
-    (and (spam-from-listed-p spam-blacklist-cache) spam-split-group)))
+    (and (spam-from-listed-p 'spam-use-blacklist) spam-split-group)))
 
 (defun spam-parse-list (file)
   (when (file-readable-p file)
@@ -1556,20 +1811,10 @@ REMOVE not nil, remove the ADDRESSES."
              (push (or pure-address address) contents)))))
       (nreverse contents))))
 
-(defun spam-from-listed-p (cache)
-  (let ((from (nnmail-fetch-field "from"))
+(defun spam-from-listed-p (type)
+  (let ((from (message-fetch-field "from"))
        found)
-    (while cache
-      (let ((address (pop cache)))
-       (unless (zerop (length address)) ; 0 for a nil address too
-         (setq address (regexp-quote address))
-         ;; fix regexp-quote's treatment of user-intended regexes
-         (while (string-match "\\\\\\*" address)
-           (setq address (replace-match ".*" t t address))))
-       (when (and address (string-match address from))
-         (setq found t
-               cache nil))))
-    found))
+    (spam-filelist-check-cache type from)))
 
 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
@@ -1631,7 +1876,7 @@ REMOVE not nil, remove the ADDRESSES."
 \f
 ;;;; Bogofilter
 (defun spam-check-bogofilter-headers (&optional score)
-  (let ((header (nnmail-fetch-field spam-bogofilter-header))
+  (let ((header (message-fetch-field spam-bogofilter-header))
        (spam-split-group (if spam-split-symbolic-return
                              'spam
                            spam-split-group)))
@@ -1654,9 +1899,9 @@ REMOVE not nil, remove the ADDRESSES."
     (set-buffer gnus-article-buffer)
     (let ((score (or (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"
@@ -1738,7 +1983,7 @@ 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."
@@ -1760,8 +2005,8 @@ 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))
@@ -1776,6 +2021,95 @@ 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 ()
+  "Get the SpamAssassin score"
+  (interactive)
+  (save-window-excursion
+    (gnus-summary-show-article t)
+    (set-buffer gnus-article-buffer)
+    (let ((score (or (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
 ;;;; Hooks
 
 ;;;###autoload
@@ -1793,7 +2127,7 @@ REMOVE not nil, remove the ADDRESSES."
   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
-  (add-hook 'gnus-summary-prepare-hook 'spam-find-spam))
+  (add-hook 'gnus-summary-prepared-hook 'spam-find-spam))
 
 (defun spam-unload-hook ()
   "Uninstall the spam.el hooks"
@@ -1811,8 +2145,4 @@ REMOVE not nil, remove the ADDRESSES."
 
 (provide 'spam)
 
-;;; spam.el ends here.
-
-(provide 'spam)
-
 ;;; spam.el ends here