* nnmail.el (nnmail-cache-insert): make sure that the
[gnus] / lisp / spam.el
index 1b3bfb4..783cc2e 100644 (file)
 (eval-and-compile
   (autoload 'spam-report-gmane "spam-report"))
 
+;; autoload gnus-registry
+(eval-and-compile
+  (autoload 'gnus-registry-store-extra-entry "gnus-registry")
+  (autoload 'gnus-registry-fetch-extra "gnus-registry"))
+
 ;; autoload query-dns
 (eval-and-compile
   (autoload 'query-dns "dns"))
@@ -74,8 +79,8 @@
 
 (defcustom spam-move-spam-nonspam-groups-only t
   "Whether spam should be moved in non-spam groups only.
-When nil, only ham and unclassified groups will have their spam moved
-to the spam-process-destination.  When t, spam will also be moved from
+When t, only ham and unclassified groups will have their spam moved
+to the spam-process-destination.  When nil, spam will also be moved from
 spam groups."
   :type 'boolean
   :group 'spam)
@@ -85,6 +90,11 @@ spam groups."
   :type 'boolean
   :group 'spam)
 
+(defcustom spam-log-to-registry nil
+  "Whether spam/ham processing should be logged in the registry."
+  :type 'boolean
+  :group 'spam)
+
 (defcustom spam-process-ham-in-spam-groups nil
   "Whether ham should be processed in spam groups."
   :type 'boolean
@@ -221,7 +231,7 @@ considered spam."
                               spam-use-spamoracle)
   "Whether the spam hooks should be installed, default to t if one of
 the spam-use-* variables is set."
-  :group 'gnus-registry
+  :group 'spam
   :type 'boolean)
 
 (defcustom spam-split-group "spam"
@@ -345,7 +355,7 @@ your main source of newsgroup names."
   :group 'spam-ifile)
 
 (defgroup spam-spamoracle nil
-  "Spam ifile configuration."
+  "Spam spamoracle configuration."
   :group 'spam)
 
 (defcustom spam-spamoracle-database nil 
@@ -525,8 +535,8 @@ spamoracle database."
 
     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
       (gnus-message 5 "Copying ham")
-      (spam-ham-move-routine
-       (gnus-parameter-ham-process-destination gnus-newsgroup-name) t))
+      (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)
@@ -548,45 +558,89 @@ spamoracle database."
       (dolist (article articles)
        (gnus-summary-mark-article article gnus-spam-mark)))))
 
-(defun spam-mark-spam-as-expired-and-move-routine (&optional group)
+(defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
   (gnus-summary-kill-process-mark)
   (let ((articles gnus-newsgroup-articles)
-       article tomove)
+       (backend-supports-deletions
+        (gnus-check-backend-function
+         'request-move-article gnus-newsgroup-name))
+       article tomove deletep)
     (dolist (article articles)
       (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
        (gnus-summary-mark-article article gnus-expirable-mark)
        (push article tomove)))
-
-    ;; now do the actual move
-    (when (and tomove
-              (stringp group))
+    
+    ;; now do the actual copies
+    (dolist (group groups)
+      (when (and tomove
+                (stringp group))
+       (dolist (article tomove)
+         (gnus-summary-set-process-mark article))
+       (when tomove
+         (if (or (not backend-supports-deletions)
+               (> (length groups) 1))
+             (progn 
+               (gnus-summary-copy-article nil group)
+               (setq deletep t))
+           (gnus-summary-move-article nil group)))))
+    
+    ;; now delete the articles, if there was a copy done, and the
+    ;; backend allows it
+    (when (and deletep backend-supports-deletions)
       (dolist (article tomove)
        (gnus-summary-set-process-mark article))
-      (when tomove (gnus-summary-move-article nil group))))
-  (gnus-summary-yank-process-mark))
+      (when tomove
+       (let ((gnus-novice-user nil))   ; don't ask me if I'm sure
+         (gnus-summary-delete-article nil))))
+    
+    (gnus-summary-yank-process-mark)))
  
-(defun spam-ham-move-routine (&optional group copy)
+(defun spam-ham-copy-or-move-routine (copy groups)
   (gnus-summary-kill-process-mark)
   (let ((articles gnus-newsgroup-articles)
-       article mark tomove)
-    (when (stringp group)              ; this routine will do nothing
-                                       ; without a valid group
-      (dolist (article articles)
-       (when (spam-group-ham-mark-p gnus-newsgroup-name
-                                    (gnus-summary-article-mark article))
-         (push article tomove)))
+       (backend-supports-deletions
+        (gnus-check-backend-function
+         'request-move-article gnus-newsgroup-name))
+       article mark todo deletep)
+    (dolist (article articles)
+      (when (spam-group-ham-mark-p gnus-newsgroup-name
+                                  (gnus-summary-article-mark article))
+       (push article todo)))
 
-      ;; now do the actual move
-      (when tomove
-       (dolist (article tomove)
+    ;; now do the actual move
+    (dolist (group groups)
+      (when (and todo (stringp group))
+       (dolist (article todo)
          (when spam-mark-ham-unread-before-move-from-spam-group
-           (gnus-summary-mark-article article gnus-unread-mark))           
+           (gnus-summary-mark-article article gnus-unread-mark))
+         (gnus-summary-set-process-mark article))
+
+       (if (or (not backend-supports-deletions)
+               (> (length groups) 1))
+           (progn 
+             (gnus-summary-copy-article nil group)
+             (setq deletep t))
+         (gnus-summary-move-article nil group))))
+    
+    ;; now delete the articles, unless a) copy is t, and there was a copy done
+    ;;                                 b) a move was done to a single group
+    ;;                                 c) backend-supports-deletions is nil
+    (unless copy
+      (when (and deletep backend-supports-deletions)
+       (dolist (article todo)
          (gnus-summary-set-process-mark article))
-       (if copy
-           (gnus-summary-copy-article nil group)
-         (gnus-summary-move-article nil group)))))
+       (when todo
+         (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
+           (gnus-summary-delete-article nil))))))
+  
   (gnus-summary-yank-process-mark))
  
+(defun spam-ham-copy-routine (&rest groups)
+  (spam-ham-copy-or-move-routine t groups))
+(defun spam-ham-move-routine (&rest groups)
+  (spam-ham-copy-or-move-routine nil groups))
 (defun spam-generic-register-routine (spam-func ham-func)
   (let ((articles gnus-newsgroup-articles)
        article mark ham-articles spam-articles)
@@ -656,6 +710,13 @@ spamoracle database."
       (mail-header-subject (gnus-data-header (assoc article (gnus-data-list nil))))
     nil))
 
+(defun spam-fetch-field-message-id-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-message-id (gnus-data-header (assoc article (gnus-data-list nil))))
+    nil))
+
 \f
 ;;;; Spam determination.
 
@@ -692,33 +753,87 @@ splitters that need to have the full message body available.")
 ;;;TODO: modify to invoke self with each specific check if invoked without specific checks
 (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 `nnmail-split-fancy', for
-example like this: (: spam-split).  It can take checks as parameters.
+This function can be used as an entry in `nnmail-split-fancy',
+for example like this: (: spam-split).  It can take checks as
+parameters.  A string as a parameter will set the
+spam-split-group to that string.
 
 See the Info node `(gnus)Fancy Mail Splitting' for more details."
   (interactive)
-  (save-excursion
-    (save-restriction
-      (dolist (check spam-list-of-statistical-checks)
-       (when (symbol-value check)
-         (widen)
-         (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
-                       (symbol-name check))
-         (return)))
-      ;;   (progn (widen) (debug (buffer-string)))
-      (let ((list-of-checks spam-list-of-checks)
-           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)))
-             (gnus-message 5 "spam-split: calling the %s function" (symbol-name (cdr pair)))
-             (setq decision (funcall (cdr pair))))))
-       (if (eq decision t)
-           nil
-         decision)))))
-  
+  (let ((spam-split-group-choice spam-split-group))
+    (dolist (check specific-checks)
+      (when (stringp check)
+       (setq spam-split-group-choice check)
+       (setq specific-checks (delq check specific-checks))))
+
+    (let ((spam-split-group spam-split-group-choice))
+      (save-excursion
+       (save-restriction
+         (dolist (check spam-list-of-statistical-checks)
+           (when (and (symbolp check) (symbol-value check))
+             (widen)
+             (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
+                           (symbol-name check))
+             (return)))
+         ;;   (progn (widen) (debug (buffer-string)))
+         (let ((list-of-checks spam-list-of-checks)
+               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)))
+                 (gnus-message 5 "spam-split: calling the %s function" 
+                               (symbol-name (cdr pair)))
+                 (setq decision (funcall (cdr pair))))))
+           (if (eq decision t)
+               nil
+             decision)))))))
+
+;;; log a ham- or spam-processor invocation to the registry
+(defun spam-log-processing-to-registry (id type classification check group)
+  (when spam-log-to-registry
+    (if (and (stringp id)
+            (stringp group)
+            (or (eq type 'incoming)
+                (eq type 'process))
+            (or (eq classification 'spam)
+                (eq classification 'ham))
+            (assoc check spam-list-of-checks))
+       (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
+              (cell (list classification check group)))
+         (push cell cell-list)
+         (gnus-registry-store-extra-entry
+          id
+          type
+          cell-list))
+
+      (gnus-message 5 (format "%s called with bad ID, type, check, or group"
+                             "spam-log-processing-to-registry")))))
+
+;;; 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
+    (if (and (stringp id)
+            (or (eq type 'incoming)
+                (eq type 'process))
+            (or (eq classification 'spam)
+                (eq classification 'ham))
+            (assoc check spam-list-of-checks))
+       (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
+             found)
+         (dolist (cell cell-list)
+           (unless found
+             (when (and (eq classification (nth 0 cell))
+                        (eq check (nth 1 cell)))
+               (setq found t))))
+         found)
+      (progn 
+       (gnus-message 5 (format "%s called with bad ID, type, check, or group"
+                               "spam-log-unregistration-needed-p"))
+       nil))))
+
+;;; set up IMAP widening if it's necessary  
 (defun spam-setup-widening ()
   (dolist (check spam-list-of-statistical-checks)
     (when (symbol-value check)
@@ -756,6 +871,12 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
 \f
 ;;;; Blackholes.
 
+(defun spam-reverse-ip-string (ip)
+  (when (stringp ip)
+    (mapconcat 'identity
+              (nreverse (split-string ip "\\."))
+              ".")))
+
 (defun spam-check-blackholes ()
   "Check the Received headers for blackholed relays."
   (let ((headers (nnmail-fetch-field "received"))
@@ -766,16 +887,17 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
        (goto-char (point-min))
        (gnus-message 5 "Checking headers for relay addresses")
        (while (re-search-forward
-               "\\[\\([0-9]+.[0-9]+.[0-9]+.[0-9]+\\)\\]" nil t)
+               "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
          (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
-         (push (mapconcat 'identity
-                          (nreverse (split-string (match-string 1) "\\."))
-                          ".")
+         (push (spam-reverse-ip-string (match-string 1))
                ips)))
       (dolist (server spam-blackhole-servers)
        (dolist (ip ips)
          (unless (and spam-blackhole-good-server-regex
-                      (string-match spam-blackhole-good-server-regex ip))
+                      ;; match the good-server-regex against the reversed (again) IP string
+                      (string-match 
+                       spam-blackhole-good-server-regex
+                       (spam-reverse-ip-string ip)))
            (unless matches
              (let ((query-string (concat ip "." server)))
                (if spam-use-dig
@@ -837,6 +959,12 @@ See the Info node `(gnus)Fancy Mail Splitting' for more details."
      nil
      ;; ham function
      (lambda (article)
+       (spam-log-processing-to-registry
+       (spam-fetch-field-message-id-fast article)
+       'process
+       'ham
+       'spam-use-BBDB 
+       gnus-newsgroup-name)
        (spam-enter-ham-BBDB (spam-fetch-field-from-fast article)))))
 
   (defun spam-check-BBDB ()
@@ -914,6 +1042,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
 (defun spam-ifile-register-spam-routine ()
   (spam-generic-register-routine 
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'spam
+      'spam-use-ifile
+      gnus-newsgroup-name)
      (spam-ifile-register-with-ifile 
       (spam-get-article-as-string article) spam-ifile-spam-category))
    nil))
@@ -922,6 +1056,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
   (spam-generic-register-routine 
    nil
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'ham
+      'spam-use-ifile
+      gnus-newsgroup-name)
      (spam-ifile-register-with-ifile 
       (spam-get-article-as-string article) spam-ifile-ham-category))))
 
@@ -943,6 +1083,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
       (defun spam-stat-register-spam-routine ()
        (spam-generic-register-routine 
         (lambda (article)
+          (spam-log-processing-to-registry 
+           (spam-fetch-field-message-id-fast article)
+           'process
+           'spam
+           'spam-use-stat
+           gnus-newsgroup-name)
           (let ((article-string (spam-get-article-as-string article)))
             (with-temp-buffer
               (insert article-string)
@@ -953,6 +1099,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
        (spam-generic-register-routine 
         nil
         (lambda (article)
+          (spam-log-processing-to-registry 
+           (spam-fetch-field-message-id-fast article)
+           'process
+           'ham
+           'spam-use-stat
+           gnus-newsgroup-name)
           (let ((article-string (spam-get-article-as-string article)))
             (with-temp-buffer
               (insert article-string)
@@ -1060,6 +1212,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
   (spam-generic-register-routine 
    ;; the spam function
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'spam
+      'spam-use-blacklist
+      gnus-newsgroup-name)
      (let ((from (spam-fetch-field-from-fast article)))
        (when (stringp from)
           (spam-enter-blacklist from))))
@@ -1072,6 +1230,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
    nil 
    ;; the ham function
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'ham
+      'spam-use-whitelist
+      gnus-newsgroup-name)
      (let ((from (spam-fetch-field-from-fast article)))
        (when (stringp from)
           (spam-enter-whitelist from))))))
@@ -1146,6 +1310,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
 (defun spam-bogofilter-register-spam-routine ()
   (spam-generic-register-routine 
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'spam
+      'spam-use-bogofilter
+      gnus-newsgroup-name)
      (spam-bogofilter-register-with-bogofilter
       (spam-get-article-as-string article) t))
    nil))
@@ -1154,6 +1324,12 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
   (spam-generic-register-routine 
    nil
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'ham
+      'spam-use-bogofilter
+      gnus-newsgroup-name)
      (spam-bogofilter-register-with-bogofilter
       (spam-get-article-as-string article) nil))))
 
@@ -1206,11 +1382,23 @@ Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
   (spam-generic-register-routine 
    nil
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'ham
+      'spam-use-spamoracle
+      gnus-newsgroup-name)
      (spam-spamoracle-learn article nil))))
 
 (defun spam-spamoracle-learn-spam ()
   (spam-generic-register-routine 
    (lambda (article)
+     (spam-log-processing-to-registry 
+      (spam-fetch-field-message-id-fast article)
+      'process
+      'spam
+      'spam-use-spamoracle
+      gnus-newsgroup-name)
      (spam-spamoracle-learn article t))
    nil))
 \f