2001-12-22 22:00:00 ShengHuo ZHU <zsh@cs.rochester.edu>

[gnus] / lisp / gnus-nocem.el

diff --git a/lisp/gnus-nocem.el b/lisp/gnus-nocem.el
index 312f3cb..d682451 100644 (file)
--- a/lisp/gnus-nocem.el
+++ b/lisp/gnus-nocem.el
@@ -1,7 +1,8 @@
 ;;; gnus-nocem.el --- NoCeM pseudo-cancellation treatment
 ;;; gnus-nocem.el --- NoCeM pseudo-cancellation treatment

-;; Copyright (C) 1995,96 Free Software Foundation, Inc.

 
 

-;; Author: Lars Magne Ingebrigtsen <larsi@ifi.uio.no>
+;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000 Free Software Foundation, Inc.
+
+;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>

 ;; Keywords: news
 
 ;; This file is part of GNU Emacs.
 ;; Keywords: news
 
 ;; This file is part of GNU Emacs.


@@ -25,28 +26,80 @@
 
 ;;; Code:
 
 
 ;;; Code:
 

+(eval-when-compile (require 'cl))
+

 (require 'gnus)
 (require 'nnmail)
 (require 'gnus)
 (require 'nnmail)

-(eval-when-compile (require 'cl))
+(require 'gnus-art)
+(require 'gnus-sum)
+(require 'gnus-range)
+
+(defgroup gnus-nocem nil
+  "NoCeM pseudo-cancellation treatment"
+  :group 'gnus-score)
+
+(defcustom gnus-nocem-groups
+  '("news.lists.filters" "news.admin.net-abuse.bulletins"
+    "alt.nocem.misc" "news.admin.net-abuse.announce")
+  "*List of groups that will be searched for NoCeM messages."
+  :group 'gnus-nocem
+  :type '(repeat (string :tag "Group")))
+
+(defcustom gnus-nocem-issuers
+  '("AutoMoose-1"                      ; CancelMoose[tm]
+    "clewis@ferret.ocunix"             ; Chris Lewis
+    "cosmo.roadkill"
+    "SpamHippo"
+    "hweede@snafu.de")
+  "*List of NoCeM issuers to pay attention to.
+
+This can also be a list of `(ISSUER CONDITION ...)' elements.
+
+See <URL:http://www.xs4all.nl/~rosalind/nocemreg/nocemreg.html> for an
+issuer registry."
+  :group 'gnus-nocem
+  :type '(repeat (choice string sexp)))
+
+(defcustom gnus-nocem-directory
+  (nnheader-concat gnus-article-save-directory "NoCeM/")
+  "*Directory where NoCeM files will be stored."
+  :group 'gnus-nocem
+  :type 'directory)
+
+(defcustom gnus-nocem-expiry-wait 15
+  "*Number of days to keep NoCeM headers in the cache."
+  :group 'gnus-nocem
+  :type 'integer)

 
 

-(defvar gnus-nocem-groups 
-  '("alt.nocem.misc" "news.admin.net-abuse.announce")
-  "*List of groups that will be searched for NoCeM messages.")
+(defcustom gnus-nocem-verifyer 'mc-verify
+  "*Function called to verify that the NoCeM message is valid.
+One likely value is `mc-verify'.  If the function in this variable
+isn't bound, the message will be used unconditionally."
+  :group 'gnus-nocem
+  :type '(radio (function-item mc-verify)
+               (function :tag "other")))

 
 

-(defvar gnus-nocem-issuers 
-  '("Automoose-1" ; The CancelMoose[tm] on autopilot.
-    "clewis@ferret.ocunix.on.ca;" ; Chris Lewis -- Canadian angel & despammer.
-    "jem@xpat.com;"  ; Jem -- Korean despammer.
-    "red@redpoll.mrfs.oh.us (Richard E. Depew)" ; Spew/bincancel guy.
-    )
-  "*List of NoCeM issuers to pay attention to.")
+(defcustom gnus-nocem-liberal-fetch nil
+  "*If t try to fetch all messages which have @@NCM in the subject.
+Otherwise don't fetch messages which have references or whose message-id
+matches an previously scanned and verified nocem message."
+  :group 'gnus-nocem
+  :type 'boolean)

 
 

-(defvar gnus-nocem-directory 
-  (concat (file-name-as-directory gnus-article-save-directory) "NoCeM/")
-  "*Directory where NoCeM files will be stored.")
+(defcustom gnus-nocem-check-article-limit 500
+  "*If non-nil, the maximum number of articles to check in any NoCeM group."
+  :group 'gnus-nocem
+  :version "21.1"
+  :type '(choice (const :tag "unlimited" nil)
+                (integer 1000)))

 
 

-(defvar gnus-nocem-expiry-wait 15
-  "*Number of days to keep NoCeM headers in the cache.")
+(defcustom gnus-nocem-check-from t
+  "Non-nil means check for valid issuers in message bodies.
+Otherwise don't bother fetching articles unless their author matches a
+valid issuer, which is much faster if you are selective about the issuers."
+  :group 'gnus-nocem
+  :version "21.1"
+  :type 'boolean)

 
 ;;; Internal variables
 
 
 ;;; Internal variables
 


@@ -54,6 +107,7 @@
 (defvar gnus-nocem-alist nil)
 (defvar gnus-nocem-touched-alist nil)
 (defvar gnus-nocem-hashtb nil)
 (defvar gnus-nocem-alist nil)
 (defvar gnus-nocem-touched-alist nil)
 (defvar gnus-nocem-hashtb nil)

+(defvar gnus-nocem-seen-message-ids nil)

 
 ;;; Functions
 
 
 ;;; Functions
 


@@ -63,53 +117,103 @@
 (defun gnus-nocem-cache-file ()
   (concat (file-name-as-directory gnus-nocem-directory) "cache"))
 
 (defun gnus-nocem-cache-file ()
   (concat (file-name-as-directory gnus-nocem-directory) "cache"))
 

+;;
+;; faster lookups for group names:
+;;
+
+(defvar gnus-nocem-real-group-hashtb nil
+  "Real-name mappings of subscribed groups.")
+
+(defun gnus-fill-real-hashtb ()
+  "Fill up a hash table with the real-name mappings from the user's active file."
+  (setq gnus-nocem-real-group-hashtb (gnus-make-hashtable
+                                     (length gnus-newsrc-alist)))
+  (mapcar (lambda (group)
+           (setq group (gnus-group-real-name (car group)))
+           (gnus-sethash group t gnus-nocem-real-group-hashtb))
+         gnus-newsrc-alist))
+

 (defun gnus-nocem-scan-groups ()
   "Scan all NoCeM groups for new NoCeM messages."
   (interactive)
   (let ((groups gnus-nocem-groups)
 (defun gnus-nocem-scan-groups ()
   "Scan all NoCeM groups for new NoCeM messages."
   (interactive)
   (let ((groups gnus-nocem-groups)

-       group active gactive articles)
-    (or (file-exists-p gnus-nocem-directory)
-       (make-directory gnus-nocem-directory t))
+       (gnus-inhibit-demon t)
+       group active gactive articles check-headers)
+    (gnus-make-directory gnus-nocem-directory)

     ;; Load any previous NoCeM headers.
     (gnus-nocem-load-cache)
     ;; Load any previous NoCeM headers.
     (gnus-nocem-load-cache)

+    ;; Get the group name mappings:
+    (gnus-fill-real-hashtb)

     ;; Read the active file if it hasn't been read yet.
     (and (file-exists-p (gnus-nocem-active-file))
         (not gnus-nocem-active)
     ;; Read the active file if it hasn't been read yet.
     (and (file-exists-p (gnus-nocem-active-file))
         (not gnus-nocem-active)

-        (condition-case ()
-            (load (gnus-nocem-active-file) t t t)
-          (error nil)))
+        (ignore-errors
+          (load (gnus-nocem-active-file) t t t)))

     ;; Go through all groups and see whether new articles have
     ;; Go through all groups and see whether new articles have

-    ;; arrived.  
+    ;; arrived.

     (while (setq group (pop groups))
       (if (not (setq gactive (gnus-activate-group group)))
     (while (setq group (pop groups))
       (if (not (setq gactive (gnus-activate-group group)))

-         () ; This group doesn't exist.
+         ()                            ; This group doesn't exist.

        (setq active (nth 1 (assoc group gnus-nocem-active)))
        (when (and (not (< (cdr gactive) (car gactive))) ; Empty group.
                   (or (not active)
                       (< (cdr active) (cdr gactive))))
          ;; Ok, there are new articles in this group, se we fetch the
          ;; headers.
        (setq active (nth 1 (assoc group gnus-nocem-active)))
        (when (and (not (< (cdr gactive) (car gactive))) ; Empty group.
                   (or (not active)
                       (< (cdr active) (cdr gactive))))
          ;; Ok, there are new articles in this group, se we fetch the
          ;; headers.

-         (let ((gnus-newsgroup-dependencies (make-vector 10 nil))
-               headers)
-           (setq headers
-                 (if (eq 'nov
-                         (gnus-retrieve-headers 
-                          (setq articles
-                                (gnus-uncompress-range
-                                 (cons 
-                                  (if active (1+ (cdr active)) (car gactive))
-                                  (cdr gactive))))
-                          group))
-                     (gnus-get-newsgroup-headers-xover articles)
-                   (gnus-get-newsgroup-headers)))
-           (while headers
-             ;; We take a closer look on all articles that have
-             ;; "@@NCM" in the subject.  
-             (when (string-match "@@NCM" (mail-header-subject (car headers)))
-               (gnus-nocem-check-article group (car headers)))
-             (setq headers (cdr headers)))))
+         (save-excursion
+           (let ((dependencies (make-vector 10 nil))
+                 headers header)
+             (with-temp-buffer
+               (setq headers
+                     (if (eq 'nov
+                             (gnus-retrieve-headers
+                              (setq articles
+                                    (gnus-uncompress-range
+                                     (cons
+                                      (if active (1+ (cdr active))
+                                        (car gactive))
+                                      (cdr gactive))))
+                              group))
+                         (gnus-get-newsgroup-headers-xover
+                          articles nil dependencies)
+                       (gnus-get-newsgroup-headers dependencies)))
+               (while (setq header (pop headers))
+                 ;; We take a closer look on all articles that have
+                 ;; "@@NCM" in the subject.  Unless we already read
+                 ;; this cross posted message.  Nocem messages
+                 ;; are not allowed to have references, so we can
+                 ;; ignore scanning followups.
+                 (and (string-match "@@NCM" (mail-header-subject header))
+                      (and gnus-nocem-check-from
+                           (let ((case-fold-search t))
+                             (catch 'ok
+                               (mapcar
+                                (lambda (author)
+                                  (if (consp author)
+                                      (setq author (car author)))
+                                  (if (string-match
+                                       author (mail-header-from header))
+                                      (throw 'ok t)))
+                                gnus-nocem-issuers)
+                               nil)))
+                      (or gnus-nocem-liberal-fetch
+                          (and (or (string= "" (mail-header-references
+                                                header))
+                                   (null (mail-header-references header)))
+                               (not (member (mail-header-message-id header)
+                                            gnus-nocem-seen-message-ids))))
+                      (push header check-headers)))
+               (let* ((i 0)
+                      (check-headers
+                       (last check-headers gnus-nocem-check-article-limit))
+                      (len (length check-headers)))
+                 (dolist (h check-headers)
+                   (gnus-message
+                    7 "Checking article %d in %s for NoCeM (%d of %d)..."
+                    (mail-header-number h) group (incf i) len)
+                   (gnus-nocem-check-article group h)))))))

        (setq gnus-nocem-active
        (setq gnus-nocem-active

-             (cons (list group gactive) 
+             (cons (list group gactive)

                    (delq (assoc group gnus-nocem-active)
                          gnus-nocem-active)))))
     ;; Save the results, if any.
                    (delq (assoc group gnus-nocem-active)
                          gnus-nocem-active)))))
     ;; Save the results, if any.


@@ -118,93 +222,142 @@
 
 (defun gnus-nocem-check-article (group header)
   "Check whether the current article is an NCM article and that we want it."
 
 (defun gnus-nocem-check-article (group header)
   "Check whether the current article is an NCM article and that we want it."

-  (nnheader-temp-write nil
-    ;; Get the article.
-    (gnus-message 7 "Checking article %d in %s for NoCeM..."
-                 (mail-header-number header) group)
-    (let ((date (mail-header-date header))
-         issuer b e)
-      (when (or (not date)
-               (nnmail-time-less 
-                (nnmail-time-since (nnmail-date-to-time date))
-                (nnmail-days-to-time gnus-nocem-expiry-wait)))
-       (gnus-request-article-this-buffer (mail-header-number header) group)
-       ;; The article has to have proper NoCeM headers.
-       (when (and (setq b (search-forward "\n@@BEGIN NCM HEADERS\n" nil t))
-                  (setq e (search-forward "\n@@BEGIN NCM BODY\n" nil t)))
-         ;; We get the name of the issuer.
-         (narrow-to-region b e)
-         (setq issuer (mail-fetch-field "issuer"))
-         (and (member issuer gnus-nocem-issuers) ; We like her...
-              (gnus-nocem-verify-issuer issuer) ; She is who she says she is..
-              (gnus-nocem-enter-article))))))) ; We gobble the message.
-  
+  ;; Get the article.
+  (let ((date (mail-header-date header))
+       (gnus-newsgroup-name group)
+       issuer b e type)
+    (when (or (not date)
+             (time-less-p
+              (time-since (date-to-time date))
+              (days-to-time gnus-nocem-expiry-wait)))
+      (gnus-request-article-this-buffer (mail-header-number header) group)
+      (goto-char (point-min))
+      (when (re-search-forward "-----BEGIN PGP MESSAGE-----" nil t)
+       (delete-region (point-min) (match-beginning 0)))
+      (when (re-search-forward "-----END PGP MESSAGE-----\n?" nil t)
+       (delete-region (match-end 0) (point-max)))
+      (goto-char (point-min))
+      ;; The article has to have proper NoCeM headers.
+      (when (and (setq b (search-forward "\n@@BEGIN NCM HEADERS\n" nil t))
+                (setq e (search-forward "\n@@BEGIN NCM BODY\n" nil t)))
+       ;; We get the name of the issuer.
+       (narrow-to-region b e)
+       (setq issuer (mail-fetch-field "issuer")
+             type (mail-fetch-field "issuer"))
+       (widen)
+       (if (not (gnus-nocem-message-wanted-p issuer type))
+           (message "invalid NoCeM issuer: %s" issuer)
+         (and (gnus-nocem-verify-issuer issuer) ; She is who she says she is.
+              (gnus-nocem-enter-article) ; We gobble the message.
+              (push (mail-header-message-id header) ; But don't come back for
+                    gnus-nocem-seen-message-ids))))))) ; second helpings.
+
+(defun gnus-nocem-message-wanted-p (issuer type)
+  (let ((issuers gnus-nocem-issuers)
+       wanted conditions condition)
+    (cond
+     ;; Do the quick check first.
+     ((member issuer issuers)
+      t)
+     ((setq conditions (cdr (assoc issuer issuers)))
+      ;; Check whether we want this type.
+      (while (setq condition (pop conditions))
+       (cond
+        ((stringp condition)
+         (setq wanted (string-match condition type)))
+        ((and (consp condition)
+              (eq (car condition) 'not)
+              (stringp (cadr condition)))
+         (setq wanted (not (string-match (cadr condition) type))))
+        (t
+         (error "Invalid NoCeM condition: %S" condition))))
+      wanted))))
+

 (defun gnus-nocem-verify-issuer (person)
   "Verify using PGP that the canceler is who she says she is."
 (defun gnus-nocem-verify-issuer (person)
   "Verify using PGP that the canceler is who she says she is."

-  t)
+  (if (fboundp gnus-nocem-verifyer)
+      (ignore-errors
+       (funcall gnus-nocem-verifyer))
+    ;; If we don't have Mailcrypt, then we use the message anyway.
+    t))

 
 (defun gnus-nocem-enter-article ()
   "Enter the current article into the NoCeM cache."
 
 (defun gnus-nocem-enter-article ()
   "Enter the current article into the NoCeM cache."

-  (widen)

   (goto-char (point-min))
   (let ((b (search-forward "\n@@BEGIN NCM BODY\n" nil t))
        (e (search-forward "\n@@END NCM BODY\n" nil t))
        (buf (current-buffer))
   (goto-char (point-min))
   (let ((b (search-forward "\n@@BEGIN NCM BODY\n" nil t))
        (e (search-forward "\n@@END NCM BODY\n" nil t))
        (buf (current-buffer))

-       ncm id)
+       ncm id group)

     (when (and b e)
       (narrow-to-region b (1+ (match-beginning 0)))
       (goto-char (point-min))
       (while (search-forward "\t" nil t)
     (when (and b e)
       (narrow-to-region b (1+ (match-beginning 0)))
       (goto-char (point-min))
       (while (search-forward "\t" nil t)

-       (when (boundp (let ((obarray gnus-newsrc-hashtb)) (read buf)))
-         (beginning-of-line)
-         (while (= (following-char) ?\t)
-           (forward-line -1))
-         (setq id (buffer-substring (point) (1- (search-forward "\t"))))
-         (push id ncm)
-         (gnus-sethash id t gnus-nocem-hashtb)
-         (forward-line 1)
-         (while (= (following-char) ?\t)
-           (forward-line 1))))
+       (cond
+        ((not (ignore-errors
+                (setq group (let ((obarray gnus-active-hashtb)) (read buf)))))
+         ;; An error.
+         )
+        ((not (symbolp group))
+         ;; Ignore invalid entries.
+         )
+        ((not (boundp group))
+         ;; Make sure all entries in the hashtb are bound.
+         (set group nil))
+        (t
+         (when (gnus-gethash (gnus-group-real-name (symbol-name group))
+                             gnus-nocem-real-group-hashtb)
+           ;; Valid group.
+           (beginning-of-line)
+           (while (eq (char-after) ?\t)
+             (forward-line -1))
+           (setq id (buffer-substring (point) (1- (search-forward "\t"))))
+           (unless (gnus-gethash id gnus-nocem-hashtb)
+             ;; only store if not already present
+             (gnus-sethash id t gnus-nocem-hashtb)
+             (push id ncm))
+           (forward-line 1)
+           (while (eq (char-after) ?\t)
+             (forward-line 1))))))

       (when ncm
        (setq gnus-nocem-touched-alist t)
        (push (cons (let ((time (current-time))) (setcdr (cdr time) nil) time)
       (when ncm
        (setq gnus-nocem-touched-alist t)
        (push (cons (let ((time (current-time))) (setcdr (cdr time) nil) time)

-                   ncm) 
-             gnus-nocem-alist)))))
+                   ncm)
+             gnus-nocem-alist))
+      t)))

 
 (defun gnus-nocem-load-cache ()
   "Load the NoCeM cache."
 
 (defun gnus-nocem-load-cache ()
   "Load the NoCeM cache."

+  (interactive)

   (unless gnus-nocem-alist
     ;; The buffer doesn't exist, so we create it and load the NoCeM
   (unless gnus-nocem-alist
     ;; The buffer doesn't exist, so we create it and load the NoCeM

-    ;; cache.  
+    ;; cache.

     (when (file-exists-p (gnus-nocem-cache-file))
       (load (gnus-nocem-cache-file) t t t)
       (gnus-nocem-alist-to-hashtb))))
     (when (file-exists-p (gnus-nocem-cache-file))
       (load (gnus-nocem-cache-file) t t t)
       (gnus-nocem-alist-to-hashtb))))

-      
+

 (defun gnus-nocem-save-cache ()
   "Save the NoCeM cache."
   (when (and gnus-nocem-alist
             gnus-nocem-touched-alist)
 (defun gnus-nocem-save-cache ()
   "Save the NoCeM cache."
   (when (and gnus-nocem-alist
             gnus-nocem-touched-alist)

-    (nnheader-temp-write (gnus-nocem-cache-file)
-      (insert (prin1-to-string
-              `(setq gnus-nocem-alist ',gnus-nocem-alist))))
+    (with-temp-file (gnus-nocem-cache-file)
+      (gnus-prin1 `(setq gnus-nocem-alist ',gnus-nocem-alist)))

     (setq gnus-nocem-touched-alist nil)))
 
 (defun gnus-nocem-save-active ()
   "Save the NoCeM active file."
     (setq gnus-nocem-touched-alist nil)))
 
 (defun gnus-nocem-save-active ()
   "Save the NoCeM active file."

-  (nnheader-temp-write (gnus-nocem-active-file)
-    (insert (prin1-to-string
-            `(setq gnus-nocem-active ',gnus-nocem-active)))))
+  (with-temp-file (gnus-nocem-active-file)
+    (gnus-prin1 `(setq gnus-nocem-active ',gnus-nocem-active))))

 
 (defun gnus-nocem-alist-to-hashtb ()
   "Create a hashtable from the Message-IDs we have."
   (let* ((alist gnus-nocem-alist)
         (pprev (cons nil alist))
         (prev pprev)
 
 (defun gnus-nocem-alist-to-hashtb ()
   "Create a hashtable from the Message-IDs we have."
   (let* ((alist gnus-nocem-alist)
         (pprev (cons nil alist))
         (prev pprev)

-        (expiry (nnmail-days-to-time gnus-nocem-expiry-wait))
+        (expiry (days-to-time gnus-nocem-expiry-wait))

         entry)
     (setq gnus-nocem-hashtb (gnus-make-hashtable (* (length alist) 51)))
     (while (setq entry (car alist))
         entry)
     (setq gnus-nocem-hashtb (gnus-make-hashtable (* (length alist) 51)))
     (while (setq entry (car alist))

-      (if (not (nnmail-time-less (nnmail-time-since (car entry)) expiry))
+      (if (not (time-less-p (time-since (car entry)) expiry))

          ;; This entry has expired, so we remove it.
          (setcdr prev (cdr alist))
        (setq prev alist)
          ;; This entry has expired, so we remove it.
          (setcdr prev (cdr alist))
        (setq prev alist)


@@ -215,16 +368,21 @@
          (setq entry (cdr entry))))
       (setq alist (cdr alist)))))
 
          (setq entry (cdr entry))))
       (setq alist (cdr alist)))))
 

+(gnus-add-shutdown 'gnus-nocem-close 'gnus)
+

 (defun gnus-nocem-close ()
   "Clear internal NoCeM variables."
   (setq gnus-nocem-alist nil
        gnus-nocem-hashtb nil
        gnus-nocem-active nil
 (defun gnus-nocem-close ()
   "Clear internal NoCeM variables."
   (setq gnus-nocem-alist nil
        gnus-nocem-hashtb nil
        gnus-nocem-active nil

-       gnus-nocem-touched-alist nil))
+       gnus-nocem-touched-alist nil
+       gnus-nocem-seen-message-ids nil
+       gnus-nocem-real-group-hashtb nil))

 
 (defun gnus-nocem-unwanted-article-p (id)
   "Say whether article ID in the current group is wanted."
 
 (defun gnus-nocem-unwanted-article-p (id)
   "Say whether article ID in the current group is wanted."

-  (gnus-gethash id gnus-nocem-hashtb))
+  (and gnus-nocem-hashtb
+       (gnus-gethash id gnus-nocem-hashtb)))

 
 (provide 'gnus-nocem)
 
 
 (provide 'gnus-nocem)