Import Gnus v5.10.5.
[elisp/gnus.git-] / lisp / spam.el
1 ;;; spam.el --- Identifying spam
2 ;; Copyright (C) 2002, 2003, 2004 Free Software Foundation, Inc.
3
4 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
5 ;; Keywords: network
6
7 ;; This file is part of GNU Emacs.
8
9 ;; GNU Emacs is free software; you can redistribute it and/or modify
10 ;; it under the terms of the GNU General Public License as published by
11 ;; the Free Software Foundation; either version 2, or (at your option)
12 ;; any later version.
13
14 ;; GNU Emacs is distributed in the hope that it will be useful,
15 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
17 ;; GNU General Public License for more details.
18
19 ;; You should have received a copy of the GNU General Public License
20 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
21 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
22 ;; Boston, MA 02111-1307, USA.
23
24 ;;; Commentary:
25
26 ;;; This module addresses a few aspects of spam control under Gnus.  Page
27 ;;; breaks are used for grouping declarations and documentation relating to
28 ;;; each particular aspect.
29
30 ;;; The integration with Gnus is not yet complete.  See various `FIXME'
31 ;;; comments, below, for supplementary explanations or discussions.
32
33 ;;; Several TODO items are marked as such
34
35 ;; TODO: spam scores, detection of spam in newsgroups, cross-server splitting,
36 ;; remote processing, training through files
37
38 ;;; Code:
39
40 (eval-when-compile (require 'cl))
41
42 (require 'gnus-sum)
43
44 (require 'gnus-uu)                      ; because of key prefix issues
45 ;;; for the definitions of group content classification and spam processors
46 (require 'gnus)
47 (require 'message)              ;for the message-fetch-field functions
48
49 ;; for nnimap-split-download-body-default
50 (eval-when-compile (require 'nnimap))
51
52 ;; autoload executable-find
53 (eval-and-compile
54   ;; executable-find is not autoloaded in Emacs 20
55   (autoload 'executable-find "executable"))
56
57 ;; autoload query-dig
58 (eval-and-compile
59   (autoload 'query-dig "dig"))
60
61 ;; autoload spam-report
62 (eval-and-compile
63   (autoload 'spam-report-gmane "spam-report"))
64
65 ;; autoload gnus-registry
66 (eval-and-compile
67   (autoload 'gnus-registry-group-count "gnus-registry")
68   (autoload 'gnus-registry-add-group "gnus-registry")
69   (autoload 'gnus-registry-store-extra-entry "gnus-registry")
70   (autoload 'gnus-registry-fetch-extra "gnus-registry"))
71
72 ;; autoload query-dns
73 (eval-and-compile
74   (autoload 'query-dns "dns"))
75
76 ;;; Main parameters.
77
78 (defgroup spam nil
79   "Spam configuration.")
80
81 (defcustom spam-directory "~/News/spam/"
82   "Directory for spam whitelists and blacklists."
83   :type 'directory
84   :group 'spam)
85
86 (defcustom spam-move-spam-nonspam-groups-only t
87   "Whether spam should be moved in non-spam groups only.
88 When t, only ham and unclassified groups will have their spam moved
89 to the spam-process-destination.  When nil, spam will also be moved from
90 spam groups."
91   :type 'boolean
92   :group 'spam)
93
94 (defcustom spam-process-ham-in-nonham-groups nil
95   "Whether ham should be processed in non-ham groups."
96   :type 'boolean
97   :group 'spam)
98
99 (defcustom spam-log-to-registry nil
100   "Whether spam/ham processing should be logged in the registry."
101   :type 'boolean
102   :group 'spam)
103
104 (defcustom spam-split-symbolic-return nil
105   "Whether `spam-split' should work with symbols or group names."
106   :type 'boolean
107   :group 'spam)
108
109 (defcustom spam-split-symbolic-return-positive nil
110   "Whether `spam-split' should ALWAYS work with symbols or group names.
111 Do not set this if you use `spam-split' in a fancy split
112   method."
113   :type 'boolean
114   :group 'spam)
115
116 (defcustom spam-process-ham-in-spam-groups nil
117   "Whether ham should be processed in spam groups."
118   :type 'boolean
119   :group 'spam)
120
121 (defcustom spam-mark-only-unseen-as-spam t
122   "Whether only unseen articles should be marked as spam in spam groups.
123 When nil, all unread articles in a spam group are marked as
124 spam.  Set this if you want to leave an article unread in a spam group
125 without losing it to the automatic spam-marking process."
126   :type 'boolean
127   :group 'spam)
128
129 (defcustom spam-mark-ham-unread-before-move-from-spam-group nil
130   "Whether ham should be marked unread before it's moved.
131 The article is moved out of a spam group according to ham-process-destination.
132 This variable is an official entry in the international Longest Variable Name
133 Competition."
134   :type 'boolean
135   :group 'spam)
136
137 (defcustom spam-disable-spam-split-during-ham-respool nil
138   "Whether `spam-split' should be ignored while resplitting ham in a process
139 destination.  This is useful to prevent ham from ending up in the same spam
140 group after the resplit.  Don't set this to t if you have spam-split as the
141 last rule in your split configuration."
142   :type 'boolean
143   :group 'spam)
144
145 (defcustom spam-autodetect-recheck-messages nil
146   "Should spam.el recheck all meessages when autodetecting?
147 Normally this is nil, so only unseen messages will be checked."
148   :type 'boolean
149   :group 'spam)
150
151 (defcustom spam-whitelist (expand-file-name "whitelist" spam-directory)
152   "The location of the whitelist.
153 The file format is one regular expression per line.
154 The regular expression is matched against the address."
155   :type 'file
156   :group 'spam)
157
158 (defcustom spam-blacklist (expand-file-name "blacklist" spam-directory)
159   "The location of the blacklist.
160 The file format is one regular expression per line.
161 The regular expression is matched against the address."
162   :type 'file
163   :group 'spam)
164
165 (defcustom spam-use-dig t
166   "Whether `query-dig' should be used instead of `query-dns'."
167   :type 'boolean
168   :group 'spam)
169
170 (defcustom spam-use-blacklist nil
171   "Whether the blacklist should be used by `spam-split'."
172   :type 'boolean
173   :group 'spam)
174
175 (defcustom spam-blacklist-ignored-regexes nil
176   "Regular expressions that the blacklist should ignore."
177   :type '(repeat (regexp :tag "Regular expression to ignore when blacklisting"))
178   :group 'spam)
179
180 (defcustom spam-use-whitelist nil
181   "Whether the whitelist should be used by `spam-split'."
182   :type 'boolean
183   :group 'spam)
184
185 (defcustom spam-use-whitelist-exclusive nil
186   "Whether whitelist-exclusive should be used by `spam-split'.
187 Exclusive whitelisting means that all messages from senders not in the whitelist
188 are considered spam."
189   :type 'boolean
190   :group 'spam)
191
192 (defcustom spam-use-blackholes nil
193   "Whether blackholes should be used by `spam-split'."
194   :type 'boolean
195   :group 'spam)
196
197 (defcustom spam-use-hashcash nil
198   "Whether hashcash payments should be detected by `spam-split'."
199   :type 'boolean
200   :group 'spam)
201
202 (defcustom spam-use-regex-headers nil
203   "Whether a header regular expression match should be used by `spam-split'.
204 Also see the variables `spam-regex-headers-spam' and `spam-regex-headers-ham'."
205   :type 'boolean
206   :group 'spam)
207
208 (defcustom spam-use-regex-body nil
209   "Whether a body regular expression match should be used by `spam-split'.
210 Also see the variables `spam-regex-body-spam' and `spam-regex-body-ham'."
211   :type 'boolean
212   :group 'spam)
213
214 (defcustom spam-use-bogofilter-headers nil
215   "Whether bogofilter headers should be used by `spam-split'.
216 Enable this if you pre-process messages with Bogofilter BEFORE Gnus sees them."
217   :type 'boolean
218   :group 'spam)
219
220 (defcustom spam-use-bogofilter nil
221   "Whether bogofilter should be invoked by `spam-split'.
222 Enable this if you want Gnus to invoke Bogofilter on new messages."
223   :type 'boolean
224   :group 'spam)
225
226 (defcustom spam-use-BBDB nil
227   "Whether BBDB should be used by `spam-split'."
228   :type 'boolean
229   :group 'spam)
230
231 (defcustom spam-use-BBDB-exclusive nil
232   "Whether BBDB-exclusive should be used by `spam-split'.
233 Exclusive BBDB means that all messages from senders not in the BBDB are
234 considered spam."
235   :type 'boolean
236   :group 'spam)
237
238 (defcustom spam-use-ifile nil
239   "Whether ifile should be used by `spam-split'."
240   :type 'boolean
241   :group 'spam)
242
243 (defcustom spam-use-stat nil
244   "Whether `spam-stat' should be used by `spam-split'."
245   :type 'boolean
246   :group 'spam)
247
248 (defcustom spam-use-spamoracle nil
249   "Whether spamoracle should be used by `spam-split'."
250   :type 'boolean
251   :group 'spam)
252
253 (defcustom spam-install-hooks (or
254                                spam-use-dig
255                                spam-use-blacklist
256                                spam-use-whitelist
257                                spam-use-whitelist-exclusive
258                                spam-use-blackholes
259                                spam-use-hashcash
260                                spam-use-regex-headers
261                                spam-use-regex-body
262                                spam-use-bogofilter-headers
263                                spam-use-bogofilter
264                                spam-use-BBDB
265                                spam-use-BBDB-exclusive
266                                spam-use-ifile
267                                spam-use-stat
268                                spam-use-spamoracle)
269   "Whether the spam hooks should be installed.
270 Default to t if one of the spam-use-* variables is set."
271   :group 'spam
272   :type 'boolean)
273
274 (defcustom spam-split-group "spam"
275   "Group name where incoming spam should be put by `spam-split'."
276   :type 'string
277   :group 'spam)
278
279 ;;; TODO: deprecate this variable, it's confusing since it's a list of strings,
280 ;;; not regular expressions
281 (defcustom spam-junk-mailgroups (cons
282                                  spam-split-group
283                                  '("mail.junk" "poste.pourriel"))
284   "Mailgroups with spam contents.
285 All unmarked article in such group receive the spam mark on group entry."
286   :type '(repeat (string :tag "Group"))
287   :group 'spam)
288
289 (defcustom spam-blackhole-servers '("bl.spamcop.net" "relays.ordb.org"
290                                     "dev.null.dk" "relays.visi.com")
291   "List of blackhole servers."
292   :type '(repeat (string :tag "Server"))
293   :group 'spam)
294
295 (defcustom spam-blackhole-good-server-regex nil
296   "String matching IP addresses that should not be checked in the blackholes."
297   :type '(radio (const nil)
298                 (regexp :format "%t: %v\n" :size 0))
299   :group 'spam)
300
301 (defcustom spam-face 'gnus-splash-face
302   "Face for spam-marked articles."
303   :type 'face
304   :group 'spam)
305
306 (defcustom spam-regex-headers-spam '("^X-Spam-Flag: YES")
307   "Regular expression for positive header spam matches."
308   :type '(repeat (regexp :tag "Regular expression to match spam header"))
309   :group 'spam)
310
311 (defcustom spam-regex-headers-ham '("^X-Spam-Flag: NO")
312   "Regular expression for positive header ham matches."
313   :type '(repeat (regexp :tag "Regular expression to match ham header"))
314   :group 'spam)
315
316 (defcustom spam-regex-body-spam '()
317   "Regular expression for positive body spam matches."
318   :type '(repeat (regexp :tag "Regular expression to match spam body"))
319   :group 'spam)
320
321 (defcustom spam-regex-body-ham '()
322   "Regular expression for positive body ham matches."
323   :type '(repeat (regexp :tag "Regular expression to match ham body"))
324   :group 'spam)
325
326 (defgroup spam-ifile nil
327   "Spam ifile configuration."
328   :group 'spam)
329
330 (defcustom spam-ifile-path (executable-find "ifile")
331   "File path of the ifile executable program."
332   :type '(choice (file :tag "Location of ifile")
333                  (const :tag "ifile is not installed"))
334   :group 'spam-ifile)
335
336 (defcustom spam-ifile-database-path nil
337   "File path of the ifile database."
338   :type '(choice (file :tag "Location of the ifile database")
339                  (const :tag "Use the default"))
340   :group 'spam-ifile)
341
342 (defcustom spam-ifile-spam-category "spam"
343   "Name of the spam ifile category."
344   :type 'string
345   :group 'spam-ifile)
346
347 (defcustom spam-ifile-ham-category nil
348   "Name of the ham ifile category.
349 If nil, the current group name will be used."
350   :type '(choice (string :tag "Use a fixed category")
351                  (const :tag "Use the current group name"))
352   :group 'spam-ifile)
353
354 (defcustom spam-ifile-all-categories nil
355   "Whether the ifile check will return all categories, or just spam.
356 Set this to t if you want to use the `spam-split' invocation of ifile as
357 your main source of newsgroup names."
358   :type 'boolean
359   :group 'spam-ifile)
360
361 (defgroup spam-bogofilter nil
362   "Spam bogofilter configuration."
363   :group 'spam)
364
365 (defcustom spam-bogofilter-path (executable-find "bogofilter")
366   "File path of the Bogofilter executable program."
367   :type '(choice (file :tag "Location of bogofilter")
368                  (const :tag "Bogofilter is not installed"))
369   :group 'spam-bogofilter)
370
371 (defcustom spam-bogofilter-header "X-Bogosity"
372   "The header that Bogofilter inserts in messages."
373   :type 'string
374   :group 'spam-bogofilter)
375
376 (defcustom spam-bogofilter-spam-switch "-s"
377   "The switch that Bogofilter uses to register spam messages."
378   :type 'string
379   :group 'spam-bogofilter)
380
381 (defcustom spam-bogofilter-ham-switch "-n"
382   "The switch that Bogofilter uses to register ham messages."
383   :type 'string
384   :group 'spam-bogofilter)
385
386 (defcustom spam-bogofilter-spam-strong-switch "-S"
387   "The switch that Bogofilter uses to unregister ham messages."
388   :type 'string
389   :group 'spam-bogofilter)
390
391 (defcustom spam-bogofilter-ham-strong-switch "-N"
392   "The switch that Bogofilter uses to unregister spam messages."
393   :type 'string
394   :group 'spam-bogofilter)
395
396 (defcustom spam-bogofilter-bogosity-positive-spam-header "^\\(Yes\\|Spam\\)"
397   "The regex on `spam-bogofilter-header' for positive spam identification."
398   :type 'regexp
399   :group 'spam-bogofilter)
400
401 (defcustom spam-bogofilter-database-directory nil
402   "Directory path of the Bogofilter databases."
403   :type '(choice (directory
404                   :tag "Location of the Bogofilter database directory")
405                  (const :tag "Use the default"))
406   :group 'spam-bogofilter)
407
408 (defgroup spam-spamoracle nil
409   "Spam spamoracle configuration."
410   :group 'spam)
411
412 (defcustom spam-spamoracle-database nil
413   "Location of spamoracle database file. When nil, use the default
414 spamoracle database."
415   :type '(choice (directory :tag "Location of spamoracle database file.")
416                  (const :tag "Use the default"))
417   :group 'spam-spamoracle)
418
419 (defcustom spam-spamoracle-binary (executable-find "spamoracle")
420   "Location of the spamoracle binary."
421   :type '(choice (directory :tag "Location of the spamoracle binary")
422                  (const :tag "Use the default"))
423   :group 'spam-spamoracle)
424
425 ;;; Key bindings for spam control.
426
427 (gnus-define-keys gnus-summary-mode-map
428   "St" spam-bogofilter-score
429   "Sx" gnus-summary-mark-as-spam
430   "Mst" spam-bogofilter-score
431   "Msx" gnus-summary-mark-as-spam
432   "\M-d" gnus-summary-mark-as-spam)
433
434 (defvar spam-old-ham-articles nil
435   "List of old ham articles, generated when a group is entered.")
436
437 (defvar spam-old-spam-articles nil
438   "List of old spam articles, generated when a group is entered.")
439
440 (defvar spam-split-disabled nil
441   "If non-nil, `spam-split' is disabled, and always returns nil.")
442
443 (defvar spam-split-last-successful-check nil
444   "`spam-split' will set this to nil or a spam-use-XYZ check if it
445   finds ham or spam.")
446
447 ;; convenience functions
448 (defun spam-xor (a b)
449   "Logical exclusive `or'."
450   (and (or a b) (not (and a b))))
451
452 (defun spam-group-ham-mark-p (group mark &optional spam)
453   (when (stringp group)
454     (let* ((marks (spam-group-ham-marks group spam))
455            (marks (if (symbolp mark)
456                       marks
457                     (mapcar 'symbol-value marks))))
458       (memq mark marks))))
459
460 (defun spam-group-spam-mark-p (group mark)
461   (spam-group-ham-mark-p group mark t))
462
463 (defun spam-group-ham-marks (group &optional spam)
464   (when (stringp group)
465     (let* ((marks (if spam
466                       (gnus-parameter-spam-marks group)
467                     (gnus-parameter-ham-marks group)))
468            (marks (car marks))
469            (marks (if (listp (car marks)) (car marks) marks)))
470       marks)))
471
472 (defun spam-group-spam-marks (group)
473   (spam-group-ham-marks group t))
474
475 (defun spam-group-spam-contents-p (group)
476   (if (stringp group)
477       (or (member group spam-junk-mailgroups)
478           (memq 'gnus-group-spam-classification-spam
479                 (gnus-parameter-spam-contents group)))
480     nil))
481
482 (defun spam-group-ham-contents-p (group)
483   (if (stringp group)
484       (memq 'gnus-group-spam-classification-ham
485             (gnus-parameter-spam-contents group))
486     nil))
487
488 (defvar spam-list-of-processors
489   '((gnus-group-spam-exit-processor-report-gmane spam spam-use-gmane)
490     (gnus-group-spam-exit-processor-bogofilter   spam spam-use-bogofilter)
491     (gnus-group-spam-exit-processor-blacklist    spam spam-use-blacklist)
492     (gnus-group-spam-exit-processor-ifile        spam spam-use-ifile)
493     (gnus-group-spam-exit-processor-stat         spam spam-use-stat)
494     (gnus-group-spam-exit-processor-spamoracle   spam spam-use-spamoracle)
495     (gnus-group-ham-exit-processor-ifile         ham spam-use-ifile)
496     (gnus-group-ham-exit-processor-bogofilter    ham spam-use-bogofilter)
497     (gnus-group-ham-exit-processor-stat          ham spam-use-stat)
498     (gnus-group-ham-exit-processor-whitelist     ham spam-use-whitelist)
499     (gnus-group-ham-exit-processor-BBDB          ham spam-use-BBDB)
500     (gnus-group-ham-exit-processor-copy          ham spam-use-ham-copy)
501     (gnus-group-ham-exit-processor-spamoracle    ham spam-use-spamoracle))
502   "The spam-list-of-processors list contains pairs associating a
503 ham/spam exit processor variable with a classification and a
504 spam-use-* variable.")
505
506 (defun spam-group-processor-p (group processor)
507   (if (and (stringp group)
508            (symbolp processor))
509       (or (member processor (nth 0 (gnus-parameter-spam-process group)))
510           (spam-group-processor-multiple-p
511            group
512            (cdr-safe (assoc processor spam-list-of-processors))))
513     nil))
514
515 (defun spam-group-processor-multiple-p (group processor-info)
516   (let* ((classification (nth 0 processor-info))
517          (check (nth 1 processor-info))
518          (parameters (nth 0 (gnus-parameter-spam-process group)))
519          found)
520     (dolist (parameter parameters)
521       (when (and (null found)
522                  (listp parameter)
523                  (eq classification (nth 0 parameter))
524                  (eq check (nth 1 parameter)))
525         (setq found t)))
526     found))
527
528 (defun spam-group-spam-processor-report-gmane-p (group)
529   (spam-group-processor-p group 'gnus-group-spam-exit-processor-report-gmane))
530
531 (defun spam-group-spam-processor-bogofilter-p (group)
532   (spam-group-processor-p group 'gnus-group-spam-exit-processor-bogofilter))
533
534 (defun spam-group-spam-processor-blacklist-p (group)
535   (spam-group-processor-p group 'gnus-group-spam-exit-processor-blacklist))
536
537 (defun spam-group-spam-processor-ifile-p (group)
538   (spam-group-processor-p group 'gnus-group-spam-exit-processor-ifile))
539
540 (defun spam-group-ham-processor-ifile-p (group)
541   (spam-group-processor-p group 'gnus-group-ham-exit-processor-ifile))
542
543 (defun spam-group-spam-processor-spamoracle-p (group)
544   (spam-group-processor-p group 'gnus-group-spam-exit-processor-spamoracle))
545
546 (defun spam-group-ham-processor-bogofilter-p (group)
547   (spam-group-processor-p group 'gnus-group-ham-exit-processor-bogofilter))
548
549 (defun spam-group-spam-processor-stat-p (group)
550   (spam-group-processor-p group 'gnus-group-spam-exit-processor-stat))
551
552 (defun spam-group-ham-processor-stat-p (group)
553   (spam-group-processor-p group 'gnus-group-ham-exit-processor-stat))
554
555 (defun spam-group-ham-processor-whitelist-p (group)
556   (spam-group-processor-p group 'gnus-group-ham-exit-processor-whitelist))
557
558 (defun spam-group-ham-processor-BBDB-p (group)
559   (spam-group-processor-p group 'gnus-group-ham-exit-processor-BBDB))
560
561 (defun spam-group-ham-processor-copy-p (group)
562   (spam-group-processor-p group 'gnus-group-ham-exit-processor-copy))
563
564 (defun spam-group-ham-processor-spamoracle-p (group)
565   (spam-group-processor-p group 'gnus-group-ham-exit-processor-spamoracle))
566
567 ;;; Summary entry and exit processing.
568
569 (defun spam-summary-prepare ()
570   (setq spam-old-ham-articles
571         (spam-list-articles gnus-newsgroup-articles 'ham))
572   (setq spam-old-spam-articles
573         (spam-list-articles gnus-newsgroup-articles 'spam))
574   (spam-mark-junk-as-spam-routine))
575
576 ;; The spam processors are invoked for any group, spam or ham or neither
577 (defun spam-summary-prepare-exit ()
578   (unless gnus-group-is-exiting-without-update-p
579     (gnus-message 6 "Exiting summary buffer and applying spam rules")
580
581     ;; first of all, unregister any articles that are no longer ham or spam
582     ;; we have to iterate over the processors, or else we'll be too slow
583     (dolist (classification '(spam ham))
584       (let* ((old-articles (if (eq classification 'spam)
585                                spam-old-spam-articles
586                              spam-old-ham-articles))
587              (new-articles (spam-list-articles
588                             gnus-newsgroup-articles
589                             classification))
590              (changed-articles (gnus-set-difference old-articles new-articles)))
591         ;; now that we have the changed articles, we go through the processors
592         (dolist (processor-param spam-list-of-processors)
593           (let ((processor (nth 0 processor-param))
594                 (processor-classification (nth 1 processor-param))
595                 (check (nth 2 processor-param))
596                 unregister-list)
597             (dolist (article changed-articles)
598               (let ((id (spam-fetch-field-message-id-fast article)))
599                 (when (spam-log-unregistration-needed-p
600                        id 'process classification check)
601                   (push article unregister-list))))
602             ;; call spam-register-routine with specific articles to unregister,
603             ;; when there are articles to unregister and the check is enabled
604             (when (and unregister-list (symbol-value check))
605               (spam-register-routine classification check t unregister-list))))))
606
607     ;; find all the spam processors applicable to this group
608     (dolist (processor-param spam-list-of-processors)
609       (let ((processor (nth 0 processor-param))
610             (classification (nth 1 processor-param))
611             (check (nth 2 processor-param)))
612         (when (and (eq 'spam classification)
613                    (spam-group-processor-p gnus-newsgroup-name processor))
614           (spam-register-routine classification check))))
615
616     (if spam-move-spam-nonspam-groups-only
617         (when (not (spam-group-spam-contents-p gnus-newsgroup-name))
618           (spam-mark-spam-as-expired-and-move-routine
619            (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
620       (gnus-message 5 "Marking spam as expired and moving it to %s"
621                     gnus-newsgroup-name)
622       (spam-mark-spam-as-expired-and-move-routine
623        (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
624
625     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
626     ;; expire spam, in case the above did not expire them
627     (gnus-message 5 "Marking spam as expired without moving it")
628     (spam-mark-spam-as-expired-and-move-routine nil)
629
630     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
631               (and (spam-group-spam-contents-p gnus-newsgroup-name)
632                    spam-process-ham-in-spam-groups)
633               spam-process-ham-in-nonham-groups)
634       ;; find all the ham processors applicable to this group
635       (dolist (processor-param spam-list-of-processors)
636         (let ((processor (nth 0 processor-param))
637               (classification (nth 1 processor-param))
638               (check (nth 2 processor-param)))
639           (when (and (eq 'ham classification)
640                      (spam-group-processor-p gnus-newsgroup-name processor))
641             (spam-register-routine classification check)))))
642
643     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
644       (gnus-message 5 "Copying ham")
645       (spam-ham-copy-routine
646        (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
647
648     ;; now move all ham articles out of spam groups
649     (when (spam-group-spam-contents-p gnus-newsgroup-name)
650       (gnus-message 5 "Moving ham messages from spam group")
651       (spam-ham-move-routine
652        (gnus-parameter-ham-process-destination gnus-newsgroup-name))))
653
654   (setq spam-old-ham-articles nil)
655   (setq spam-old-spam-articles nil))
656
657 (defun spam-mark-junk-as-spam-routine ()
658   ;; check the global list of group names spam-junk-mailgroups and the
659   ;; group parameters
660   (when (spam-group-spam-contents-p gnus-newsgroup-name)
661     (gnus-message 5 "Marking %s articles as spam"
662                   (if spam-mark-only-unseen-as-spam
663                       "unseen"
664                     "unread"))
665     (let ((articles (if spam-mark-only-unseen-as-spam
666                         gnus-newsgroup-unseen
667                       gnus-newsgroup-unreads)))
668       (dolist (article articles)
669         (gnus-summary-mark-article article gnus-spam-mark)))))
670
671 (defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
672   (if (and (car-safe groups) (listp (car-safe groups)))
673       (apply 'spam-mark-spam-as-expired-and-move-routine (car groups))
674     (gnus-summary-kill-process-mark)
675     (let ((articles gnus-newsgroup-articles)
676           (backend-supports-deletions
677            (gnus-check-backend-function
678             'request-move-article gnus-newsgroup-name))
679           article tomove deletep)
680       (dolist (article articles)
681         (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
682           (gnus-summary-mark-article article gnus-expirable-mark)
683           (push article tomove)))
684
685       ;; now do the actual copies
686       (dolist (group groups)
687         (when (and tomove
688                    (stringp group))
689           (dolist (article tomove)
690             (gnus-summary-set-process-mark article))
691           (when tomove
692             (if (or (not backend-supports-deletions)
693                     (> (length groups) 1))
694                 (progn
695                   (gnus-summary-copy-article nil group)
696                   (setq deletep t))
697               (gnus-summary-move-article nil group)))))
698
699       ;; now delete the articles, if there was a copy done, and the
700       ;; backend allows it
701       (when (and deletep backend-supports-deletions)
702         (dolist (article tomove)
703           (gnus-summary-set-process-mark article))
704         (when tomove
705           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
706             (gnus-summary-delete-article nil))))
707
708       (gnus-summary-yank-process-mark))))
709
710 (defun spam-ham-copy-or-move-routine (copy groups)
711   (gnus-summary-kill-process-mark)
712   (let ((todo (spam-list-articles gnus-newsgroup-articles 'ham))
713         (backend-supports-deletions
714          (gnus-check-backend-function
715           'request-move-article gnus-newsgroup-name))
716         (respool-method (gnus-find-method-for-group gnus-newsgroup-name))
717         article mark todo deletep respool)
718
719     (when (member 'respool groups)
720       (setq respool t)                  ; boolean for later
721       (setq groups '("fake"))) ; when respooling, groups are dynamic so fake it
722
723     ;; now do the actual move
724     (dolist (group groups)
725       (when (and todo (stringp group))
726         (dolist (article todo)
727           (when spam-mark-ham-unread-before-move-from-spam-group
728             (gnus-summary-mark-article article gnus-unread-mark))
729           (gnus-summary-set-process-mark article))
730
731         (if respool                        ; respooling is with a "fake" group
732             (let ((spam-split-disabled
733                    (or spam-split-disabled
734                        spam-disable-spam-split-during-ham-respool)))
735               (gnus-summary-respool-article nil respool-method))
736           (if (or (not backend-supports-deletions) ; else, we are not respooling
737                   (> (length groups) 1))
738               (progn                ; if copying, copy and set deletep
739                 (gnus-summary-copy-article nil group)
740                 (setq deletep t))
741             (gnus-summary-move-article nil group))))) ; else move articles
742
743     ;; now delete the articles, unless a) copy is t, and there was a copy done
744     ;;                                 b) a move was done to a single group
745     ;;                                 c) backend-supports-deletions is nil
746     (unless copy
747       (when (and deletep backend-supports-deletions)
748         (dolist (article todo)
749           (gnus-summary-set-process-mark article))
750         (when todo
751           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
752             (gnus-summary-delete-article nil))))))
753
754   (gnus-summary-yank-process-mark))
755
756 (defun spam-ham-copy-routine (&rest groups)
757   (if (and (car-safe groups) (listp (car-safe groups)))
758       (apply 'spam-ham-copy-routine (car groups))
759     (spam-ham-copy-or-move-routine t groups)))
760
761 (defun spam-ham-move-routine (&rest groups)
762   (if (and (car-safe groups) (listp (car-safe groups)))
763       (apply 'spam-ham-move-routine (car groups))
764     (spam-ham-copy-or-move-routine nil groups)))
765
766 (eval-and-compile
767   (defalias 'spam-point-at-eol (if (fboundp 'point-at-eol)
768                                    'point-at-eol
769                                  'line-end-position)))
770
771 (defun spam-get-article-as-string (article)
772   (let ((article-buffer (spam-get-article-as-buffer article))
773         article-string)
774     (when article-buffer
775       (save-window-excursion
776         (set-buffer article-buffer)
777         (setq article-string (buffer-string))))
778     article-string))
779
780 (defun spam-get-article-as-buffer (article)
781   (let ((article-buffer))
782     (when (numberp article)
783       (save-window-excursion
784         (gnus-summary-goto-subject article)
785         (gnus-summary-show-article t)
786         (setq article-buffer (get-buffer gnus-article-buffer))))
787     article-buffer))
788
789 ;; disabled for now
790 ;; (defun spam-get-article-as-filename (article)
791 ;;   (let ((article-filename))
792 ;;     (when (numberp article)
793 ;;       (nnml-possibly-change-directory
794 ;;        (gnus-group-real-name gnus-newsgroup-name))
795 ;;       (setq article-filename (expand-file-name
796 ;;                              (int-to-string article) nnml-current-directory)))
797 ;;     (if (file-exists-p article-filename)
798 ;;      article-filename
799 ;;       nil)))
800
801 (defun spam-fetch-field-from-fast (article)
802   "Fetch the `from' field quickly, using the internal gnus-data-list function"
803   (if (and (numberp article)
804            (assoc article (gnus-data-list nil)))
805       (mail-header-from
806        (gnus-data-header (assoc article (gnus-data-list nil))))
807     nil))
808
809 (defun spam-fetch-field-subject-fast (article)
810   "Fetch the `subject' field quickly, using the internal
811   gnus-data-list function"
812   (if (and (numberp article)
813            (assoc article (gnus-data-list nil)))
814       (mail-header-subject
815        (gnus-data-header (assoc article (gnus-data-list nil))))
816     nil))
817
818 (defun spam-fetch-field-message-id-fast (article)
819   "Fetch the `Message-ID' field quickly, using the internal
820   gnus-data-list function"
821   (if (and (numberp article)
822            (assoc article (gnus-data-list nil)))
823       (mail-header-message-id
824        (gnus-data-header (assoc article (gnus-data-list nil))))
825     nil))
826
827 \f
828 ;;;; Spam determination.
829
830 (defvar spam-list-of-checks
831   '((spam-use-blacklist          . spam-check-blacklist)
832     (spam-use-regex-headers      . spam-check-regex-headers)
833     (spam-use-regex-body         . spam-check-regex-body)
834     (spam-use-whitelist          . spam-check-whitelist)
835     (spam-use-BBDB               . spam-check-BBDB)
836     (spam-use-ifile              . spam-check-ifile)
837     (spam-use-spamoracle         . spam-check-spamoracle)
838     (spam-use-stat               . spam-check-stat)
839     (spam-use-blackholes         . spam-check-blackholes)
840     (spam-use-hashcash           . spam-check-hashcash)
841     (spam-use-bogofilter-headers . spam-check-bogofilter-headers)
842     (spam-use-bogofilter         . spam-check-bogofilter))
843   "The spam-list-of-checks list contains pairs associating a
844 parameter variable with a spam checking function.  If the
845 parameter variable is true, then the checking function is called,
846 and its value decides what happens.  Each individual check may
847 return nil, t, or a mailgroup name.  The value nil means that the
848 check does not yield a decision, and so, that further checks are
849 needed.  The value t means that the message is definitely not
850 spam, and that further spam checks should be inhibited.
851 Otherwise, a mailgroup name or the symbol 'spam (depending on
852 spam-split-symbolic-return) is returned where the mail should go,
853 and further checks are also inhibited.  The usual mailgroup name
854 is the value of `spam-split-group', meaning that the message is
855 definitely a spam.")
856
857 (defvar spam-list-of-statistical-checks
858   '(spam-use-ifile
859     spam-use-regex-body
860     spam-use-stat
861     spam-use-bogofilter
862     spam-use-spamoracle)
863   "The spam-list-of-statistical-checks list contains all the mail
864 splitters that need to have the full message body available.")
865
866 ;;;TODO: modify to invoke self with each check if invoked without specifics
867 (defun spam-split (&rest specific-checks)
868   "Split this message into the `spam' group if it is spam.
869 This function can be used as an entry in the variable `nnmail-split-fancy',
870 for example like this: (: spam-split).  It can take checks as
871 parameters.  A string as a parameter will set the
872 spam-split-group to that string.
873
874 See the Info node `(gnus)Fancy Mail Splitting' for more details."
875   (interactive)
876   (setq spam-split-last-successful-check nil)
877   (unless spam-split-disabled
878     (let ((spam-split-group-choice spam-split-group))
879       (dolist (check specific-checks)
880         (when (stringp check)
881           (setq spam-split-group-choice check)
882           (setq specific-checks (delq check specific-checks))))
883
884       (let ((spam-split-group spam-split-group-choice))
885         (save-excursion
886           (save-restriction
887             (dolist (check spam-list-of-statistical-checks)
888               (when (and (symbolp check) (symbol-value check))
889                 (widen)
890                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
891                               (symbol-name check))
892                 (return)))
893             ;;   (progn (widen) (debug (buffer-string)))
894             (let ((list-of-checks spam-list-of-checks)
895                   decision)
896               (while (and list-of-checks (not decision))
897                 (let ((pair (pop list-of-checks)))
898                   (when (and (symbol-value (car pair))
899                              (or (null specific-checks)
900                                  (memq (car pair) specific-checks)))
901                     (gnus-message 5 "spam-split: calling the %s function"
902                                   (symbol-name (cdr pair)))
903                     (setq decision (funcall (cdr pair)))
904                     ;; if we got a decision at all, save the current check
905                     (when decision
906                       (setq spam-split-last-successful-check (car pair)))
907
908                     (when (eq decision 'spam)
909                       (if spam-split-symbolic-return
910                           (setq decision spam-split-group)
911                         (gnus-error
912                          5
913                          (format "spam-split got %s but %s is nil"
914                                  (symbol-name decision)
915                                  (symbol-name spam-split-symbolic-return))))))))
916               (if (eq decision t)
917                   (if spam-split-symbolic-return-positive 'ham nil)
918                 decision))))))))
919
920 (defun spam-find-spam ()
921   "This function will detect spam in the current newsgroup using spam-split."
922   (interactive)
923
924   (let* ((group gnus-newsgroup-name)
925          (autodetect (gnus-parameter-spam-autodetect group))
926          (methods (gnus-parameter-spam-autodetect-methods group))
927          (first-method (nth 0 methods)))
928   (when (and autodetect
929              (not (equal first-method 'none)))
930     (mapcar
931      (lambda (article)
932        (let ((id (spam-fetch-field-message-id-fast article))
933              (subject (spam-fetch-field-subject-fast article))
934              (sender (spam-fetch-field-from-fast article)))
935          (unless (and spam-log-to-registry
936                       (spam-log-registered-p id 'incoming))
937            (let* ((spam-split-symbolic-return t)
938                   (spam-split-symbolic-return-positive t)
939                   (split-return
940                    (with-temp-buffer
941                      (gnus-request-article-this-buffer
942                       article
943                       group)
944                      (if (or (null first-method)
945                              (equal first-method 'default))
946                          (spam-split)
947                        (apply 'spam-split methods)))))
948              (if (equal split-return 'spam)
949                  (gnus-summary-mark-article article gnus-spam-mark))
950
951              (when (and split-return spam-log-to-registry)
952                (when (zerop (gnus-registry-group-count id))
953                  (gnus-registry-add-group
954                   id group subject sender))
955
956                (spam-log-processing-to-registry
957                 id
958                 'incoming
959                 split-return
960                 spam-split-last-successful-check
961                 group))))))
962      (if spam-autodetect-recheck-messages
963          gnus-newsgroup-articles
964        gnus-newsgroup-unseen)))))
965
966 (defvar spam-registration-functions
967   ;; first the ham register, second the spam register function
968   ;; third the ham unregister, fourth the spam unregister function
969   '((spam-use-blacklist  nil
970                          spam-blacklist-register-routine
971                          nil
972                          spam-blacklist-unregister-routine)
973     (spam-use-whitelist  spam-whitelist-register-routine
974                          nil
975                          spam-whitelist-unregister-routine
976                          nil)
977     (spam-use-BBDB       spam-BBDB-register-routine
978                          nil
979                          spam-BBDB-unregister-routine
980                          nil)
981     (spam-use-ifile      spam-ifile-register-ham-routine
982                          spam-ifile-register-spam-routine
983                          spam-ifile-unregister-ham-routine
984                          spam-ifile-unregister-spam-routine)
985     (spam-use-spamoracle spam-spamoracle-learn-ham
986                          spam-spamoracle-learn-spam
987                          spam-spamoracle-unlearn-ham
988                          spam-spamoracle-unlearn-spam)
989     (spam-use-stat       spam-stat-register-ham-routine
990                          spam-stat-register-spam-routine
991                          spam-stat-unregister-ham-routine
992                          spam-stat-unregister-spam-routine)
993     ;; note that spam-use-gmane is not a legitimate check
994     (spam-use-gmane      nil
995                          spam-report-gmane-register-routine
996                          ;; does Gmane support unregistration?
997                          nil
998                          nil)
999     (spam-use-bogofilter spam-bogofilter-register-ham-routine
1000                          spam-bogofilter-register-spam-routine
1001                          spam-bogofilter-unregister-ham-routine
1002                          spam-bogofilter-unregister-spam-routine))
1003   "The spam-registration-functions list contains pairs
1004 associating a parameter variable with the ham and spam
1005 registration functions, and the ham and spam unregistration
1006 functions")
1007
1008 (defun spam-classification-valid-p (classification)
1009   (or  (eq classification 'spam)
1010        (eq classification 'ham)))
1011
1012 (defun spam-process-type-valid-p (process-type)
1013   (or  (eq process-type 'incoming)
1014        (eq process-type 'process)))
1015
1016 (defun spam-registration-check-valid-p (check)
1017   (assoc check spam-registration-functions))
1018
1019 (defun spam-unregistration-check-valid-p (check)
1020   (assoc check spam-registration-functions))
1021
1022 (defun spam-registration-function (classification check)
1023   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1024     (if (eq classification 'spam)
1025         (nth 1 flist)
1026       (nth 0 flist))))
1027
1028 (defun spam-unregistration-function (classification check)
1029   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1030     (if (eq classification 'spam)
1031         (nth 3 flist)
1032       (nth 2 flist))))
1033
1034 (defun spam-list-articles (articles classification)
1035   (let ((mark-check (if (eq classification 'spam)
1036                         'spam-group-spam-mark-p
1037                       'spam-group-ham-mark-p))
1038         list mark-cache-yes mark-cache-no)
1039     (dolist (article articles)
1040       (let ((mark (gnus-summary-article-mark article)))
1041         (unless (memq mark mark-cache-no)
1042           (if (memq mark mark-cache-yes)
1043               (push article list)
1044             ;; else, we have to actually check the mark
1045             (if (funcall mark-check
1046                          gnus-newsgroup-name
1047                          mark)
1048                 (progn
1049                   (push article list)
1050                   (push mark mark-cache-yes))
1051               (push mark mark-cache-no))))))
1052     list))
1053
1054 (defun spam-register-routine (classification
1055                               check
1056                               &optional unregister
1057                               specific-articles)
1058   (when (and (spam-classification-valid-p classification)
1059              (spam-registration-check-valid-p check))
1060     (let* ((register-function
1061             (spam-registration-function classification check))
1062            (unregister-function
1063             (spam-unregistration-function classification check))
1064            (run-function (if unregister
1065                              unregister-function
1066                            register-function))
1067            (log-function (if unregister
1068                              'spam-log-undo-registration
1069                            'spam-log-processing-to-registry))
1070            article articles)
1071
1072       (when run-function
1073         ;; make list of articles, using specific-articles if given
1074         (setq articles (or specific-articles
1075                            (spam-list-articles
1076                             gnus-newsgroup-articles
1077                             classification)))
1078         ;; process them
1079         (gnus-message 5 "%s %d %s articles with classification %s, check %s"
1080                       (if unregister "Unregistering" "Registering")
1081                       (length articles)
1082                       (if specific-articles "specific" "")
1083                       (symbol-name classification)
1084                       (symbol-name check))
1085         (funcall run-function articles)
1086         ;; now log all the registrations (or undo them, depending on unregister)
1087         (dolist (article articles)
1088           (funcall log-function
1089                    (spam-fetch-field-message-id-fast article)
1090                    'process
1091                    classification
1092                    check
1093                    gnus-newsgroup-name))))))
1094
1095 ;;; log a ham- or spam-processor invocation to the registry
1096 (defun spam-log-processing-to-registry (id type classification check group)
1097   (when spam-log-to-registry
1098     (if (and (stringp id)
1099              (stringp group)
1100              (spam-process-type-valid-p type)
1101              (spam-classification-valid-p classification)
1102              (spam-registration-check-valid-p check))
1103         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1104               (cell (list classification check group)))
1105           (push cell cell-list)
1106           (gnus-registry-store-extra-entry
1107            id
1108            type
1109            cell-list))
1110
1111       (gnus-message 5 (format "%s called with bad ID, type, classification, check, or group"
1112                               "spam-log-processing-to-registry")))))
1113
1114 ;;; check if a ham- or spam-processor registration has been done
1115 (defun spam-log-registered-p (id type)
1116   (when spam-log-to-registry
1117     (if (and (stringp id)
1118              (spam-process-type-valid-p type))
1119         (cdr-safe (gnus-registry-fetch-extra id type))
1120       (progn
1121         (gnus-message 5 (format "%s called with bad ID, type, classification, or check"
1122                                 "spam-log-registered-p"))
1123         nil))))
1124
1125 ;;; check if a ham- or spam-processor registration needs to be undone
1126 (defun spam-log-unregistration-needed-p (id type classification check)
1127   (when spam-log-to-registry
1128     (if (and (stringp id)
1129              (spam-process-type-valid-p type)
1130              (spam-classification-valid-p classification)
1131              (spam-registration-check-valid-p check))
1132         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1133               found)
1134           (dolist (cell cell-list)
1135             (unless found
1136               (when (and (eq classification (nth 0 cell))
1137                          (eq check (nth 1 cell)))
1138                 (setq found t))))
1139           found)
1140       (progn
1141         (gnus-message 5 (format "%s called with bad ID, type, classification, or check"
1142                                 "spam-log-unregistration-needed-p"))
1143         nil))))
1144
1145
1146 ;;; undo a ham- or spam-processor registration (the group is not used)
1147 (defun spam-log-undo-registration (id type classification check &optional group)
1148   (when (and spam-log-to-registry
1149              (spam-log-unregistration-needed-p id type classification check))
1150     (if (and (stringp id)
1151              (spam-process-type-valid-p type)
1152              (spam-classification-valid-p classification)
1153              (spam-registration-check-valid-p check))
1154         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1155               new-cell-list found)
1156           (dolist (cell cell-list)
1157             (unless (and (eq classification (nth 0 cell))
1158                          (eq check (nth 1 cell)))
1159               (push cell new-cell-list)))
1160           (gnus-registry-store-extra-entry
1161            id
1162            type
1163            new-cell-list))
1164       (progn
1165         (gnus-message 5 (format "%s called with bad ID, type, check, or group"
1166                                 "spam-log-undo-registration"))
1167         nil))))
1168
1169 ;;; set up IMAP widening if it's necessary
1170 (defun spam-setup-widening ()
1171   (dolist (check spam-list-of-statistical-checks)
1172     (when (symbol-value check)
1173       (setq nnimap-split-download-body-default t))))
1174
1175 \f
1176 ;;;; Regex body
1177
1178 (defun spam-check-regex-body ()
1179   (let ((spam-regex-headers-ham spam-regex-body-ham)
1180         (spam-regex-headers-spam spam-regex-body-spam))
1181     (spam-check-regex-headers t)))
1182
1183 \f
1184 ;;;; Regex headers
1185
1186 (defun spam-check-regex-headers (&optional body)
1187   (let ((type (if body "body" "header"))
1188         (spam-split-group (if spam-split-symbolic-return
1189                               'spam
1190                             spam-split-group))
1191         ret found)
1192     (dolist (h-regex spam-regex-headers-ham)
1193       (unless found
1194         (goto-char (point-min))
1195         (when (re-search-forward h-regex nil t)
1196           (message "Ham regex %s search positive." type)
1197           (setq found t))))
1198     (dolist (s-regex spam-regex-headers-spam)
1199       (unless found
1200         (goto-char (point-min))
1201         (when (re-search-forward s-regex nil t)
1202           (message "Spam regex %s search positive." type)
1203           (setq found t)
1204           (setq ret spam-split-group))))
1205     ret))
1206
1207 \f
1208 ;;;; Blackholes.
1209
1210 (defun spam-reverse-ip-string (ip)
1211   (when (stringp ip)
1212     (mapconcat 'identity
1213                (nreverse (split-string ip "\\."))
1214                ".")))
1215
1216 (defun spam-check-blackholes ()
1217   "Check the Received headers for blackholed relays."
1218   (let ((headers (nnmail-fetch-field "received"))
1219         (spam-split-group (if spam-split-symbolic-return
1220                               'spam
1221                             spam-split-group))
1222         ips matches)
1223     (when headers
1224       (with-temp-buffer
1225         (insert headers)
1226         (goto-char (point-min))
1227         (gnus-message 5 "Checking headers for relay addresses")
1228         (while (re-search-forward
1229                 "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
1230           (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
1231           (push (spam-reverse-ip-string (match-string 1))
1232                 ips)))
1233       (dolist (server spam-blackhole-servers)
1234         (dolist (ip ips)
1235           (unless (and spam-blackhole-good-server-regex
1236                        ;; match the good-server-regex against the reversed (again) IP string
1237                        (string-match
1238                         spam-blackhole-good-server-regex
1239                         (spam-reverse-ip-string ip)))
1240             (unless matches
1241               (let ((query-string (concat ip "." server)))
1242                 (if spam-use-dig
1243                     (let ((query-result (query-dig query-string)))
1244                       (when query-result
1245                         (gnus-message 5 "(DIG): positive blackhole check '%s'"
1246                                       query-result)
1247                         (push (list ip server query-result)
1248                               matches)))
1249                   ;; else, if not using dig.el
1250                   (when (query-dns query-string)
1251                     (gnus-message 5 "positive blackhole check")
1252                     (push (list ip server (query-dns query-string 'TXT))
1253                           matches)))))))))
1254     (when matches
1255       spam-split-group)))
1256 \f
1257 ;;;; Hashcash.
1258
1259 (condition-case nil
1260     (progn
1261       (require 'hashcash)
1262
1263       (defun spam-check-hashcash ()
1264         "Check the headers for hashcash payments."
1265         (mail-check-payment)))   ;mail-check-payment returns a boolean
1266
1267   (file-error (progn
1268                 (defalias 'mail-check-payment 'ignore)
1269                 (defalias 'spam-check-hashcash 'ignore))))
1270 \f
1271 ;;;; BBDB
1272
1273 ;;; original idea for spam-check-BBDB from Alexander Kotelnikov
1274 ;;; <sacha@giotto.sj.ru>
1275
1276 ;; all this is done inside a condition-case to trap errors
1277
1278 (condition-case nil
1279     (progn
1280       (require 'bbdb)
1281       (require 'bbdb-com)
1282
1283       (defun spam-enter-ham-BBDB (addresses &optional remove)
1284         "Enter an address into the BBDB; implies ham (non-spam) sender"
1285         (dolist (from addresses)
1286           (when (stringp from)
1287             (let* ((parsed-address (gnus-extract-address-components from))
1288                    (name (or (nth 0 parsed-address) "Ham Sender"))
1289                    (remove-function (if remove
1290                                         'bbdb-delete-record-internal
1291                                       'ignore))
1292                    (net-address (nth 1 parsed-address))
1293                    (record (and net-address
1294                                 (bbdb-search-simple nil net-address))))
1295               (when net-address
1296                 (gnus-message 5 "%s address %s %s BBDB"
1297                               (if remove "Deleting" "Adding")
1298                               from
1299                               (if remove "from" "to"))
1300                 (if record
1301                     (funcall remove-function record)
1302                   (bbdb-create-internal name nil net-address nil nil
1303                                         "ham sender added by spam.el")))))))
1304
1305       (defun spam-BBDB-register-routine (articles &optional unregister)
1306         (let (addresses)
1307           (dolist (article articles)
1308             (when (stringp (spam-fetch-field-from-fast article))
1309               (push (spam-fetch-field-from-fast article) addresses)))
1310           ;; now do the register/unregister action
1311           (spam-enter-ham-BBDB addresses unregister)))
1312
1313       (defun spam-BBDB-unregister-routine (articles)
1314         (spam-BBDB-register-routine articles t))
1315
1316       (defun spam-check-BBDB ()
1317         "Mail from people in the BBDB is classified as ham or non-spam"
1318         (let ((who (nnmail-fetch-field "from"))
1319               (spam-split-group (if spam-split-symbolic-return
1320                                     'spam
1321                                   spam-split-group)))
1322           (when who
1323             (setq who (nth 1 (gnus-extract-address-components who)))
1324             (if (bbdb-search-simple nil who)
1325                 t
1326               (if spam-use-BBDB-exclusive
1327                   spam-split-group
1328                 nil))))))
1329
1330   (file-error (progn
1331                 (defalias 'bbdb-search-simple 'ignore)
1332                 (defalias 'spam-check-BBDB 'ignore)
1333                 (defalias 'spam-BBDB-register-routine 'ignore)
1334                 (defalias 'spam-enter-ham-BBDB 'ignore)
1335                 (defalias 'bbdb-create-internal 'ignore)
1336                 (defalias 'bbdb-delete-record-internal 'ignore)
1337                 (defalias 'bbdb-records 'ignore))))
1338
1339 \f
1340 ;;;; ifile
1341
1342 ;;; check the ifile backend; return nil if the mail was NOT classified
1343 ;;; as spam
1344
1345 (defun spam-get-ifile-database-parameter ()
1346   "Get the command-line parameter for ifile's database from
1347   spam-ifile-database-path."
1348   (if spam-ifile-database-path
1349       (format "--db-file=%s" spam-ifile-database-path)
1350     nil))
1351
1352 (defun spam-check-ifile ()
1353   "Check the ifile backend for the classification of this message."
1354   (let ((article-buffer-name (buffer-name))
1355         (spam-split-group (if spam-split-symbolic-return
1356                               'spam
1357                             spam-split-group))
1358         category return)
1359     (with-temp-buffer
1360       (let ((temp-buffer-name (buffer-name))
1361             (db-param (spam-get-ifile-database-parameter)))
1362         (save-excursion
1363           (set-buffer article-buffer-name)
1364           (apply 'call-process-region
1365                  (point-min) (point-max) spam-ifile-path
1366                  nil temp-buffer-name nil "-c"
1367                  (if db-param `(,db-param "-q") `("-q"))))
1368         ;; check the return now (we're back in the temp buffer)
1369         (goto-char (point-min))
1370         (if (not (eobp))
1371             (setq category (buffer-substring (point) (spam-point-at-eol))))
1372         (when (not (zerop (length category))) ; we need a category here
1373           (if spam-ifile-all-categories
1374               (setq return category)
1375             ;; else, if spam-ifile-all-categories is not set...
1376             (when (string-equal spam-ifile-spam-category category)
1377               (setq return spam-split-group)))))) ; note return is nil otherwise
1378     return))
1379
1380 (defun spam-ifile-register-with-ifile (articles category &optional unregister)
1381   "Register an article, given as a string, with a category.
1382 Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
1383   (let ((category (or category gnus-newsgroup-name))
1384         (add-or-delete-option (if unregister "-d" "-i"))
1385         (db (spam-get-ifile-database-parameter))
1386         parameters)
1387     (with-temp-buffer
1388       (dolist (article articles)
1389         (let ((article-string (spam-get-article-as-string article)))
1390           (when (stringp article-string)
1391             (insert article-string))))
1392       (apply 'call-process-region
1393              (point-min) (point-max) spam-ifile-path
1394              nil nil nil
1395              add-or-delete-option category
1396              (if db `(,db "-h") `("-h"))))))
1397
1398 (defun spam-ifile-register-spam-routine (articles &optional unregister)
1399   (spam-ifile-register-with-ifile articles spam-ifile-spam-category unregister))
1400
1401 (defun spam-ifile-unregister-spam-routine (articles)
1402   (spam-ifile-register-spam-routine articles t))
1403
1404 (defun spam-ifile-register-ham-routine (articles &optional unregister)
1405   (spam-ifile-register-with-ifile articles spam-ifile-ham-category unregister))
1406
1407 (defun spam-ifile-unregister-ham-routine (articles)
1408   (spam-ifile-register-ham-routine articles t))
1409
1410 \f
1411 ;;;; spam-stat
1412
1413 (condition-case nil
1414     (progn
1415       (let ((spam-stat-install-hooks nil))
1416         (require 'spam-stat))
1417
1418       (defun spam-check-stat ()
1419         "Check the spam-stat backend for the classification of this message"
1420         (let ((spam-split-group (if spam-split-symbolic-return
1421                                     'spam
1422                                   spam-split-group))
1423               (spam-stat-split-fancy-spam-group spam-split-group) ; override
1424               (spam-stat-buffer (buffer-name)) ; stat the current buffer
1425               category return)
1426           (spam-stat-split-fancy)))
1427
1428       (defun spam-stat-register-spam-routine (articles &optional unregister)
1429         (dolist (article articles)
1430           (let ((article-string (spam-get-article-as-string article)))
1431             (with-temp-buffer
1432               (insert article-string)
1433               (if unregister
1434                   (spam-stat-buffer-change-to-non-spam)
1435               (spam-stat-buffer-is-spam))))))
1436
1437       (defun spam-stat-unregister-spam-routine (articles)
1438         (spam-stat-register-spam-routine articles t))
1439
1440       (defun spam-stat-register-ham-routine (articles &optional unregister)
1441         (dolist (article articles)
1442           (let ((article-string (spam-get-article-as-string article)))
1443             (with-temp-buffer
1444               (insert article-string)
1445               (if unregister
1446                   (spam-stat-buffer-change-to-spam)
1447               (spam-stat-buffer-is-non-spam))))))
1448
1449       (defun spam-stat-unregister-ham-routine (articles)
1450         (spam-stat-register-ham-routine articles t))
1451
1452       (defun spam-maybe-spam-stat-load ()
1453         (when spam-use-stat (spam-stat-load)))
1454
1455       (defun spam-maybe-spam-stat-save ()
1456         (when spam-use-stat (spam-stat-save))))
1457
1458   (file-error (progn
1459                 (defalias 'spam-stat-load 'ignore)
1460                 (defalias 'spam-stat-save 'ignore)
1461                 (defalias 'spam-maybe-spam-stat-load 'ignore)
1462                 (defalias 'spam-maybe-spam-stat-save 'ignore)
1463                 (defalias 'spam-stat-register-ham-routine 'ignore)
1464                 (defalias 'spam-stat-unregister-ham-routine 'ignore)
1465                 (defalias 'spam-stat-register-spam-routine 'ignore)
1466                 (defalias 'spam-stat-unregister-spam-routine 'ignore)
1467                 (defalias 'spam-stat-buffer-is-spam 'ignore)
1468                 (defalias 'spam-stat-buffer-change-to-spam 'ignore)
1469                 (defalias 'spam-stat-buffer-is-non-spam 'ignore)
1470                 (defalias 'spam-stat-buffer-change-to-non-spam 'ignore)
1471                 (defalias 'spam-stat-split-fancy 'ignore)
1472                 (defalias 'spam-check-stat 'ignore))))
1473
1474 \f
1475
1476 ;;;; Blacklists and whitelists.
1477
1478 (defvar spam-whitelist-cache nil)
1479 (defvar spam-blacklist-cache nil)
1480
1481 (defun spam-kill-whole-line ()
1482   (beginning-of-line)
1483   (let ((kill-whole-line t))
1484     (kill-line)))
1485
1486 ;;; address can be a list, too
1487 (defun spam-enter-whitelist (address &optional remove)
1488   "Enter ADDRESS (list or single) into the whitelist.
1489 With a non-nil REMOVE, remove them."
1490   (interactive "sAddress: ")
1491   (spam-enter-list address spam-whitelist remove)
1492   (setq spam-whitelist-cache nil))
1493
1494 ;;; address can be a list, too
1495 (defun spam-enter-blacklist (address &optional remove)
1496   "Enter ADDRESS (list or single) into the blacklist.
1497 With a non-nil REMOVE, remove them."
1498   (interactive "sAddress: ")
1499   (spam-enter-list address spam-blacklist remove)
1500   (setq spam-blacklist-cache nil))
1501
1502 (defun spam-enter-list (addresses file &optional remove)
1503   "Enter ADDRESSES into the given FILE.
1504 Either the whitelist or the blacklist files can be used.  With
1505 REMOVE not nil, remove the ADDRESSES."
1506   (if (stringp addresses)
1507       (spam-enter-list (list addresses) file remove)
1508     ;; else, we have a list of addresses here
1509     (unless (file-exists-p (file-name-directory file))
1510       (make-directory (file-name-directory file) t))
1511     (save-excursion
1512       (set-buffer
1513        (find-file-noselect file))
1514       (dolist (a addresses)
1515         (when (stringp a)
1516           (goto-char (point-min))
1517           (if (re-search-forward (regexp-quote a) nil t)
1518               ;; found the address
1519               (when remove
1520                 (spam-kill-whole-line))
1521             ;; else, the address was not found
1522             (unless remove
1523               (goto-char (point-max))
1524               (unless (bobp)
1525                 (insert "\n"))
1526               (insert a "\n")))))
1527       (save-buffer))))
1528
1529 ;;; returns t if the sender is in the whitelist, nil or
1530 ;;; spam-split-group otherwise
1531 (defun spam-check-whitelist ()
1532   ;; FIXME!  Should it detect when file timestamps change?
1533   (let ((spam-split-group (if spam-split-symbolic-return
1534                               'spam
1535                             spam-split-group)))
1536     (unless spam-whitelist-cache
1537       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
1538     (if (spam-from-listed-p spam-whitelist-cache)
1539         t
1540       (if spam-use-whitelist-exclusive
1541           spam-split-group
1542         nil))))
1543
1544 (defun spam-check-blacklist ()
1545   ;; FIXME!  Should it detect when file timestamps change?
1546   (let ((spam-split-group (if spam-split-symbolic-return
1547                               'spam
1548                             spam-split-group)))
1549     (unless spam-blacklist-cache
1550       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
1551     (and (spam-from-listed-p spam-blacklist-cache) spam-split-group)))
1552
1553 (defun spam-parse-list (file)
1554   (when (file-readable-p file)
1555     (let (contents address)
1556       (with-temp-buffer
1557         (insert-file-contents file)
1558         (while (not (eobp))
1559           (setq address (buffer-substring (point) (spam-point-at-eol)))
1560           (forward-line 1)
1561           ;; insert the e-mail address if detected, otherwise the raw data
1562           (unless (zerop (length address))
1563             (let ((pure-address (nth 1 (gnus-extract-address-components address))))
1564               (push (or pure-address address) contents)))))
1565       (nreverse contents))))
1566
1567 (defun spam-from-listed-p (cache)
1568   (let ((from (nnmail-fetch-field "from"))
1569         found)
1570     (while cache
1571       (let ((address (pop cache)))
1572         (unless (zerop (length address)) ; 0 for a nil address too
1573           (setq address (regexp-quote address))
1574           ;; fix regexp-quote's treatment of user-intended regexes
1575           (while (string-match "\\\\\\*" address)
1576             (setq address (replace-match ".*" t t address))))
1577         (when (and address (string-match address from))
1578           (setq found t
1579                 cache nil))))
1580     found))
1581
1582 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
1583   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
1584         (declassification (if blacklist 'ham 'spam))
1585         (enter-function
1586          (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
1587         (remove-function
1588          (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
1589         from addresses unregister-list)
1590     (dolist (article articles)
1591       (let ((from (spam-fetch-field-from-fast article))
1592             (id (spam-fetch-field-message-id-fast article))
1593             sender-ignored)
1594         (when (stringp from)
1595           (dolist (ignore-regex spam-blacklist-ignored-regexes)
1596             (when (and (not sender-ignored)
1597                        (stringp ignore-regex)
1598                        (string-match ignore-regex from))
1599               (setq sender-ignored t)))
1600           ;; remember the messages we need to unregister, unless remove is set
1601           (when (and
1602                  (null unregister)
1603                  (spam-log-unregistration-needed-p
1604                   id 'process declassification de-symbol))
1605             (push from unregister-list))
1606           (unless sender-ignored
1607             (push from addresses)))))
1608
1609     (if unregister
1610         (funcall enter-function addresses t) ; unregister all these addresses
1611       ;; else, register normally and unregister what we need to
1612       (funcall remove-function unregister-list t)
1613       (dolist (article unregister-list)
1614         (spam-log-undo-registration
1615          (spam-fetch-field-message-id-fast article)
1616          'process
1617          declassification
1618          de-symbol))
1619       (funcall enter-function addresses nil))))
1620
1621 (defun spam-blacklist-unregister-routine (articles)
1622   (spam-blacklist-register-routine articles t))
1623
1624 (defun spam-blacklist-register-routine (articles &optional unregister)
1625   (spam-filelist-register-routine articles t unregister))
1626
1627 (defun spam-whitelist-unregister-routine (articles)
1628   (spam-whitelist-register-routine articles t))
1629
1630 (defun spam-whitelist-register-routine (articles &optional unregister)
1631   (spam-filelist-register-routine articles nil unregister))
1632
1633 \f
1634 ;;;; Spam-report glue
1635 (defun spam-report-gmane-register-routine (articles)
1636   (when articles
1637     (apply 'spam-report-gmane articles)))
1638
1639 \f
1640 ;;;; Bogofilter
1641 (defun spam-check-bogofilter-headers (&optional score)
1642   (let ((header (nnmail-fetch-field spam-bogofilter-header))
1643         (spam-split-group (if spam-split-symbolic-return
1644                               'spam
1645                             spam-split-group)))
1646     (when header                        ; return nil when no header
1647       (if score                         ; scoring mode
1648           (if (string-match "spamicity=\\([0-9.]+\\)" header)
1649               (match-string 1 header)
1650             "0")
1651         ;; spam detection mode
1652         (when (string-match spam-bogofilter-bogosity-positive-spam-header
1653                             header)
1654           spam-split-group)))))
1655
1656 ;; return something sensible if the score can't be determined
1657 (defun spam-bogofilter-score ()
1658   "Get the Bogofilter spamicity score"
1659   (interactive)
1660   (save-window-excursion
1661     (gnus-summary-show-article t)
1662     (set-buffer gnus-article-buffer)
1663     (let ((score (or (spam-check-bogofilter-headers t)
1664                      (spam-check-bogofilter t))))
1665       (message "Spamicity score %s" score)
1666       (or score "0"))
1667     (gnus-summary-show-article)))
1668
1669 (defun spam-check-bogofilter (&optional score)
1670   "Check the Bogofilter backend for the classification of this message"
1671   (let ((article-buffer-name (buffer-name))
1672         (db spam-bogofilter-database-directory)
1673         return)
1674     (with-temp-buffer
1675       (let ((temp-buffer-name (buffer-name)))
1676         (save-excursion
1677           (set-buffer article-buffer-name)
1678           (apply 'call-process-region
1679                  (point-min) (point-max)
1680                  spam-bogofilter-path
1681                  nil temp-buffer-name nil
1682                  (if db `("-d" ,db "-v") `("-v"))))
1683         (setq return (spam-check-bogofilter-headers score))))
1684     return))
1685
1686 (defun spam-bogofilter-register-with-bogofilter (articles
1687                                                  spam
1688                                                  &optional unregister)
1689   "Register an article, given as a string, as spam or non-spam."
1690   (dolist (article articles)
1691     (let ((article-string (spam-get-article-as-string article))
1692           (db spam-bogofilter-database-directory)
1693           (switch (if unregister
1694                       (if spam
1695                           spam-bogofilter-spam-strong-switch
1696                         spam-bogofilter-ham-strong-switch)
1697                     (if spam
1698                         spam-bogofilter-spam-switch
1699                       spam-bogofilter-ham-switch))))
1700       (when (stringp article-string)
1701         (with-temp-buffer
1702           (insert article-string)
1703
1704           (apply 'call-process-region
1705                  (point-min) (point-max)
1706                  spam-bogofilter-path
1707                  nil nil nil switch
1708                  (if db `("-d" ,db "-v") `("-v"))))))))
1709
1710 (defun spam-bogofilter-register-spam-routine (articles &optional unregister)
1711   (spam-bogofilter-register-with-bogofilter articles t unregister))
1712
1713 (defun spam-bogofilter-unregister-spam-routine (articles)
1714   (spam-bogofilter-register-spam-routine articles t))
1715
1716 (defun spam-bogofilter-register-ham-routine (articles &optional unregister)
1717   (spam-bogofilter-register-with-bogofilter articles nil unregister))
1718
1719 (defun spam-bogofilter-unregister-ham-routine (articles)
1720   (spam-bogofilter-register-ham-routine articles t))
1721
1722
1723 \f
1724 ;;;; spamoracle
1725 (defun spam-check-spamoracle ()
1726   "Run spamoracle on an article to determine whether it's spam."
1727   (let ((article-buffer-name (buffer-name))
1728         (spam-split-group (if spam-split-symbolic-return
1729                               'spam
1730                             spam-split-group)))
1731     (with-temp-buffer
1732       (let ((temp-buffer-name (buffer-name)))
1733         (save-excursion
1734           (set-buffer article-buffer-name)
1735           (let ((status
1736                  (apply 'call-process-region
1737                         (point-min) (point-max)
1738                         spam-spamoracle-binary
1739                         nil temp-buffer-name nil
1740                         (if spam-spamoracle-database
1741                             `("-f" ,spam-spamoracle-database "mark")
1742                           '("mark")))))
1743             (if (eq 0 status)
1744                 (progn
1745                   (set-buffer temp-buffer-name)
1746                   (goto-char (point-min))
1747                   (when (re-search-forward "^X-Spam: yes;" nil t)
1748                     spam-split-group))
1749               (error "Error running spamoracle" status))))))))
1750
1751 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
1752   "Run spamoracle in training mode."
1753   (with-temp-buffer
1754     (let ((temp-buffer-name (buffer-name)))
1755       (save-excursion
1756         (goto-char (point-min))
1757         (dolist (article articles)
1758           (insert (spam-get-article-as-string article)))
1759         (let* ((arg (if (spam-xor unregister article-is-spam-p)
1760                         "-spam"
1761                       "-good"))
1762                (status
1763                 (apply 'call-process-region
1764                        (point-min) (point-max)
1765                        spam-spamoracle-binary
1766                        nil temp-buffer-name nil
1767                        (if spam-spamoracle-database
1768                            `("-f" ,spam-spamoracle-database
1769                              "add" ,arg)
1770                          `("add" ,arg)))))
1771           (when (not (eq 0 status))
1772             (error "Error running spamoracle" status)))))))
1773
1774 (defun spam-spamoracle-learn-ham (articles &optional unregister)
1775   (spam-spamoracle-learn articles nil unregister))
1776
1777 (defun spam-spamoracle-unlearn-ham (articles &optional unregister)
1778   (spam-spamoracle-learn-ham articles t))
1779
1780 (defun spam-spamoracle-learn-spam (articles &optional unregister)
1781   (spam-spamoracle-learn articles t unregister))
1782
1783 (defun spam-spamoracle-unlearn-spam (articles &optional unregister)
1784   (spam-spamoracle-learn-spam articles t))
1785
1786 \f
1787 ;;;; Hooks
1788
1789 ;;;###autoload
1790 (defun spam-initialize ()
1791   "Install the spam.el hooks and do other initialization"
1792   (interactive)
1793   (setq spam-install-hooks t)
1794   ;; TODO: How do we redo this every time spam-face is customized?
1795   (push '((eq mark gnus-spam-mark) . spam-face)
1796         gnus-summary-highlight)
1797   ;; Add hooks for loading and saving the spam stats
1798   (add-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
1799   (add-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
1800   (add-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
1801   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
1802   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
1803   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
1804   (add-hook 'gnus-summary-prepare-hook 'spam-find-spam))
1805
1806 (defun spam-unload-hook ()
1807   "Uninstall the spam.el hooks"
1808   (interactive)
1809   (remove-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
1810   (remove-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
1811   (remove-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
1812   (remove-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
1813   (remove-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
1814   (remove-hook 'gnus-get-new-news-hook 'spam-setup-widening)
1815   (remove-hook 'gnus-summary-prepare-hook 'spam-find-spam))
1816
1817 (when spam-install-hooks
1818   (spam-initialize))
1819
1820 (provide 'spam)
1821
1822 ;;; spam.el ends here.
1823
1824 (provide 'spam)
1825
1826 ;;; spam.el ends here