* spam.el (spam-point-at-eol): Replace with point-at-eol.
[gnus] / 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 (defun spam-get-article-as-string (article)
767   (let ((article-buffer (spam-get-article-as-buffer article))
768         article-string)
769     (when article-buffer
770       (save-window-excursion
771         (set-buffer article-buffer)
772         (setq article-string (buffer-string))))
773     article-string))
774
775 (defun spam-get-article-as-buffer (article)
776   (let ((article-buffer))
777     (when (numberp article)
778       (save-window-excursion
779         (gnus-summary-goto-subject article)
780         (gnus-summary-show-article t)
781         (setq article-buffer (get-buffer gnus-article-buffer))))
782     article-buffer))
783
784 ;; disabled for now
785 ;; (defun spam-get-article-as-filename (article)
786 ;;   (let ((article-filename))
787 ;;     (when (numberp article)
788 ;;       (nnml-possibly-change-directory
789 ;;        (gnus-group-real-name gnus-newsgroup-name))
790 ;;       (setq article-filename (expand-file-name
791 ;;                              (int-to-string article) nnml-current-directory)))
792 ;;     (if (file-exists-p article-filename)
793 ;;      article-filename
794 ;;       nil)))
795
796 (defun spam-fetch-field-from-fast (article)
797   "Fetch the `from' field quickly, using the internal gnus-data-list function"
798   (if (and (numberp article)
799            (assoc article (gnus-data-list nil)))
800       (mail-header-from
801        (gnus-data-header (assoc article (gnus-data-list nil))))
802     nil))
803
804 (defun spam-fetch-field-subject-fast (article)
805   "Fetch the `subject' field quickly, using the internal
806   gnus-data-list function"
807   (if (and (numberp article)
808            (assoc article (gnus-data-list nil)))
809       (mail-header-subject
810        (gnus-data-header (assoc article (gnus-data-list nil))))
811     nil))
812
813 (defun spam-fetch-field-message-id-fast (article)
814   "Fetch the `Message-ID' field quickly, using the internal
815   gnus-data-list function"
816   (if (and (numberp article)
817            (assoc article (gnus-data-list nil)))
818       (mail-header-message-id
819        (gnus-data-header (assoc article (gnus-data-list nil))))
820     nil))
821
822 \f
823 ;;;; Spam determination.
824
825 (defvar spam-list-of-checks
826   '((spam-use-blacklist          . spam-check-blacklist)
827     (spam-use-regex-headers      . spam-check-regex-headers)
828     (spam-use-regex-body         . spam-check-regex-body)
829     (spam-use-whitelist          . spam-check-whitelist)
830     (spam-use-BBDB               . spam-check-BBDB)
831     (spam-use-ifile              . spam-check-ifile)
832     (spam-use-spamoracle         . spam-check-spamoracle)
833     (spam-use-stat               . spam-check-stat)
834     (spam-use-blackholes         . spam-check-blackholes)
835     (spam-use-hashcash           . spam-check-hashcash)
836     (spam-use-bogofilter-headers . spam-check-bogofilter-headers)
837     (spam-use-bogofilter         . spam-check-bogofilter))
838   "The spam-list-of-checks list contains pairs associating a
839 parameter variable with a spam checking function.  If the
840 parameter variable is true, then the checking function is called,
841 and its value decides what happens.  Each individual check may
842 return nil, t, or a mailgroup name.  The value nil means that the
843 check does not yield a decision, and so, that further checks are
844 needed.  The value t means that the message is definitely not
845 spam, and that further spam checks should be inhibited.
846 Otherwise, a mailgroup name or the symbol 'spam (depending on
847 spam-split-symbolic-return) is returned where the mail should go,
848 and further checks are also inhibited.  The usual mailgroup name
849 is the value of `spam-split-group', meaning that the message is
850 definitely a spam.")
851
852 (defvar spam-list-of-statistical-checks
853   '(spam-use-ifile
854     spam-use-regex-body
855     spam-use-stat
856     spam-use-bogofilter
857     spam-use-spamoracle)
858   "The spam-list-of-statistical-checks list contains all the mail
859 splitters that need to have the full message body available.")
860
861 ;;;TODO: modify to invoke self with each check if invoked without specifics
862 (defun spam-split (&rest specific-checks)
863   "Split this message into the `spam' group if it is spam.
864 This function can be used as an entry in the variable `nnmail-split-fancy',
865 for example like this: (: spam-split).  It can take checks as
866 parameters.  A string as a parameter will set the
867 spam-split-group to that string.
868
869 See the Info node `(gnus)Fancy Mail Splitting' for more details."
870   (interactive)
871   (setq spam-split-last-successful-check nil)
872   (unless spam-split-disabled
873     (let ((spam-split-group-choice spam-split-group))
874       (dolist (check specific-checks)
875         (when (stringp check)
876           (setq spam-split-group-choice check)
877           (setq specific-checks (delq check specific-checks))))
878
879       (let ((spam-split-group spam-split-group-choice))
880         (save-excursion
881           (save-restriction
882             (dolist (check spam-list-of-statistical-checks)
883               (when (and (symbolp check) (symbol-value check))
884                 (widen)
885                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
886                               (symbol-name check))
887                 (return)))
888             ;;   (progn (widen) (debug (buffer-string)))
889             (let ((list-of-checks spam-list-of-checks)
890                   decision)
891               (while (and list-of-checks (not decision))
892                 (let ((pair (pop list-of-checks)))
893                   (when (and (symbol-value (car pair))
894                              (or (null specific-checks)
895                                  (memq (car pair) specific-checks)))
896                     (gnus-message 5 "spam-split: calling the %s function"
897                                   (symbol-name (cdr pair)))
898                     (setq decision (funcall (cdr pair)))
899                     ;; if we got a decision at all, save the current check
900                     (when decision
901                       (setq spam-split-last-successful-check (car pair)))
902
903                     (when (eq decision 'spam)
904                       (if spam-split-symbolic-return
905                           (setq decision spam-split-group)
906                         (gnus-error
907                          5
908                          (format "spam-split got %s but %s is nil"
909                                  (symbol-name decision)
910                                  (symbol-name spam-split-symbolic-return))))))))
911               (if (eq decision t)
912                   (if spam-split-symbolic-return-positive 'ham nil)
913                 decision))))))))
914
915 (defun spam-find-spam ()
916   "This function will detect spam in the current newsgroup using spam-split."
917   (interactive)
918
919   (let* ((group gnus-newsgroup-name)
920          (autodetect (gnus-parameter-spam-autodetect group))
921          (methods (gnus-parameter-spam-autodetect-methods group))
922          (first-method (nth 0 methods)))
923   (when (and autodetect
924              (not (equal first-method 'none)))
925     (mapcar
926      (lambda (article)
927        (let ((id (spam-fetch-field-message-id-fast article))
928              (subject (spam-fetch-field-subject-fast article))
929              (sender (spam-fetch-field-from-fast article)))
930          (unless (and spam-log-to-registry
931                       (spam-log-registered-p id 'incoming))
932            (let* ((spam-split-symbolic-return t)
933                   (spam-split-symbolic-return-positive t)
934                   (split-return
935                    (with-temp-buffer
936                      (gnus-request-article-this-buffer
937                       article
938                       group)
939                      (if (or (null first-method)
940                              (equal first-method 'default))
941                          (spam-split)
942                        (apply 'spam-split methods)))))
943              (if (equal split-return 'spam)
944                  (gnus-summary-mark-article article gnus-spam-mark))
945
946              (when (and split-return spam-log-to-registry)
947                (when (zerop (gnus-registry-group-count id))
948                  (gnus-registry-add-group
949                   id group subject sender))
950
951                (spam-log-processing-to-registry
952                 id
953                 'incoming
954                 split-return
955                 spam-split-last-successful-check
956                 group))))))
957      (if spam-autodetect-recheck-messages
958          gnus-newsgroup-articles
959        gnus-newsgroup-unseen)))))
960
961 (defvar spam-registration-functions
962   ;; first the ham register, second the spam register function
963   ;; third the ham unregister, fourth the spam unregister function
964   '((spam-use-blacklist  nil
965                          spam-blacklist-register-routine
966                          nil
967                          spam-blacklist-unregister-routine)
968     (spam-use-whitelist  spam-whitelist-register-routine
969                          nil
970                          spam-whitelist-unregister-routine
971                          nil)
972     (spam-use-BBDB       spam-BBDB-register-routine
973                          nil
974                          spam-BBDB-unregister-routine
975                          nil)
976     (spam-use-ifile      spam-ifile-register-ham-routine
977                          spam-ifile-register-spam-routine
978                          spam-ifile-unregister-ham-routine
979                          spam-ifile-unregister-spam-routine)
980     (spam-use-spamoracle spam-spamoracle-learn-ham
981                          spam-spamoracle-learn-spam
982                          spam-spamoracle-unlearn-ham
983                          spam-spamoracle-unlearn-spam)
984     (spam-use-stat       spam-stat-register-ham-routine
985                          spam-stat-register-spam-routine
986                          spam-stat-unregister-ham-routine
987                          spam-stat-unregister-spam-routine)
988     ;; note that spam-use-gmane is not a legitimate check
989     (spam-use-gmane      nil
990                          spam-report-gmane-register-routine
991                          ;; does Gmane support unregistration?
992                          nil
993                          nil)
994     (spam-use-bogofilter spam-bogofilter-register-ham-routine
995                          spam-bogofilter-register-spam-routine
996                          spam-bogofilter-unregister-ham-routine
997                          spam-bogofilter-unregister-spam-routine))
998   "The spam-registration-functions list contains pairs
999 associating a parameter variable with the ham and spam
1000 registration functions, and the ham and spam unregistration
1001 functions")
1002
1003 (defun spam-classification-valid-p (classification)
1004   (or  (eq classification 'spam)
1005        (eq classification 'ham)))
1006
1007 (defun spam-process-type-valid-p (process-type)
1008   (or  (eq process-type 'incoming)
1009        (eq process-type 'process)))
1010
1011 (defun spam-registration-check-valid-p (check)
1012   (assoc check spam-registration-functions))
1013
1014 (defun spam-unregistration-check-valid-p (check)
1015   (assoc check spam-registration-functions))
1016
1017 (defun spam-registration-function (classification check)
1018   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1019     (if (eq classification 'spam)
1020         (nth 1 flist)
1021       (nth 0 flist))))
1022
1023 (defun spam-unregistration-function (classification check)
1024   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1025     (if (eq classification 'spam)
1026         (nth 3 flist)
1027       (nth 2 flist))))
1028
1029 (defun spam-list-articles (articles classification)
1030   (let ((mark-check (if (eq classification 'spam)
1031                         'spam-group-spam-mark-p
1032                       'spam-group-ham-mark-p))
1033         list mark-cache-yes mark-cache-no)
1034     (dolist (article articles)
1035       (let ((mark (gnus-summary-article-mark article)))
1036         (unless (memq mark mark-cache-no)
1037           (if (memq mark mark-cache-yes)
1038               (push article list)
1039             ;; else, we have to actually check the mark
1040             (if (funcall mark-check
1041                          gnus-newsgroup-name
1042                          mark)
1043                 (progn
1044                   (push article list)
1045                   (push mark mark-cache-yes))
1046               (push mark mark-cache-no))))))
1047     list))
1048
1049 (defun spam-register-routine (classification
1050                               check
1051                               &optional unregister
1052                               specific-articles)
1053   (when (and (spam-classification-valid-p classification)
1054              (spam-registration-check-valid-p check))
1055     (let* ((register-function
1056             (spam-registration-function classification check))
1057            (unregister-function
1058             (spam-unregistration-function classification check))
1059            (run-function (if unregister
1060                              unregister-function
1061                            register-function))
1062            (log-function (if unregister
1063                              'spam-log-undo-registration
1064                            'spam-log-processing-to-registry))
1065            article articles)
1066
1067       (when run-function
1068         ;; make list of articles, using specific-articles if given
1069         (setq articles (or specific-articles
1070                            (spam-list-articles
1071                             gnus-newsgroup-articles
1072                             classification)))
1073         ;; process them
1074         (gnus-message 5 "%s %d %s articles with classification %s, check %s"
1075                       (if unregister "Unregistering" "Registering")
1076                       (length articles)
1077                       (if specific-articles "specific" "")
1078                       (symbol-name classification)
1079                       (symbol-name check))
1080         (funcall run-function articles)
1081         ;; now log all the registrations (or undo them, depending on unregister)
1082         (dolist (article articles)
1083           (funcall log-function
1084                    (spam-fetch-field-message-id-fast article)
1085                    'process
1086                    classification
1087                    check
1088                    gnus-newsgroup-name))))))
1089
1090 ;;; log a ham- or spam-processor invocation to the registry
1091 (defun spam-log-processing-to-registry (id type classification check group)
1092   (when spam-log-to-registry
1093     (if (and (stringp id)
1094              (stringp group)
1095              (spam-process-type-valid-p type)
1096              (spam-classification-valid-p classification)
1097              (spam-registration-check-valid-p check))
1098         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1099               (cell (list classification check group)))
1100           (push cell cell-list)
1101           (gnus-registry-store-extra-entry
1102            id
1103            type
1104            cell-list))
1105
1106       (gnus-message 5 (format "%s called with bad ID, type, classification, check, or group"
1107                               "spam-log-processing-to-registry")))))
1108
1109 ;;; check if a ham- or spam-processor registration has been done
1110 (defun spam-log-registered-p (id type)
1111   (when spam-log-to-registry
1112     (if (and (stringp id)
1113              (spam-process-type-valid-p type))
1114         (cdr-safe (gnus-registry-fetch-extra id type))
1115       (progn
1116         (gnus-message 5 (format "%s called with bad ID, type, classification, or check"
1117                                 "spam-log-registered-p"))
1118         nil))))
1119
1120 ;;; check if a ham- or spam-processor registration needs to be undone
1121 (defun spam-log-unregistration-needed-p (id type classification check)
1122   (when spam-log-to-registry
1123     (if (and (stringp id)
1124              (spam-process-type-valid-p type)
1125              (spam-classification-valid-p classification)
1126              (spam-registration-check-valid-p check))
1127         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1128               found)
1129           (dolist (cell cell-list)
1130             (unless found
1131               (when (and (eq classification (nth 0 cell))
1132                          (eq check (nth 1 cell)))
1133                 (setq found t))))
1134           found)
1135       (progn
1136         (gnus-message 5 (format "%s called with bad ID, type, classification, or check"
1137                                 "spam-log-unregistration-needed-p"))
1138         nil))))
1139
1140
1141 ;;; undo a ham- or spam-processor registration (the group is not used)
1142 (defun spam-log-undo-registration (id type classification check &optional group)
1143   (when (and spam-log-to-registry
1144              (spam-log-unregistration-needed-p id type classification check))
1145     (if (and (stringp id)
1146              (spam-process-type-valid-p type)
1147              (spam-classification-valid-p classification)
1148              (spam-registration-check-valid-p check))
1149         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1150               new-cell-list found)
1151           (dolist (cell cell-list)
1152             (unless (and (eq classification (nth 0 cell))
1153                          (eq check (nth 1 cell)))
1154               (push cell new-cell-list)))
1155           (gnus-registry-store-extra-entry
1156            id
1157            type
1158            new-cell-list))
1159       (progn
1160         (gnus-message 5 (format "%s called with bad ID, type, check, or group"
1161                                 "spam-log-undo-registration"))
1162         nil))))
1163
1164 ;;; set up IMAP widening if it's necessary
1165 (defun spam-setup-widening ()
1166   (dolist (check spam-list-of-statistical-checks)
1167     (when (symbol-value check)
1168       (setq nnimap-split-download-body-default t))))
1169
1170 \f
1171 ;;;; Regex body
1172
1173 (defun spam-check-regex-body ()
1174   (let ((spam-regex-headers-ham spam-regex-body-ham)
1175         (spam-regex-headers-spam spam-regex-body-spam))
1176     (spam-check-regex-headers t)))
1177
1178 \f
1179 ;;;; Regex headers
1180
1181 (defun spam-check-regex-headers (&optional body)
1182   (let ((type (if body "body" "header"))
1183         (spam-split-group (if spam-split-symbolic-return
1184                               'spam
1185                             spam-split-group))
1186         ret found)
1187     (dolist (h-regex spam-regex-headers-ham)
1188       (unless found
1189         (goto-char (point-min))
1190         (when (re-search-forward h-regex nil t)
1191           (message "Ham regex %s search positive." type)
1192           (setq found t))))
1193     (dolist (s-regex spam-regex-headers-spam)
1194       (unless found
1195         (goto-char (point-min))
1196         (when (re-search-forward s-regex nil t)
1197           (message "Spam regex %s search positive." type)
1198           (setq found t)
1199           (setq ret spam-split-group))))
1200     ret))
1201
1202 \f
1203 ;;;; Blackholes.
1204
1205 (defun spam-reverse-ip-string (ip)
1206   (when (stringp ip)
1207     (mapconcat 'identity
1208                (nreverse (split-string ip "\\."))
1209                ".")))
1210
1211 (defun spam-check-blackholes ()
1212   "Check the Received headers for blackholed relays."
1213   (let ((headers (nnmail-fetch-field "received"))
1214         (spam-split-group (if spam-split-symbolic-return
1215                               'spam
1216                             spam-split-group))
1217         ips matches)
1218     (when headers
1219       (with-temp-buffer
1220         (insert headers)
1221         (goto-char (point-min))
1222         (gnus-message 5 "Checking headers for relay addresses")
1223         (while (re-search-forward
1224                 "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
1225           (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
1226           (push (spam-reverse-ip-string (match-string 1))
1227                 ips)))
1228       (dolist (server spam-blackhole-servers)
1229         (dolist (ip ips)
1230           (unless (and spam-blackhole-good-server-regex
1231                        ;; match the good-server-regex against the reversed (again) IP string
1232                        (string-match
1233                         spam-blackhole-good-server-regex
1234                         (spam-reverse-ip-string ip)))
1235             (unless matches
1236               (let ((query-string (concat ip "." server)))
1237                 (if spam-use-dig
1238                     (let ((query-result (query-dig query-string)))
1239                       (when query-result
1240                         (gnus-message 5 "(DIG): positive blackhole check '%s'"
1241                                       query-result)
1242                         (push (list ip server query-result)
1243                               matches)))
1244                   ;; else, if not using dig.el
1245                   (when (query-dns query-string)
1246                     (gnus-message 5 "positive blackhole check")
1247                     (push (list ip server (query-dns query-string 'TXT))
1248                           matches)))))))))
1249     (when matches
1250       spam-split-group)))
1251 \f
1252 ;;;; Hashcash.
1253
1254 (condition-case nil
1255     (progn
1256       (require 'hashcash)
1257
1258       (defun spam-check-hashcash ()
1259         "Check the headers for hashcash payments."
1260         (mail-check-payment)))   ;mail-check-payment returns a boolean
1261
1262   (file-error (progn
1263                 (defalias 'mail-check-payment 'ignore)
1264                 (defalias 'spam-check-hashcash 'ignore))))
1265 \f
1266 ;;;; BBDB
1267
1268 ;;; original idea for spam-check-BBDB from Alexander Kotelnikov
1269 ;;; <sacha@giotto.sj.ru>
1270
1271 ;; all this is done inside a condition-case to trap errors
1272
1273 (condition-case nil
1274     (progn
1275       (require 'bbdb)
1276       (require 'bbdb-com)
1277
1278       (defun spam-enter-ham-BBDB (addresses &optional remove)
1279         "Enter an address into the BBDB; implies ham (non-spam) sender"
1280         (dolist (from addresses)
1281           (when (stringp from)
1282             (let* ((parsed-address (gnus-extract-address-components from))
1283                    (name (or (nth 0 parsed-address) "Ham Sender"))
1284                    (remove-function (if remove
1285                                         'bbdb-delete-record-internal
1286                                       'ignore))
1287                    (net-address (nth 1 parsed-address))
1288                    (record (and net-address
1289                                 (bbdb-search-simple nil net-address))))
1290               (when net-address
1291                 (gnus-message 5 "%s address %s %s BBDB"
1292                               (if remove "Deleting" "Adding")
1293                               from
1294                               (if remove "from" "to"))
1295                 (if record
1296                     (funcall remove-function record)
1297                   (bbdb-create-internal name nil net-address nil nil
1298                                         "ham sender added by spam.el")))))))
1299
1300       (defun spam-BBDB-register-routine (articles &optional unregister)
1301         (let (addresses)
1302           (dolist (article articles)
1303             (when (stringp (spam-fetch-field-from-fast article))
1304               (push (spam-fetch-field-from-fast article) addresses)))
1305           ;; now do the register/unregister action
1306           (spam-enter-ham-BBDB addresses unregister)))
1307
1308       (defun spam-BBDB-unregister-routine (articles)
1309         (spam-BBDB-register-routine articles t))
1310
1311       (defun spam-check-BBDB ()
1312         "Mail from people in the BBDB is classified as ham or non-spam"
1313         (let ((who (nnmail-fetch-field "from"))
1314               (spam-split-group (if spam-split-symbolic-return
1315                                     'spam
1316                                   spam-split-group)))
1317           (when who
1318             (setq who (nth 1 (gnus-extract-address-components who)))
1319             (if (bbdb-search-simple nil who)
1320                 t
1321               (if spam-use-BBDB-exclusive
1322                   spam-split-group
1323                 nil))))))
1324
1325   (file-error (progn
1326                 (defalias 'bbdb-search-simple 'ignore)
1327                 (defalias 'spam-check-BBDB 'ignore)
1328                 (defalias 'spam-BBDB-register-routine 'ignore)
1329                 (defalias 'spam-enter-ham-BBDB 'ignore)
1330                 (defalias 'bbdb-create-internal 'ignore)
1331                 (defalias 'bbdb-delete-record-internal 'ignore)
1332                 (defalias 'bbdb-records 'ignore))))
1333
1334 \f
1335 ;;;; ifile
1336
1337 ;;; check the ifile backend; return nil if the mail was NOT classified
1338 ;;; as spam
1339
1340 (defun spam-get-ifile-database-parameter ()
1341   "Get the command-line parameter for ifile's database from
1342   spam-ifile-database-path."
1343   (if spam-ifile-database-path
1344       (format "--db-file=%s" spam-ifile-database-path)
1345     nil))
1346
1347 (defun spam-check-ifile ()
1348   "Check the ifile backend for the classification of this message."
1349   (let ((article-buffer-name (buffer-name))
1350         (spam-split-group (if spam-split-symbolic-return
1351                               'spam
1352                             spam-split-group))
1353         category return)
1354     (with-temp-buffer
1355       (let ((temp-buffer-name (buffer-name))
1356             (db-param (spam-get-ifile-database-parameter)))
1357         (save-excursion
1358           (set-buffer article-buffer-name)
1359           (apply 'call-process-region
1360                  (point-min) (point-max) spam-ifile-path
1361                  nil temp-buffer-name nil "-c"
1362                  (if db-param `(,db-param "-q") `("-q"))))
1363         ;; check the return now (we're back in the temp buffer)
1364         (goto-char (point-min))
1365         (if (not (eobp))
1366             (setq category (buffer-substring (point) (point-at-eol))))
1367         (when (not (zerop (length category))) ; we need a category here
1368           (if spam-ifile-all-categories
1369               (setq return category)
1370             ;; else, if spam-ifile-all-categories is not set...
1371             (when (string-equal spam-ifile-spam-category category)
1372               (setq return spam-split-group)))))) ; note return is nil otherwise
1373     return))
1374
1375 (defun spam-ifile-register-with-ifile (articles category &optional unregister)
1376   "Register an article, given as a string, with a category.
1377 Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
1378   (let ((category (or category gnus-newsgroup-name))
1379         (add-or-delete-option (if unregister "-d" "-i"))
1380         (db (spam-get-ifile-database-parameter))
1381         parameters)
1382     (with-temp-buffer
1383       (dolist (article articles)
1384         (let ((article-string (spam-get-article-as-string article)))
1385           (when (stringp article-string)
1386             (insert article-string))))
1387       (apply 'call-process-region
1388              (point-min) (point-max) spam-ifile-path
1389              nil nil nil
1390              add-or-delete-option category
1391              (if db `(,db "-h") `("-h"))))))
1392
1393 (defun spam-ifile-register-spam-routine (articles &optional unregister)
1394   (spam-ifile-register-with-ifile articles spam-ifile-spam-category unregister))
1395
1396 (defun spam-ifile-unregister-spam-routine (articles)
1397   (spam-ifile-register-spam-routine articles t))
1398
1399 (defun spam-ifile-register-ham-routine (articles &optional unregister)
1400   (spam-ifile-register-with-ifile articles spam-ifile-ham-category unregister))
1401
1402 (defun spam-ifile-unregister-ham-routine (articles)
1403   (spam-ifile-register-ham-routine articles t))
1404
1405 \f
1406 ;;;; spam-stat
1407
1408 (condition-case nil
1409     (progn
1410       (let ((spam-stat-install-hooks nil))
1411         (require 'spam-stat))
1412
1413       (defun spam-check-stat ()
1414         "Check the spam-stat backend for the classification of this message"
1415         (let ((spam-split-group (if spam-split-symbolic-return
1416                                     'spam
1417                                   spam-split-group))
1418               (spam-stat-split-fancy-spam-group spam-split-group) ; override
1419               (spam-stat-buffer (buffer-name)) ; stat the current buffer
1420               category return)
1421           (spam-stat-split-fancy)))
1422
1423       (defun spam-stat-register-spam-routine (articles &optional unregister)
1424         (dolist (article articles)
1425           (let ((article-string (spam-get-article-as-string article)))
1426             (with-temp-buffer
1427               (insert article-string)
1428               (if unregister
1429                   (spam-stat-buffer-change-to-non-spam)
1430               (spam-stat-buffer-is-spam))))))
1431
1432       (defun spam-stat-unregister-spam-routine (articles)
1433         (spam-stat-register-spam-routine articles t))
1434
1435       (defun spam-stat-register-ham-routine (articles &optional unregister)
1436         (dolist (article articles)
1437           (let ((article-string (spam-get-article-as-string article)))
1438             (with-temp-buffer
1439               (insert article-string)
1440               (if unregister
1441                   (spam-stat-buffer-change-to-spam)
1442               (spam-stat-buffer-is-non-spam))))))
1443
1444       (defun spam-stat-unregister-ham-routine (articles)
1445         (spam-stat-register-ham-routine articles t))
1446
1447       (defun spam-maybe-spam-stat-load ()
1448         (when spam-use-stat (spam-stat-load)))
1449
1450       (defun spam-maybe-spam-stat-save ()
1451         (when spam-use-stat (spam-stat-save))))
1452
1453   (file-error (progn
1454                 (defalias 'spam-stat-load 'ignore)
1455                 (defalias 'spam-stat-save 'ignore)
1456                 (defalias 'spam-maybe-spam-stat-load 'ignore)
1457                 (defalias 'spam-maybe-spam-stat-save 'ignore)
1458                 (defalias 'spam-stat-register-ham-routine 'ignore)
1459                 (defalias 'spam-stat-unregister-ham-routine 'ignore)
1460                 (defalias 'spam-stat-register-spam-routine 'ignore)
1461                 (defalias 'spam-stat-unregister-spam-routine 'ignore)
1462                 (defalias 'spam-stat-buffer-is-spam 'ignore)
1463                 (defalias 'spam-stat-buffer-change-to-spam 'ignore)
1464                 (defalias 'spam-stat-buffer-is-non-spam 'ignore)
1465                 (defalias 'spam-stat-buffer-change-to-non-spam 'ignore)
1466                 (defalias 'spam-stat-split-fancy 'ignore)
1467                 (defalias 'spam-check-stat 'ignore))))
1468
1469 \f
1470
1471 ;;;; Blacklists and whitelists.
1472
1473 (defvar spam-whitelist-cache nil)
1474 (defvar spam-blacklist-cache nil)
1475
1476 (defun spam-kill-whole-line ()
1477   (beginning-of-line)
1478   (let ((kill-whole-line t))
1479     (kill-line)))
1480
1481 ;;; address can be a list, too
1482 (defun spam-enter-whitelist (address &optional remove)
1483   "Enter ADDRESS (list or single) into the whitelist.
1484 With a non-nil REMOVE, remove them."
1485   (interactive "sAddress: ")
1486   (spam-enter-list address spam-whitelist remove)
1487   (setq spam-whitelist-cache nil))
1488
1489 ;;; address can be a list, too
1490 (defun spam-enter-blacklist (address &optional remove)
1491   "Enter ADDRESS (list or single) into the blacklist.
1492 With a non-nil REMOVE, remove them."
1493   (interactive "sAddress: ")
1494   (spam-enter-list address spam-blacklist remove)
1495   (setq spam-blacklist-cache nil))
1496
1497 (defun spam-enter-list (addresses file &optional remove)
1498   "Enter ADDRESSES into the given FILE.
1499 Either the whitelist or the blacklist files can be used.  With
1500 REMOVE not nil, remove the ADDRESSES."
1501   (if (stringp addresses)
1502       (spam-enter-list (list addresses) file remove)
1503     ;; else, we have a list of addresses here
1504     (unless (file-exists-p (file-name-directory file))
1505       (make-directory (file-name-directory file) t))
1506     (save-excursion
1507       (set-buffer
1508        (find-file-noselect file))
1509       (dolist (a addresses)
1510         (when (stringp a)
1511           (goto-char (point-min))
1512           (if (re-search-forward (regexp-quote a) nil t)
1513               ;; found the address
1514               (when remove
1515                 (spam-kill-whole-line))
1516             ;; else, the address was not found
1517             (unless remove
1518               (goto-char (point-max))
1519               (unless (bobp)
1520                 (insert "\n"))
1521               (insert a "\n")))))
1522       (save-buffer))))
1523
1524 ;;; returns t if the sender is in the whitelist, nil or
1525 ;;; spam-split-group otherwise
1526 (defun spam-check-whitelist ()
1527   ;; FIXME!  Should it detect when file timestamps change?
1528   (let ((spam-split-group (if spam-split-symbolic-return
1529                               'spam
1530                             spam-split-group)))
1531     (unless spam-whitelist-cache
1532       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
1533     (if (spam-from-listed-p spam-whitelist-cache)
1534         t
1535       (if spam-use-whitelist-exclusive
1536           spam-split-group
1537         nil))))
1538
1539 (defun spam-check-blacklist ()
1540   ;; FIXME!  Should it detect when file timestamps change?
1541   (let ((spam-split-group (if spam-split-symbolic-return
1542                               'spam
1543                             spam-split-group)))
1544     (unless spam-blacklist-cache
1545       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
1546     (and (spam-from-listed-p spam-blacklist-cache) spam-split-group)))
1547
1548 (defun spam-parse-list (file)
1549   (when (file-readable-p file)
1550     (let (contents address)
1551       (with-temp-buffer
1552         (insert-file-contents file)
1553         (while (not (eobp))
1554           (setq address (buffer-substring (point) (point-at-eol)))
1555           (forward-line 1)
1556           ;; insert the e-mail address if detected, otherwise the raw data
1557           (unless (zerop (length address))
1558             (let ((pure-address (nth 1 (gnus-extract-address-components address))))
1559               (push (or pure-address address) contents)))))
1560       (nreverse contents))))
1561
1562 (defun spam-from-listed-p (cache)
1563   (let ((from (nnmail-fetch-field "from"))
1564         found)
1565     (while cache
1566       (let ((address (pop cache)))
1567         (unless (zerop (length address)) ; 0 for a nil address too
1568           (setq address (regexp-quote address))
1569           ;; fix regexp-quote's treatment of user-intended regexes
1570           (while (string-match "\\\\\\*" address)
1571             (setq address (replace-match ".*" t t address))))
1572         (when (and address (string-match address from))
1573           (setq found t
1574                 cache nil))))
1575     found))
1576
1577 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
1578   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
1579         (declassification (if blacklist 'ham 'spam))
1580         (enter-function
1581          (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
1582         (remove-function
1583          (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
1584         from addresses unregister-list)
1585     (dolist (article articles)
1586       (let ((from (spam-fetch-field-from-fast article))
1587             (id (spam-fetch-field-message-id-fast article))
1588             sender-ignored)
1589         (when (stringp from)
1590           (dolist (ignore-regex spam-blacklist-ignored-regexes)
1591             (when (and (not sender-ignored)
1592                        (stringp ignore-regex)
1593                        (string-match ignore-regex from))
1594               (setq sender-ignored t)))
1595           ;; remember the messages we need to unregister, unless remove is set
1596           (when (and
1597                  (null unregister)
1598                  (spam-log-unregistration-needed-p
1599                   id 'process declassification de-symbol))
1600             (push from unregister-list))
1601           (unless sender-ignored
1602             (push from addresses)))))
1603
1604     (if unregister
1605         (funcall enter-function addresses t) ; unregister all these addresses
1606       ;; else, register normally and unregister what we need to
1607       (funcall remove-function unregister-list t)
1608       (dolist (article unregister-list)
1609         (spam-log-undo-registration
1610          (spam-fetch-field-message-id-fast article)
1611          'process
1612          declassification
1613          de-symbol))
1614       (funcall enter-function addresses nil))))
1615
1616 (defun spam-blacklist-unregister-routine (articles)
1617   (spam-blacklist-register-routine articles t))
1618
1619 (defun spam-blacklist-register-routine (articles &optional unregister)
1620   (spam-filelist-register-routine articles t unregister))
1621
1622 (defun spam-whitelist-unregister-routine (articles)
1623   (spam-whitelist-register-routine articles t))
1624
1625 (defun spam-whitelist-register-routine (articles &optional unregister)
1626   (spam-filelist-register-routine articles nil unregister))
1627
1628 \f
1629 ;;;; Spam-report glue
1630 (defun spam-report-gmane-register-routine (articles)
1631   (when articles
1632     (apply 'spam-report-gmane articles)))
1633
1634 \f
1635 ;;;; Bogofilter
1636 (defun spam-check-bogofilter-headers (&optional score)
1637   (let ((header (nnmail-fetch-field spam-bogofilter-header))
1638         (spam-split-group (if spam-split-symbolic-return
1639                               'spam
1640                             spam-split-group)))
1641     (when header                        ; return nil when no header
1642       (if score                         ; scoring mode
1643           (if (string-match "spamicity=\\([0-9.]+\\)" header)
1644               (match-string 1 header)
1645             "0")
1646         ;; spam detection mode
1647         (when (string-match spam-bogofilter-bogosity-positive-spam-header
1648                             header)
1649           spam-split-group)))))
1650
1651 ;; return something sensible if the score can't be determined
1652 (defun spam-bogofilter-score ()
1653   "Get the Bogofilter spamicity score"
1654   (interactive)
1655   (save-window-excursion
1656     (gnus-summary-show-article t)
1657     (set-buffer gnus-article-buffer)
1658     (let ((score (or (spam-check-bogofilter-headers t)
1659                      (spam-check-bogofilter t))))
1660       (message "Spamicity score %s" score)
1661       (or score "0"))
1662     (gnus-summary-show-article)))
1663
1664 (defun spam-check-bogofilter (&optional score)
1665   "Check the Bogofilter backend for the classification of this message"
1666   (let ((article-buffer-name (buffer-name))
1667         (db spam-bogofilter-database-directory)
1668         return)
1669     (with-temp-buffer
1670       (let ((temp-buffer-name (buffer-name)))
1671         (save-excursion
1672           (set-buffer article-buffer-name)
1673           (apply 'call-process-region
1674                  (point-min) (point-max)
1675                  spam-bogofilter-path
1676                  nil temp-buffer-name nil
1677                  (if db `("-d" ,db "-v") `("-v"))))
1678         (setq return (spam-check-bogofilter-headers score))))
1679     return))
1680
1681 (defun spam-bogofilter-register-with-bogofilter (articles
1682                                                  spam
1683                                                  &optional unregister)
1684   "Register an article, given as a string, as spam or non-spam."
1685   (dolist (article articles)
1686     (let ((article-string (spam-get-article-as-string article))
1687           (db spam-bogofilter-database-directory)
1688           (switch (if unregister
1689                       (if spam
1690                           spam-bogofilter-spam-strong-switch
1691                         spam-bogofilter-ham-strong-switch)
1692                     (if spam
1693                         spam-bogofilter-spam-switch
1694                       spam-bogofilter-ham-switch))))
1695       (when (stringp article-string)
1696         (with-temp-buffer
1697           (insert article-string)
1698
1699           (apply 'call-process-region
1700                  (point-min) (point-max)
1701                  spam-bogofilter-path
1702                  nil nil nil switch
1703                  (if db `("-d" ,db "-v") `("-v"))))))))
1704
1705 (defun spam-bogofilter-register-spam-routine (articles &optional unregister)
1706   (spam-bogofilter-register-with-bogofilter articles t unregister))
1707
1708 (defun spam-bogofilter-unregister-spam-routine (articles)
1709   (spam-bogofilter-register-spam-routine articles t))
1710
1711 (defun spam-bogofilter-register-ham-routine (articles &optional unregister)
1712   (spam-bogofilter-register-with-bogofilter articles nil unregister))
1713
1714 (defun spam-bogofilter-unregister-ham-routine (articles)
1715   (spam-bogofilter-register-ham-routine articles t))
1716
1717
1718 \f
1719 ;;;; spamoracle
1720 (defun spam-check-spamoracle ()
1721   "Run spamoracle on an article to determine whether it's spam."
1722   (let ((article-buffer-name (buffer-name))
1723         (spam-split-group (if spam-split-symbolic-return
1724                               'spam
1725                             spam-split-group)))
1726     (with-temp-buffer
1727       (let ((temp-buffer-name (buffer-name)))
1728         (save-excursion
1729           (set-buffer article-buffer-name)
1730           (let ((status
1731                  (apply 'call-process-region
1732                         (point-min) (point-max)
1733                         spam-spamoracle-binary
1734                         nil temp-buffer-name nil
1735                         (if spam-spamoracle-database
1736                             `("-f" ,spam-spamoracle-database "mark")
1737                           '("mark")))))
1738             (if (eq 0 status)
1739                 (progn
1740                   (set-buffer temp-buffer-name)
1741                   (goto-char (point-min))
1742                   (when (re-search-forward "^X-Spam: yes;" nil t)
1743                     spam-split-group))
1744               (error "Error running spamoracle" status))))))))
1745
1746 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
1747   "Run spamoracle in training mode."
1748   (with-temp-buffer
1749     (let ((temp-buffer-name (buffer-name)))
1750       (save-excursion
1751         (goto-char (point-min))
1752         (dolist (article articles)
1753           (insert (spam-get-article-as-string article)))
1754         (let* ((arg (if (spam-xor unregister article-is-spam-p)
1755                         "-spam"
1756                       "-good"))
1757                (status
1758                 (apply 'call-process-region
1759                        (point-min) (point-max)
1760                        spam-spamoracle-binary
1761                        nil temp-buffer-name nil
1762                        (if spam-spamoracle-database
1763                            `("-f" ,spam-spamoracle-database
1764                              "add" ,arg)
1765                          `("add" ,arg)))))
1766           (when (not (eq 0 status))
1767             (error "Error running spamoracle" status)))))))
1768
1769 (defun spam-spamoracle-learn-ham (articles &optional unregister)
1770   (spam-spamoracle-learn articles nil unregister))
1771
1772 (defun spam-spamoracle-unlearn-ham (articles &optional unregister)
1773   (spam-spamoracle-learn-ham articles t))
1774
1775 (defun spam-spamoracle-learn-spam (articles &optional unregister)
1776   (spam-spamoracle-learn articles t unregister))
1777
1778 (defun spam-spamoracle-unlearn-spam (articles &optional unregister)
1779   (spam-spamoracle-learn-spam articles t))
1780
1781 \f
1782 ;;;; Hooks
1783
1784 ;;;###autoload
1785 (defun spam-initialize ()
1786   "Install the spam.el hooks and do other initialization"
1787   (interactive)
1788   (setq spam-install-hooks t)
1789   ;; TODO: How do we redo this every time spam-face is customized?
1790   (push '((eq mark gnus-spam-mark) . spam-face)
1791         gnus-summary-highlight)
1792   ;; Add hooks for loading and saving the spam stats
1793   (add-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
1794   (add-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
1795   (add-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
1796   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
1797   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
1798   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
1799   (add-hook 'gnus-summary-prepare-hook 'spam-find-spam))
1800
1801 (defun spam-unload-hook ()
1802   "Uninstall the spam.el hooks"
1803   (interactive)
1804   (remove-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
1805   (remove-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
1806   (remove-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
1807   (remove-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
1808   (remove-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
1809   (remove-hook 'gnus-get-new-news-hook 'spam-setup-widening)
1810   (remove-hook 'gnus-summary-prepare-hook 'spam-find-spam))
1811
1812 (when spam-install-hooks
1813   (spam-initialize))
1814
1815 (provide 'spam)
1816
1817 ;;; spam.el ends here.
1818
1819 (provide 'spam)
1820
1821 ;;; spam.el ends here