Import Gnus v5.10.3.
[elisp/gnus.git-] / lisp / spam.el
1 ;; TODO: spam scores, detection of spam in newsgroups, cross-server splitting, remote processing, training through files
2
3 ;;; spam.el --- Identifying spam
4 ;; Copyright (C) 2002, 2003 Free Software Foundation, Inc.
5
6 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
7 ;; Keywords: network
8
9 ;; This file is part of GNU Emacs.
10
11 ;; GNU Emacs is free software; you can redistribute it and/or modify
12 ;; it under the terms of the GNU General Public License as published by
13 ;; the Free Software Foundation; either version 2, or (at your option)
14 ;; any later version.
15
16 ;; GNU Emacs is distributed in the hope that it will be useful,
17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
19 ;; GNU General Public License for more details.
20
21 ;; You should have received a copy of the GNU General Public License
22 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
23 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
24 ;; Boston, MA 02111-1307, USA.
25
26 ;;; Commentary:
27
28 ;;; This module addresses a few aspects of spam control under Gnus.  Page
29 ;;; breaks are used for grouping declarations and documentation relating to
30 ;;; each particular aspect.
31
32 ;;; The integration with Gnus is not yet complete.  See various `FIXME'
33 ;;; comments, below, for supplementary explanations or discussions.
34
35 ;;; Several TODO items are marked as such
36
37 ;;; Code:
38
39 (eval-when-compile (require 'cl))
40
41 (require 'gnus-sum)
42
43 (require 'gnus-uu)                      ; because of key prefix issues
44 ;;; for the definitions of group content classification and spam processors
45 (require 'gnus) 
46 (require 'message)              ;for the message-fetch-field functions
47
48 ;; for nnimap-split-download-body-default
49 (eval-when-compile (require 'nnimap))
50
51 ;; autoload executable-find
52 (eval-and-compile
53   ;; executable-find is not autoloaded in Emacs 20
54   (autoload 'executable-find "executable"))
55
56 ;; autoload query-dig
57 (eval-and-compile
58   (autoload 'query-dig "dig"))
59
60 ;; autoload spam-report
61 (eval-and-compile
62   (autoload 'spam-report-gmane "spam-report"))
63
64 ;; autoload gnus-registry
65 (eval-and-compile
66   (autoload 'gnus-registry-group-count "gnus-registry")
67   (autoload 'gnus-registry-add-group "gnus-registry")
68   (autoload 'gnus-registry-store-extra-entry "gnus-registry")
69   (autoload 'gnus-registry-fetch-extra "gnus-registry"))
70
71 ;; autoload query-dns
72 (eval-and-compile
73   (autoload 'query-dns "dns"))
74
75 ;;; Main parameters.
76
77 (defgroup spam nil
78   "Spam configuration.")
79
80 (defcustom spam-directory "~/News/spam/"
81   "Directory for spam whitelists and blacklists."
82   :type 'directory
83   :group 'spam)
84
85 (defcustom spam-move-spam-nonspam-groups-only t
86   "Whether spam should be moved in non-spam groups only.
87 When t, only ham and unclassified groups will have their spam moved
88 to the spam-process-destination.  When nil, spam will also be moved from
89 spam groups."
90   :type 'boolean
91   :group 'spam)
92
93 (defcustom spam-process-ham-in-nonham-groups nil
94   "Whether ham should be processed in non-ham groups."
95   :type 'boolean
96   :group 'spam)
97
98 (defcustom spam-log-to-registry nil
99   "Whether spam/ham processing should be logged in the registry."
100   :type 'boolean
101   :group 'spam)
102
103 (defcustom spam-split-symbolic-return nil
104   "Whether spam-split should work with symbols or group names."
105   :type 'boolean
106   :group 'spam)
107
108 (defcustom spam-split-symbolic-return-positive nil
109   "Whether spam-split should ALWAYS work with symbols or group
110   names.  Do not set this if you use spam-split in a fancy split
111   method."
112   :type 'boolean
113   :group 'spam)
114
115 (defcustom spam-process-ham-in-spam-groups nil
116   "Whether ham should be processed in spam groups."
117   :type 'boolean
118   :group 'spam)
119
120 (defcustom spam-mark-only-unseen-as-spam t
121   "Whether only unseen articles should be marked as spam in spam
122 groups.  When nil, all unread articles in a spam group are marked as
123 spam.  Set this if you want to leave an article unread in a spam group
124 without losing it to the automatic spam-marking process."
125   :type 'boolean
126   :group 'spam)
127
128 (defcustom spam-mark-ham-unread-before-move-from-spam-group nil
129   "Whether ham should be marked unread before it's moved out of a spam
130 group according to ham-process-destination.  This variable is an
131 official entry in the international Longest Variable Name
132 Competition."
133   :type 'boolean
134   :group 'spam)
135
136 (defcustom spam-disable-spam-split-during-ham-respool nil
137   "Whether spam-split should be ignored while resplitting ham in
138 a process destination.  This is useful to prevent ham from ending
139 up in the same spam group after the resplit.  Don't set this to t
140 if you have spam-split as the last rule in your split
141 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, default to t if one of
270 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.  If nil, the current group name will
349 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) ; logical exclusive or
449   (and (or a b) (not (and a b))))
450
451 (defun spam-group-ham-mark-p (group mark &optional spam)
452   (when (stringp group)
453     (let* ((marks (spam-group-ham-marks group spam))
454            (marks (if (symbolp mark) 
455                       marks 
456                     (mapcar 'symbol-value marks))))
457       (memq mark marks))))
458
459 (defun spam-group-spam-mark-p (group mark)
460   (spam-group-ham-mark-p group mark t))
461
462 (defun spam-group-ham-marks (group &optional spam)
463   (when (stringp group)
464     (let* ((marks (if spam
465                       (gnus-parameter-spam-marks group)
466                     (gnus-parameter-ham-marks group)))
467            (marks (car marks))
468            (marks (if (listp (car marks)) (car marks) marks)))
469       marks)))
470
471 (defun spam-group-spam-marks (group)
472   (spam-group-ham-marks group t))
473
474 (defun spam-group-spam-contents-p (group)
475   (if (stringp group)
476       (or (member group spam-junk-mailgroups)
477           (memq 'gnus-group-spam-classification-spam 
478                 (gnus-parameter-spam-contents group)))
479     nil))
480   
481 (defun spam-group-ham-contents-p (group)
482   (if (stringp group)
483       (memq 'gnus-group-spam-classification-ham 
484             (gnus-parameter-spam-contents group))
485     nil))
486
487 (defvar spam-list-of-processors
488   '((gnus-group-spam-exit-processor-report-gmane spam spam-use-gmane)
489     (gnus-group-spam-exit-processor-bogofilter   spam spam-use-bogofilter)
490     (gnus-group-spam-exit-processor-blacklist    spam spam-use-blacklist)
491     (gnus-group-spam-exit-processor-ifile        spam spam-use-ifile)
492     (gnus-group-spam-exit-processor-stat         spam spam-use-stat)
493     (gnus-group-spam-exit-processor-spamoracle   spam spam-use-spamoracle)
494     (gnus-group-ham-exit-processor-ifile         ham spam-use-ifile)
495     (gnus-group-ham-exit-processor-bogofilter    ham spam-use-bogofilter)
496     (gnus-group-ham-exit-processor-stat          ham spam-use-stat)
497     (gnus-group-ham-exit-processor-whitelist     ham spam-use-whitelist)
498     (gnus-group-ham-exit-processor-BBDB          ham spam-use-BBDB)
499     (gnus-group-ham-exit-processor-copy          ham spam-use-ham-copy)
500     (gnus-group-ham-exit-processor-spamoracle    ham spam-use-spamoracle))
501   "The spam-list-of-processors list contains pairs associating a
502 ham/spam exit processor variable with a classification and a
503 spam-use-* variable.")
504
505 (defun spam-group-processor-p (group processor)
506   (if (and (stringp group)
507            (symbolp processor))
508       (or (member processor (nth 0 (gnus-parameter-spam-process group)))
509           (spam-group-processor-multiple-p 
510            group 
511            (cdr-safe (assoc processor spam-list-of-processors))))
512     nil))
513
514 (defun spam-group-processor-multiple-p (group processor-info)
515   (let* ((classification (nth 0 processor-info))
516          (check (nth 1 processor-info))
517          (parameters (nth 0 (gnus-parameter-spam-process group)))
518          found)
519     (dolist (parameter parameters)
520       (when (and (null found)
521                  (listp parameter)
522                  (eq classification (nth 0 parameter))
523                  (eq check (nth 1 parameter)))
524         (setq found t)))
525     found))
526
527 (defun spam-group-spam-processor-report-gmane-p (group)
528   (spam-group-processor-p group 'gnus-group-spam-exit-processor-report-gmane))
529
530 (defun spam-group-spam-processor-bogofilter-p (group)
531   (spam-group-processor-p group 'gnus-group-spam-exit-processor-bogofilter))
532
533 (defun spam-group-spam-processor-blacklist-p (group)
534   (spam-group-processor-p group 'gnus-group-spam-exit-processor-blacklist))
535
536 (defun spam-group-spam-processor-ifile-p (group)
537   (spam-group-processor-p group 'gnus-group-spam-exit-processor-ifile))
538
539 (defun spam-group-ham-processor-ifile-p (group)
540   (spam-group-processor-p group 'gnus-group-ham-exit-processor-ifile))
541
542 (defun spam-group-spam-processor-spamoracle-p (group)
543   (spam-group-processor-p group 'gnus-group-spam-exit-processor-spamoracle))
544
545 (defun spam-group-ham-processor-bogofilter-p (group)
546   (spam-group-processor-p group 'gnus-group-ham-exit-processor-bogofilter))
547
548 (defun spam-group-spam-processor-stat-p (group)
549   (spam-group-processor-p group 'gnus-group-spam-exit-processor-stat))
550
551 (defun spam-group-ham-processor-stat-p (group)
552   (spam-group-processor-p group 'gnus-group-ham-exit-processor-stat))
553
554 (defun spam-group-ham-processor-whitelist-p (group)
555   (spam-group-processor-p group 'gnus-group-ham-exit-processor-whitelist))
556
557 (defun spam-group-ham-processor-BBDB-p (group)
558   (spam-group-processor-p group 'gnus-group-ham-exit-processor-BBDB))
559
560 (defun spam-group-ham-processor-copy-p (group)
561   (spam-group-processor-p group 'gnus-group-ham-exit-processor-copy))
562
563 (defun spam-group-ham-processor-spamoracle-p (group)
564   (spam-group-processor-p group 'gnus-group-ham-exit-processor-spamoracle))
565
566 ;;; Summary entry and exit processing.
567
568 (defun spam-summary-prepare ()
569   (setq spam-old-ham-articles 
570         (spam-list-articles gnus-newsgroup-articles 'ham))
571   (setq spam-old-spam-articles 
572         (spam-list-articles gnus-newsgroup-articles 'spam))
573   (spam-mark-junk-as-spam-routine))
574
575 ;; The spam processors are invoked for any group, spam or ham or neither
576 (defun spam-summary-prepare-exit ()
577   (unless gnus-group-is-exiting-without-update-p
578     (gnus-message 6 "Exiting summary buffer and applying spam rules")
579
580     ;; first of all, unregister any articles that are no longer ham or spam
581     ;; we have to iterate over the processors, or else we'll be too slow
582     (dolist (classification '(spam ham))
583       (let* ((old-articles (if (eq classification 'spam)
584                                spam-old-spam-articles 
585                              spam-old-ham-articles))
586              (new-articles (spam-list-articles 
587                             gnus-newsgroup-articles 
588                             classification))
589              (changed-articles (gnus-set-difference old-articles new-articles)))
590         ;; now that we have the changed articles, we go through the processors
591         (dolist (processor-param spam-list-of-processors)
592           (let ((processor (nth 0 processor-param))
593                 (processor-classification (nth 1 processor-param))
594                 (check (nth 2 processor-param))
595                 unregister-list)
596             (dolist (article changed-articles)
597               (let ((id (spam-fetch-field-message-id-fast article)))
598                 (when (spam-log-unregistration-needed-p 
599                        id 'process classification check)
600                   (push article unregister-list))))
601             ;; call spam-register-routine with specific articles to unregister,
602             ;; when there are articles to unregister and the check is enabled
603             (when (and unregister-list (symbol-value check))
604               (spam-register-routine classification check t unregister-list))))))
605       
606     ;; find all the spam processors applicable to this group
607     (dolist (processor-param spam-list-of-processors)
608       (let ((processor (nth 0 processor-param))
609             (classification (nth 1 processor-param))
610             (check (nth 2 processor-param)))
611         (when (and (eq 'spam classification)
612                    (spam-group-processor-p gnus-newsgroup-name processor))
613           (spam-register-routine classification check))))
614
615     (if spam-move-spam-nonspam-groups-only      
616         (when (not (spam-group-spam-contents-p gnus-newsgroup-name))
617           (spam-mark-spam-as-expired-and-move-routine
618            (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
619       (gnus-message 5 "Marking spam as expired and moving it to %s" 
620                     gnus-newsgroup-name)
621       (spam-mark-spam-as-expired-and-move-routine 
622        (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
623
624     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
625     ;; expire spam, in case the above did not expire them
626     (gnus-message 5 "Marking spam as expired without moving it")
627     (spam-mark-spam-as-expired-and-move-routine nil)
628
629     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
630               (and (spam-group-spam-contents-p gnus-newsgroup-name)
631                    spam-process-ham-in-spam-groups)
632               spam-process-ham-in-nonham-groups)
633       ;; find all the ham processors applicable to this group
634       (dolist (processor-param spam-list-of-processors)
635         (let ((processor (nth 0 processor-param))
636               (classification (nth 1 processor-param))
637               (check (nth 2 processor-param)))
638           (when (and (eq 'ham classification)
639                      (spam-group-processor-p gnus-newsgroup-name processor))
640             (spam-register-routine classification check)))))
641
642     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
643       (gnus-message 5 "Copying ham")
644       (spam-ham-copy-routine
645        (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
646
647     ;; now move all ham articles out of spam groups
648     (when (spam-group-spam-contents-p gnus-newsgroup-name)
649       (gnus-message 5 "Moving ham messages from spam group")
650       (spam-ham-move-routine
651        (gnus-parameter-ham-process-destination gnus-newsgroup-name))))
652
653   (setq spam-old-ham-articles nil)
654   (setq spam-old-spam-articles nil))
655
656 (defun spam-mark-junk-as-spam-routine ()
657   ;; check the global list of group names spam-junk-mailgroups and the
658   ;; group parameters
659   (when (spam-group-spam-contents-p gnus-newsgroup-name)
660     (gnus-message 5 "Marking %s articles as spam"
661                   (if spam-mark-only-unseen-as-spam 
662                       "unseen"
663                     "unread"))
664     (let ((articles (if spam-mark-only-unseen-as-spam 
665                         gnus-newsgroup-unseen
666                       gnus-newsgroup-unreads)))
667       (dolist (article articles)
668         (gnus-summary-mark-article article gnus-spam-mark)))))
669
670 (defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
671   (if (and (car-safe groups) (listp (car-safe groups)))
672       (apply 'spam-mark-spam-as-expired-and-move-routine (car groups))
673     (gnus-summary-kill-process-mark)
674     (let ((articles gnus-newsgroup-articles)
675           (backend-supports-deletions
676            (gnus-check-backend-function
677             'request-move-article gnus-newsgroup-name))
678           article tomove deletep)
679       (dolist (article articles)
680         (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
681           (gnus-summary-mark-article article gnus-expirable-mark)
682           (push article tomove)))
683     
684       ;; now do the actual copies
685       (dolist (group groups)
686         (when (and tomove
687                    (stringp group))
688           (dolist (article tomove)
689             (gnus-summary-set-process-mark article))
690           (when tomove
691             (if (or (not backend-supports-deletions)
692                     (> (length groups) 1))
693                 (progn 
694                   (gnus-summary-copy-article nil group)
695                   (setq deletep t))
696               (gnus-summary-move-article nil group)))))
697     
698       ;; now delete the articles, if there was a copy done, and the
699       ;; backend allows it
700       (when (and deletep backend-supports-deletions)
701         (dolist (article tomove)
702           (gnus-summary-set-process-mark article))
703         (when tomove
704           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
705             (gnus-summary-delete-article nil))))
706     
707       (gnus-summary-yank-process-mark))))
708  
709 (defun spam-ham-copy-or-move-routine (copy groups)
710   (gnus-summary-kill-process-mark)
711   (let ((articles gnus-newsgroup-articles)
712         (backend-supports-deletions
713          (gnus-check-backend-function
714           'request-move-article gnus-newsgroup-name))
715         (respool-method (gnus-find-method-for-group gnus-newsgroup-name))
716         article mark todo deletep respool)
717     (dolist (article articles)
718       (when (spam-group-ham-mark-p gnus-newsgroup-name
719                                    (gnus-summary-article-mark article))
720         (push article todo)))
721
722     (when (member 'respool groups)
723       (setq respool t)                  ; boolean for later
724       (setq groups '("fake"))) ; when respooling, groups are dynamic so fake it
725
726     ;; now do the actual move
727     (dolist (group groups)
728       (when (and todo (stringp group))
729         (dolist (article todo)
730           (when spam-mark-ham-unread-before-move-from-spam-group
731             (gnus-summary-mark-article article gnus-unread-mark))
732           (gnus-summary-set-process-mark article))
733
734         (if respool                        ; respooling is with a "fake" group
735             (let ((spam-split-disabled
736                    (or spam-split-disabled
737                        spam-disable-spam-split-during-ham-respool)))
738               (gnus-summary-respool-article nil respool-method))
739           (if (or (not backend-supports-deletions) ; else, we are not respooling
740                   (> (length groups) 1))
741               (progn                ; if copying, copy and set deletep
742                 (gnus-summary-copy-article nil group)
743                 (setq deletep t))
744             (gnus-summary-move-article nil group))))) ; else move articles
745     
746     ;; now delete the articles, unless a) copy is t, and there was a copy done
747     ;;                                 b) a move was done to a single group
748     ;;                                 c) backend-supports-deletions is nil
749     (unless copy
750       (when (and deletep backend-supports-deletions)
751         (dolist (article todo)
752           (gnus-summary-set-process-mark article))
753         (when todo
754           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
755             (gnus-summary-delete-article nil))))))
756   
757   (gnus-summary-yank-process-mark))
758  
759 (defun spam-ham-copy-routine (&rest groups)
760   (if (and (car-safe groups) (listp (car-safe groups)))
761       (apply 'spam-ham-copy-routine (car groups))
762     (spam-ham-copy-or-move-routine t groups)))
763  
764 (defun spam-ham-move-routine (&rest groups)
765   (if (and (car-safe groups) (listp (car-safe groups)))
766       (apply 'spam-ham-move-routine (car groups))
767     (spam-ham-copy-or-move-routine nil groups)))
768  
769 (eval-and-compile
770   (defalias 'spam-point-at-eol (if (fboundp 'point-at-eol)
771                                    'point-at-eol
772                                  'line-end-position)))
773
774 (defun spam-get-article-as-string (article)
775   (let ((article-buffer (spam-get-article-as-buffer article))
776         article-string)
777     (when article-buffer
778       (save-window-excursion
779         (set-buffer article-buffer)
780         (setq article-string (buffer-string))))
781     article-string))
782
783 (defun spam-get-article-as-buffer (article)
784   (let ((article-buffer))
785     (when (numberp article)
786       (save-window-excursion
787         (gnus-summary-goto-subject article)
788         (gnus-summary-show-article t)
789         (setq article-buffer (get-buffer gnus-article-buffer))))
790     article-buffer))
791
792 ;; disabled for now
793 ;; (defun spam-get-article-as-filename (article)
794 ;;   (let ((article-filename))
795 ;;     (when (numberp article)
796 ;;       (nnml-possibly-change-directory 
797 ;;        (gnus-group-real-name gnus-newsgroup-name))
798 ;;       (setq article-filename (expand-file-name 
799 ;;                              (int-to-string article) nnml-current-directory)))
800 ;;     (if (file-exists-p article-filename)
801 ;;      article-filename
802 ;;       nil)))
803
804 (defun spam-fetch-field-from-fast (article)
805   "Fetch the `from' field quickly, using the internal gnus-data-list function"
806   (if (and (numberp article)
807            (assoc article (gnus-data-list nil)))
808       (mail-header-from 
809        (gnus-data-header (assoc article (gnus-data-list nil))))
810     nil))
811
812 (defun spam-fetch-field-subject-fast (article)
813   "Fetch the `subject' field quickly, using the internal
814   gnus-data-list function"
815   (if (and (numberp article)
816            (assoc article (gnus-data-list nil)))
817       (mail-header-subject 
818        (gnus-data-header (assoc article (gnus-data-list nil))))
819     nil))
820
821 (defun spam-fetch-field-message-id-fast (article)
822   "Fetch the `Message-ID' field quickly, using the internal
823   gnus-data-list function"
824   (if (and (numberp article)
825            (assoc article (gnus-data-list nil)))
826       (mail-header-message-id 
827        (gnus-data-header (assoc article (gnus-data-list nil))))
828     nil))
829
830 \f
831 ;;;; Spam determination.
832
833 (defvar spam-list-of-checks
834   '((spam-use-blacklist          . spam-check-blacklist)
835     (spam-use-regex-headers      . spam-check-regex-headers)
836     (spam-use-regex-body         . spam-check-regex-body)
837     (spam-use-whitelist          . spam-check-whitelist)
838     (spam-use-BBDB               . spam-check-BBDB)
839     (spam-use-ifile              . spam-check-ifile)
840     (spam-use-spamoracle         . spam-check-spamoracle)
841     (spam-use-stat               . spam-check-stat)
842     (spam-use-blackholes         . spam-check-blackholes)
843     (spam-use-hashcash           . spam-check-hashcash)
844     (spam-use-bogofilter-headers . spam-check-bogofilter-headers)
845     (spam-use-bogofilter         . spam-check-bogofilter))
846   "The spam-list-of-checks list contains pairs associating a
847 parameter variable with a spam checking function.  If the
848 parameter variable is true, then the checking function is called,
849 and its value decides what happens.  Each individual check may
850 return nil, t, or a mailgroup name.  The value nil means that the
851 check does not yield a decision, and so, that further checks are
852 needed.  The value t means that the message is definitely not
853 spam, and that further spam checks should be inhibited.
854 Otherwise, a mailgroup name or the symbol 'spam (depending on
855 spam-split-symbolic-return) is returned where the mail should go,
856 and further checks are also inhibited.  The usual mailgroup name
857 is the value of `spam-split-group', meaning that the message is
858 definitely a spam.")
859
860 (defvar spam-list-of-statistical-checks 
861   '(spam-use-ifile
862     spam-use-regex-body 
863     spam-use-stat 
864     spam-use-bogofilter
865     spam-use-spamoracle)
866   "The spam-list-of-statistical-checks list contains all the mail
867 splitters that need to have the full message body available.")
868
869 ;;;TODO: modify to invoke self with each check if invoked without specifics
870 (defun spam-split (&rest specific-checks)
871   "Split this message into the `spam' group if it is spam.
872 This function can be used as an entry in `nnmail-split-fancy',
873 for example like this: (: spam-split).  It can take checks as
874 parameters.  A string as a parameter will set the
875 spam-split-group to that string.
876
877 See the Info node `(gnus)Fancy Mail Splitting' for more details."
878   (interactive)
879   (setq spam-split-last-successful-check nil)
880   (unless spam-split-disabled
881     (let ((spam-split-group-choice spam-split-group))
882       (dolist (check specific-checks)
883         (when (stringp check)
884           (setq spam-split-group-choice check)
885           (setq specific-checks (delq check specific-checks))))
886       
887       (let ((spam-split-group spam-split-group-choice))
888         (save-excursion
889           (save-restriction
890             (dolist (check spam-list-of-statistical-checks)
891               (when (and (symbolp check) (symbol-value check))
892                 (widen)
893                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
894                               (symbol-name check))
895                 (return)))
896             ;;   (progn (widen) (debug (buffer-string)))
897             (let ((list-of-checks spam-list-of-checks)
898                   decision)
899               (while (and list-of-checks (not decision))
900                 (let ((pair (pop list-of-checks)))
901                   (when (and (symbol-value (car pair))
902                              (or (null specific-checks)
903                                  (memq (car pair) specific-checks)))
904                     (gnus-message 5 "spam-split: calling the %s function" 
905                                   (symbol-name (cdr pair)))
906                     (setq decision (funcall (cdr pair)))
907                     ;; if we got a decision at all, save the current check
908                     (when decision
909                       (setq spam-split-last-successful-check (car pair)))
910
911                     (when (eq decision 'spam)
912                       (if spam-split-symbolic-return
913                           (setq decision spam-split-group)
914                         (gnus-error
915                          5 
916                          (format "spam-split got %s but %s is nil"
917                                  (symbol-name decision)
918                                  (symbol-name spam-split-symbolic-return))))))))
919               (if (eq decision t)
920                   (if spam-split-symbolic-return-positive 'ham nil)
921                 decision))))))))
922
923 (defun spam-find-spam ()
924   "This function will detect spam in the current newsgroup using spam-split"
925   (interactive)
926   
927   (let* ((group gnus-newsgroup-name)
928          (autodetect (gnus-parameter-spam-autodetect group))
929          (methods (gnus-parameter-spam-autodetect-methods group))
930          (first-method (nth 0 methods)))
931   (when (and autodetect 
932              (not (equal first-method 'none)))
933     (mapcar
934      (lambda (article)
935        (let ((id (spam-fetch-field-message-id-fast article))
936              (subject (spam-fetch-field-subject-fast article))
937              (sender (spam-fetch-field-from-fast article)))
938          (unless (and spam-log-to-registry
939                       (spam-log-registered-p id 'incoming))
940            (let* ((spam-split-symbolic-return t)
941                   (spam-split-symbolic-return-positive t)
942                   (split-return
943                    (with-temp-buffer
944                      (gnus-request-article-this-buffer 
945                       article 
946                       group)
947                      (if (or (null first-method)
948                              (equal first-method 'default))
949                          (spam-split)
950                        (apply 'spam-split methods)))))
951              (if (equal split-return 'spam)
952                  (gnus-summary-mark-article article gnus-spam-mark))
953
954              (when (and split-return spam-log-to-registry)
955                (when (zerop (gnus-registry-group-count id))
956                  (gnus-registry-add-group
957                   id group subject sender))
958
959                (spam-log-processing-to-registry 
960                 id
961                 'incoming
962                 split-return
963                 spam-split-last-successful-check
964                 group))))))
965      (if spam-autodetect-recheck-messages
966          gnus-newsgroup-articles
967        gnus-newsgroup-unseen)))))
968
969 (defvar spam-registration-functions
970   ;; first the ham register, second the spam register function
971   ;; third the ham unregister, fourth the spam unregister function
972   '((spam-use-blacklist  nil 
973                          spam-blacklist-register-routine
974                          nil
975                          spam-blacklist-unregister-routine)
976     (spam-use-whitelist  spam-whitelist-register-routine
977                          nil
978                          spam-whitelist-unregister-routine
979                          nil)
980     (spam-use-BBDB       spam-BBDB-register-routine 
981                          nil
982                          spam-BBDB-unregister-routine 
983                          nil)
984     (spam-use-ifile      spam-ifile-register-ham-routine 
985                          spam-ifile-register-spam-routine
986                          spam-ifile-unregister-ham-routine 
987                          spam-ifile-unregister-spam-routine)
988     (spam-use-spamoracle spam-spamoracle-learn-ham 
989                          spam-spamoracle-learn-spam
990                          spam-spamoracle-unlearn-ham 
991                          spam-spamoracle-unlearn-spam)
992     (spam-use-stat       spam-stat-register-ham-routine 
993                          spam-stat-register-spam-routine
994                          spam-stat-unregister-ham-routine 
995                          spam-stat-unregister-spam-routine)
996     ;; note that spam-use-gmane is not a legitimate check
997     (spam-use-gmane      nil 
998                          spam-report-gmane-register-routine
999                          ;; does Gmane support unregistration?
1000                          nil
1001                          nil)
1002     (spam-use-bogofilter spam-bogofilter-register-ham-routine 
1003                          spam-bogofilter-register-spam-routine
1004                          spam-bogofilter-unregister-ham-routine 
1005                          spam-bogofilter-unregister-spam-routine))
1006   "The spam-registration-functions list contains pairs
1007 associating a parameter variable with the ham and spam
1008 registration functions, and the ham and spam unregistration
1009 functions")
1010
1011 (defun spam-classification-valid-p (classification)
1012   (or  (eq classification 'spam)
1013        (eq classification 'ham)))
1014
1015 (defun spam-process-type-valid-p (process-type)
1016   (or  (eq process-type 'incoming)
1017        (eq process-type 'process)))
1018
1019 (defun spam-registration-check-valid-p (check)
1020   (assoc check spam-registration-functions))
1021
1022 (defun spam-unregistration-check-valid-p (check)
1023   (assoc check spam-registration-functions))
1024
1025 (defun spam-registration-function (classification check)
1026   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1027     (if (eq classification 'spam)
1028         (nth 1 flist)
1029       (nth 0 flist))))
1030
1031 (defun spam-unregistration-function (classification check)
1032   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1033     (if (eq classification 'spam)
1034         (nth 3 flist)
1035       (nth 2 flist))))
1036
1037 (defun spam-list-articles (articles classification)
1038   (let ((mark-check (if (eq classification 'spam) 
1039                         'spam-group-spam-mark-p 
1040                       'spam-group-ham-mark-p))
1041         mark list)
1042     (dolist (article articles)
1043       (when (funcall mark-check 
1044                      gnus-newsgroup-name 
1045                      (gnus-summary-article-mark article))
1046         (push article list)))
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) (spam-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.  With a
1484   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.  With a
1492   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) (spam-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.