(spam-bsfilter-path): Use `executable-find' instead of
[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 query-dig
53 (eval-and-compile
54   (autoload 'query-dig "dig"))
55
56 ;; autoload spam-report
57 (eval-and-compile
58   (autoload 'spam-report-gmane "spam-report"))
59
60 ;; autoload gnus-registry
61 (eval-and-compile
62   (autoload 'gnus-registry-group-count "gnus-registry")
63   (autoload 'gnus-registry-add-group "gnus-registry")
64   (autoload 'gnus-registry-store-extra-entry "gnus-registry")
65   (autoload 'gnus-registry-fetch-extra "gnus-registry"))
66
67 ;; autoload query-dns
68 (eval-and-compile
69   (autoload 'query-dns "dns"))
70
71 ;;; Main parameters.
72
73 (defgroup spam nil
74   "Spam configuration.")
75
76 (defcustom spam-directory (nnheader-concat gnus-directory "spam/")
77   "Directory for spam whitelists and blacklists."
78   :type 'directory
79   :group 'spam)
80
81 (defcustom spam-move-spam-nonspam-groups-only t
82   "Whether spam should be moved in non-spam groups only.
83 When t, only ham and unclassified groups will have their spam moved
84 to the spam-process-destination.  When nil, spam will also be moved from
85 spam groups."
86   :type 'boolean
87   :group 'spam)
88
89 (defcustom spam-process-ham-in-nonham-groups nil
90   "Whether ham should be processed in non-ham groups."
91   :type 'boolean
92   :group 'spam)
93
94 (defcustom spam-log-to-registry nil
95   "Whether spam/ham processing should be logged in the registry."
96   :type 'boolean
97   :group 'spam)
98
99 (defcustom spam-split-symbolic-return nil
100   "Whether `spam-split' should work with symbols or group names."
101   :type 'boolean
102   :group 'spam)
103
104 (defcustom spam-split-symbolic-return-positive nil
105   "Whether `spam-split' should ALWAYS work with symbols or group names.
106 Do not set this if you use `spam-split' in a fancy split
107   method."
108   :type 'boolean
109   :group 'spam)
110
111 (defcustom spam-process-ham-in-spam-groups nil
112   "Whether ham should be processed in spam groups."
113   :type 'boolean
114   :group 'spam)
115
116 (defcustom spam-mark-only-unseen-as-spam t
117   "Whether only unseen articles should be marked as spam in spam groups.
118 When nil, all unread articles in a spam group are marked as
119 spam.  Set this if you want to leave an article unread in a spam group
120 without losing it to the automatic spam-marking process."
121   :type 'boolean
122   :group 'spam)
123
124 (defcustom spam-mark-ham-unread-before-move-from-spam-group nil
125   "Whether ham should be marked unread before it's moved.
126 The article is moved out of a spam group according to ham-process-destination.
127 This variable is an official entry in the international Longest Variable Name
128 Competition."
129   :type 'boolean
130   :group 'spam)
131
132 (defcustom spam-disable-spam-split-during-ham-respool nil
133   "Whether `spam-split' should be ignored while resplitting ham.
134 This is useful to prevent ham from ending up in the same spam
135 group after the resplit.  Don't set this to t if you have `spam-split' as the
136 last rule in your split configuration."
137   :type 'boolean
138   :group 'spam)
139
140 (defcustom spam-autodetect-recheck-messages nil
141   "Should spam.el recheck all meessages when autodetecting?
142 Normally this is nil, so only unseen messages will be checked."
143   :type 'boolean
144   :group 'spam)
145
146 (defcustom spam-whitelist (expand-file-name "whitelist" spam-directory)
147   "The location of the whitelist.
148 The file format is one regular expression per line.
149 The regular expression is matched against the address."
150   :type 'file
151   :group 'spam)
152
153 (defcustom spam-blacklist (expand-file-name "blacklist" spam-directory)
154   "The location of the blacklist.
155 The file format is one regular expression per line.
156 The regular expression is matched against the address."
157   :type 'file
158   :group 'spam)
159
160 (defcustom spam-use-dig t
161   "Whether `query-dig' should be used instead of `query-dns'."
162   :type 'boolean
163   :group 'spam)
164
165 (defcustom spam-use-gmane-xref nil
166   "Whether the Gmane spam xref should be used by `spam-split'."
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-bsfilter-headers nil
227   "Whether bsfilter headers should be used by `spam-split'.
228 Enable this if you pre-process messages with Bsfilter BEFORE Gnus sees them."
229   :type 'boolean
230   :group 'spam)
231
232 (defcustom spam-use-bsfilter nil
233   "Whether bsfilter should be invoked by `spam-split'.
234 Enable this if you want Gnus to invoke Bsfilter on new messages."
235   :type 'boolean
236   :group 'spam)
237
238 (defcustom spam-use-BBDB nil
239   "Whether BBDB should be used by `spam-split'."
240   :type 'boolean
241   :group 'spam)
242
243 (defcustom spam-use-BBDB-exclusive nil
244   "Whether BBDB-exclusive should be used by `spam-split'.
245 Exclusive BBDB means that all messages from senders not in the BBDB are
246 considered spam."
247   :type 'boolean
248   :group 'spam)
249
250 (defcustom spam-use-ifile nil
251   "Whether ifile should be used by `spam-split'."
252   :type 'boolean
253   :group 'spam)
254
255 (defcustom spam-use-stat nil
256   "Whether `spam-stat' should be used by `spam-split'."
257   :type 'boolean
258   :group 'spam)
259
260 (defcustom spam-use-spamoracle nil
261   "Whether spamoracle should be used by `spam-split'."
262   :type 'boolean
263   :group 'spam)
264
265 (defcustom spam-use-spamassassin nil
266   "Whether spamassassin should be invoked by `spam-split'.
267 Enable this if you want Gnus to invoke SpamAssassin on new messages."
268   :type 'boolean
269   :group 'spam)
270
271 (defcustom spam-use-spamassassin-headers nil
272   "Whether spamassassin headers should be checked by `spam-split'.
273 Enable this if you pre-process messages with SpamAssassin BEFORE Gnus sees
274 them."
275   :type 'boolean
276   :group 'spam)
277
278 (defcustom spam-install-hooks (or
279                                spam-use-dig
280                                spam-use-gmane-xref
281                                spam-use-blacklist
282                                spam-use-whitelist
283                                spam-use-whitelist-exclusive
284                                spam-use-blackholes
285                                spam-use-hashcash
286                                spam-use-regex-headers
287                                spam-use-regex-body
288                                spam-use-bogofilter
289                                spam-use-bogofilter-headers
290                                spam-use-spamassassin
291                                spam-use-spamassassin-headers
292                                spam-use-bsfilter
293                                spam-use-bsfilter-headers
294                                spam-use-BBDB
295                                spam-use-BBDB-exclusive
296                                spam-use-ifile
297                                spam-use-stat
298                                spam-use-spamoracle)
299   "Whether the spam hooks should be installed.
300 Default to t if one of the spam-use-* variables is set."
301   :group 'spam
302   :type 'boolean)
303
304 (defcustom spam-split-group "spam"
305   "Group name where incoming spam should be put by `spam-split'."
306   :type 'string
307   :group 'spam)
308
309 ;;; TODO: deprecate this variable, it's confusing since it's a list of strings,
310 ;;; not regular expressions
311 (defcustom spam-junk-mailgroups (cons
312                                  spam-split-group
313                                  '("mail.junk" "poste.pourriel"))
314   "Mailgroups with spam contents.
315 All unmarked article in such group receive the spam mark on group entry."
316   :type '(repeat (string :tag "Group"))
317   :group 'spam)
318
319
320 (defcustom spam-gmane-xref-spam-group "gmane.spam.detected"
321   "The group where spam xrefs can be found on Gmane.
322 Only meaningful if you enable `spam-use-gmane-xref'."
323   :type 'string
324   :group 'spam)
325
326 (defcustom spam-blackhole-servers '("bl.spamcop.net" "relays.ordb.org"
327                                     "dev.null.dk" "relays.visi.com")
328   "List of blackhole servers.
329 Only meaningful if you enable `spam-use-blackholes'."
330   :type '(repeat (string :tag "Server"))
331   :group 'spam)
332
333 (defcustom spam-blackhole-good-server-regex nil
334   "String matching IP addresses that should not be checked in the blackholes.
335 Only meaningful if you enable `spam-use-blackholes'."
336   :type '(radio (const nil)
337                 (regexp :format "%t: %v\n" :size 0))
338   :group 'spam)
339
340 (defcustom spam-face 'gnus-splash-face
341   "Face for spam-marked articles."
342   :type 'face
343   :group 'spam)
344
345 (defcustom spam-regex-headers-spam '("^X-Spam-Flag: YES")
346   "Regular expression for positive header spam matches.
347 Only meaningful if you enable `spam-use-regex-headers'."
348   :type '(repeat (regexp :tag "Regular expression to match spam header"))
349   :group 'spam)
350
351 (defcustom spam-regex-headers-ham '("^X-Spam-Flag: NO")
352   "Regular expression for positive header ham matches.
353 Only meaningful if you enable `spam-use-regex-headers'."
354   :type '(repeat (regexp :tag "Regular expression to match ham header"))
355   :group 'spam)
356
357 (defcustom spam-regex-body-spam '()
358   "Regular expression for positive body spam matches.
359 Only meaningful if you enable `spam-use-regex-body'."
360   :type '(repeat (regexp :tag "Regular expression to match spam body"))
361   :group 'spam)
362
363 (defcustom spam-regex-body-ham '()
364   "Regular expression for positive body ham matches.
365 Only meaningful if you enable `spam-use-regex-body'."
366   :type '(repeat (regexp :tag "Regular expression to match ham body"))
367   :group 'spam)
368
369 (defgroup spam-ifile nil
370   "Spam ifile configuration."
371   :group 'spam)
372
373 (defcustom spam-ifile-path (executable-find "ifile")
374   "File path of the ifile executable program."
375   :type '(choice (file :tag "Location of ifile")
376                  (const :tag "ifile is not installed"))
377   :group 'spam-ifile)
378
379 (defcustom spam-ifile-database-path nil
380   "File path of the ifile database."
381   :type '(choice (file :tag "Location of the ifile database")
382                  (const :tag "Use the default"))
383   :group 'spam-ifile)
384
385 (defcustom spam-ifile-spam-category "spam"
386   "Name of the spam ifile category."
387   :type 'string
388   :group 'spam-ifile)
389
390 (defcustom spam-ifile-ham-category nil
391   "Name of the ham ifile category.
392 If nil, the current group name will be used."
393   :type '(choice (string :tag "Use a fixed category")
394                  (const :tag "Use the current group name"))
395   :group 'spam-ifile)
396
397 (defcustom spam-ifile-all-categories nil
398   "Whether the ifile check will return all categories, or just spam.
399 Set this to t if you want to use the `spam-split' invocation of ifile as
400 your main source of newsgroup names."
401   :type 'boolean
402   :group 'spam-ifile)
403
404 (defgroup spam-bogofilter nil
405   "Spam bogofilter configuration."
406   :group 'spam)
407
408 (defcustom spam-bogofilter-path (executable-find "bogofilter")
409   "File path of the Bogofilter executable program."
410   :type '(choice (file :tag "Location of bogofilter")
411                  (const :tag "Bogofilter is not installed"))
412   :group 'spam-bogofilter)
413
414 (defcustom spam-bogofilter-header "X-Bogosity"
415   "The header that Bogofilter inserts in messages."
416   :type 'string
417   :group 'spam-bogofilter)
418
419 (defcustom spam-bogofilter-spam-switch "-s"
420   "The switch that Bogofilter uses to register spam messages."
421   :type 'string
422   :group 'spam-bogofilter)
423
424 (defcustom spam-bogofilter-ham-switch "-n"
425   "The switch that Bogofilter uses to register ham messages."
426   :type 'string
427   :group 'spam-bogofilter)
428
429 (defcustom spam-bogofilter-spam-strong-switch "-S"
430   "The switch that Bogofilter uses to unregister ham messages."
431   :type 'string
432   :group 'spam-bogofilter)
433
434 (defcustom spam-bogofilter-ham-strong-switch "-N"
435   "The switch that Bogofilter uses to unregister spam messages."
436   :type 'string
437   :group 'spam-bogofilter)
438
439 (defcustom spam-bogofilter-bogosity-positive-spam-header "^\\(Yes\\|Spam\\)"
440   "The regex on `spam-bogofilter-header' for positive spam identification."
441   :type 'regexp
442   :group 'spam-bogofilter)
443
444 (defcustom spam-bogofilter-database-directory nil
445   "Directory path of the Bogofilter databases."
446   :type '(choice (directory
447                   :tag "Location of the Bogofilter database directory")
448                  (const :tag "Use the default"))
449   :group 'spam-bogofilter)
450
451 (defgroup spam-bsfilter nil
452   "Spam bsfilter configuration."
453   :group 'spam)
454
455 (defcustom spam-bsfilter-path (executable-find "bsfilter")
456   "File path of the Bsfilter executable program."
457   :type '(choice (file :tag "Location of bsfilter")
458                  (const :tag "Bsfilter is not installed"))
459   :group 'spam-bsfilter)
460
461 (defcustom spam-bsfilter-header "X-Spam-Flag"
462   "The header inserted by Bsfilter to flag spam."
463   :type 'string
464   :group 'spam-bsfilter)
465
466 (defcustom spam-bsfilter-probability-header "X-Spam-Probability"
467   "The header that Bsfilter inserts in messages."
468   :type 'string
469   :group 'spam-bsfilter)
470
471 (defcustom spam-bsfilter-spam-switch "--add-spam"
472   "The switch that Bsfilter uses to register spam messages."
473   :type 'string
474   :group 'spam-bsfilter)
475
476 (defcustom spam-bsfilter-ham-switch "--add-ham"
477   "The switch that Bsfilter uses to register ham messages."
478   :type 'string
479   :group 'spam-bsfilter)
480
481 (defcustom spam-bsfilter-spam-strong-switch "--sub-spam"
482   "The switch that Bsfilter uses to unregister ham messages."
483   :type 'string
484   :group 'spam-bsfilter)
485
486 (defcustom spam-bsfilter-ham-strong-switch "--sub-clean"
487   "The switch that Bsfilter uses to unregister spam messages."
488   :type 'string
489   :group 'spam-bsfilter)
490
491 (defcustom spam-bsfilter-database-directory nil
492   "Directory path of the Bsfilter databases."
493   :type '(choice (directory
494                   :tag "Location of the Bsfilter database directory")
495                  (const :tag "Use the default"))
496   :group 'spam-bsfilter)
497
498 (defgroup spam-spamoracle nil
499   "Spam spamoracle configuration."
500   :group 'spam)
501
502 (defcustom spam-spamoracle-database nil
503   "Location of spamoracle database file.
504 When nil, use the default spamoracle database."
505   :type '(choice (directory :tag "Location of spamoracle database file.")
506                  (const :tag "Use the default"))
507   :group 'spam-spamoracle)
508
509 (defcustom spam-spamoracle-binary (executable-find "spamoracle")
510   "Location of the spamoracle binary."
511   :type '(choice (directory :tag "Location of the spamoracle binary")
512                  (const :tag "Use the default"))
513   :group 'spam-spamoracle)
514
515 (defgroup spam-spamassassin nil
516   "Spam SpamAssassin configuration."
517   :group 'spam)
518
519 (defcustom spam-spamassassin-path (executable-find "spamassassin")
520   "File path of the spamassassin executable program.
521 Hint: set this to \"spamc\" if you have spamd running.  See the spamc and
522 spamd man pages for more information on these programs."
523   :type '(choice (file :tag "Location of spamc")
524                  (const :tag "spamassassin is not installed"))
525   :group 'spam-spamassassin)
526
527 (defcustom spam-spamassassin-arguments ()
528   "Arguments to pass to the spamassassin executable.
529 This must be a list.  For example, `(\"-C\" \"configfile\")'."
530   :type '(restricted-sexp :match-alternatives (listp))
531   :group 'spam-spamassassin)
532
533 (defcustom spam-spamassassin-spam-flag-header "X-Spam-Flag"
534   "The header inserted by SpamAssassin to flag spam."
535   :type 'string
536   :group 'spam-spamassassin)
537
538 (defcustom spam-spamassassin-positive-spam-flag-header "YES"
539   "The regex on `spam-spamassassin-spam-flag-header' for positive spam
540 identification"
541   :type 'string
542   :group 'spam-spamassassin)
543
544 (defcustom spam-spamassassin-spam-status-header "X-Spam-Status"
545   "The header inserted by SpamAssassin, giving extended scoring information"
546   :type 'string
547   :group 'spam-spamassassin)
548
549 (defcustom spam-sa-learn-path (executable-find "sa-learn")
550   "File path of the sa-learn executable program."
551   :type '(choice (file :tag "Location of spamassassin")
552                  (const :tag "spamassassin is not installed"))
553   :group 'spam-spamassassin)
554
555 (defcustom spam-sa-learn-rebuild t
556   "Whether sa-learn should rebuild the database every time it is called
557 Enable this if you want sa-learn to rebuild the database automatically.  Doing
558 this will slightly increase the running time of the spam registration process.
559 If you choose not to do this, you will have to run \"sa-learn --rebuild\" in
560 order for SpamAssassin to recognize the new registered spam."
561   :type 'boolean
562   :group 'spam-spamassassin)
563
564 (defcustom spam-sa-learn-spam-switch "--spam"
565   "The switch that sa-learn uses to register spam messages"
566   :type 'string
567   :group 'spam-spamassassin)
568
569 (defcustom spam-sa-learn-ham-switch "--ham"
570   "The switch that sa-learn uses to register ham messages"
571   :type 'string
572   :group 'spam-spamassassin)
573
574 (defcustom spam-sa-learn-unregister-switch "--forget"
575   "The switch that sa-learn uses to unregister messages messages"
576   :type 'string
577   :group 'spam-spamassassin)
578
579 ;;; Key bindings for spam control.
580
581 (gnus-define-keys gnus-summary-mode-map
582   "St" spam-generic-score
583   "Sx" gnus-summary-mark-as-spam
584   "Mst" spam-generic-score
585   "Msx" gnus-summary-mark-as-spam
586   "\M-d" gnus-summary-mark-as-spam)
587
588 (defvar spam-cache-lookups t
589   "Whether spam.el will try to cache lookups using `spam-caches'.")
590
591 (defvar spam-caches (make-hash-table
592                      :size 10
593                      :test 'equal)
594   "Cache of spam detection entries.")
595
596 (defvar spam-old-ham-articles nil
597   "List of old ham articles, generated when a group is entered.")
598
599 (defvar spam-old-spam-articles nil
600   "List of old spam articles, generated when a group is entered.")
601
602 (defvar spam-split-disabled nil
603   "If non-nil, `spam-split' is disabled, and always returns nil.")
604
605 (defvar spam-split-last-successful-check nil
606   "Internal variable.
607 `spam-split' will set this to nil or a spam-use-XYZ check if it
608 finds ham or spam.")
609
610 ;; convenience functions
611 (defun spam-clear-cache (symbol)
612   "Clear the spam-caches entry for a check."
613   (remhash symbol spam-caches))
614
615 (defun spam-xor (a b)
616   "Logical A xor B."
617   (and (or a b) (not (and a b))))
618
619 (defun spam-group-ham-mark-p (group mark &optional spam)
620   "Checks if MARK is considered a ham mark in GROUP."
621   (when (stringp group)
622     (let* ((marks (spam-group-ham-marks group spam))
623            (marks (if (symbolp mark)
624                       marks
625                     (mapcar 'symbol-value marks))))
626       (memq mark marks))))
627
628 (defun spam-group-spam-mark-p (group mark)
629   "Checks if MARK is considered a spam mark in GROUP."
630   (spam-group-ham-mark-p group mark t))
631
632 (defun spam-group-ham-marks (group &optional spam)
633   "In GROUP, get all the ham marks."
634   (when (stringp group)
635     (let* ((marks (if spam
636                       (gnus-parameter-spam-marks group)
637                     (gnus-parameter-ham-marks group)))
638            (marks (car marks))
639            (marks (if (listp (car marks)) (car marks) marks)))
640       marks)))
641
642 (defun spam-group-spam-marks (group)
643   "In GROUP, get all the spam marks."
644   (spam-group-ham-marks group t))
645
646 (defun spam-group-spam-contents-p (group)
647   "Is GROUP a spam group?"
648   (if (stringp group)
649       (or (member group spam-junk-mailgroups)
650           (memq 'gnus-group-spam-classification-spam
651                 (gnus-parameter-spam-contents group)))
652     nil))
653
654 (defun spam-group-ham-contents-p (group)
655   "Is GROUP a ham group?"
656   (if (stringp group)
657       (memq 'gnus-group-spam-classification-ham
658             (gnus-parameter-spam-contents group))
659     nil))
660
661 (defvar spam-list-of-processors
662   '((gnus-group-spam-exit-processor-report-gmane spam spam-use-gmane)
663     (gnus-group-spam-exit-processor-bogofilter   spam spam-use-bogofilter)
664     (gnus-group-spam-exit-processor-bsfilter     spam spam-use-bsfilter)
665     (gnus-group-spam-exit-processor-blacklist    spam spam-use-blacklist)
666     (gnus-group-spam-exit-processor-ifile        spam spam-use-ifile)
667     (gnus-group-spam-exit-processor-stat         spam spam-use-stat)
668     (gnus-group-spam-exit-processor-spamoracle   spam spam-use-spamoracle)
669     (gnus-group-spam-exit-processor-spamassassin spam spam-use-spamassassin)
670     (gnus-group-ham-exit-processor-ifile         ham spam-use-ifile)
671     (gnus-group-ham-exit-processor-bogofilter    ham spam-use-bogofilter)
672     (gnus-group-ham-exit-processor-bsfilter      ham spam-use-bsfilter)
673     (gnus-group-ham-exit-processor-stat          ham spam-use-stat)
674     (gnus-group-ham-exit-processor-whitelist     ham spam-use-whitelist)
675     (gnus-group-ham-exit-processor-BBDB          ham spam-use-BBDB)
676     (gnus-group-ham-exit-processor-copy          ham spam-use-ham-copy)
677     (gnus-group-ham-exit-processor-spamassassin  ham spam-use-spamassassin)
678     (gnus-group-ham-exit-processor-spamoracle    ham spam-use-spamoracle))
679   "The `spam-list-of-processors' list.
680 This list contains pairs associating a ham/spam exit processor
681 variable with a classification and a spam-use-* variable.")
682
683 (defun spam-group-processor-p (group processor)
684   (if (and (stringp group)
685            (symbolp processor))
686       (or (member processor (nth 0 (gnus-parameter-spam-process group)))
687           (spam-group-processor-multiple-p
688            group
689            (cdr-safe (assoc processor spam-list-of-processors))))
690     nil))
691
692 (defun spam-group-processor-multiple-p (group processor-info)
693   (let* ((classification (nth 0 processor-info))
694          (check (nth 1 processor-info))
695          (parameters (nth 0 (gnus-parameter-spam-process group)))
696          found)
697     (dolist (parameter parameters)
698       (when (and (null found)
699                  (listp parameter)
700                  (eq classification (nth 0 parameter))
701                  (eq check (nth 1 parameter)))
702         (setq found t)))
703     found))
704
705 (defun spam-group-spam-processor-report-gmane-p (group)
706   (spam-group-processor-p group 'gnus-group-spam-exit-processor-report-gmane))
707
708 (defun spam-group-spam-processor-bogofilter-p (group)
709   (spam-group-processor-p group 'gnus-group-spam-exit-processor-bogofilter))
710
711 (defun spam-group-spam-processor-blacklist-p (group)
712   (spam-group-processor-p group 'gnus-group-spam-exit-processor-blacklist))
713
714 (defun spam-group-spam-processor-ifile-p (group)
715   (spam-group-processor-p group 'gnus-group-spam-exit-processor-ifile))
716
717 (defun spam-group-ham-processor-ifile-p (group)
718   (spam-group-processor-p group 'gnus-group-ham-exit-processor-ifile))
719
720 (defun spam-group-spam-processor-spamoracle-p (group)
721   (spam-group-processor-p group 'gnus-group-spam-exit-processor-spamoracle))
722
723 (defun spam-group-ham-processor-bogofilter-p (group)
724   (spam-group-processor-p group 'gnus-group-ham-exit-processor-bogofilter))
725
726 (defun spam-group-spam-processor-stat-p (group)
727   (spam-group-processor-p group 'gnus-group-spam-exit-processor-stat))
728
729 (defun spam-group-ham-processor-stat-p (group)
730   (spam-group-processor-p group 'gnus-group-ham-exit-processor-stat))
731
732 (defun spam-group-ham-processor-whitelist-p (group)
733   (spam-group-processor-p group 'gnus-group-ham-exit-processor-whitelist))
734
735 (defun spam-group-ham-processor-BBDB-p (group)
736   (spam-group-processor-p group 'gnus-group-ham-exit-processor-BBDB))
737
738 (defun spam-group-ham-processor-copy-p (group)
739   (spam-group-processor-p group 'gnus-group-ham-exit-processor-copy))
740
741 (defun spam-group-ham-processor-spamoracle-p (group)
742   (spam-group-processor-p group 'gnus-group-ham-exit-processor-spamoracle))
743
744 (defun spam-report-articles-gmane (n)
745   "Report the current message as spam.
746 Respects the process/prefix convention."
747   (interactive "P")
748   (dolist (article (gnus-summary-work-articles n))
749     (gnus-summary-remove-process-mark article)
750     (spam-report-gmane article)))
751
752 (defun spam-necessary-extra-headers ()
753   "Return the extra headers spam.el thinks are necessary."
754   (let (list)
755     (when (or spam-use-spamassassin
756               spam-use-spamassassin-headers
757               spam-use-regex-headers)
758       (push 'X-Spam-Status list))
759     list))
760
761 (defun spam-user-format-function-S (headers)
762   (when headers
763     (spam-summary-score headers)))
764
765 (defun spam-article-sort-by-spam-status (h1 h2)
766   "Sort articles by score."
767   (let (result)
768     (dolist (header (spam-necessary-extra-headers))
769       (let ((s1 (spam-summary-score h1 header))
770             (s2 (spam-summary-score h2 header)))
771       (unless (= s1 s2)
772         (setq result (< s1 s2))
773         (return))))
774     result))
775
776 (defun spam-extra-header-to-number (header headers)
777   "Transform an extra header to a number."
778   (if (gnus-extra-header header headers)
779       (cond
780        ((eq header 'X-Spam-Status)
781         (string-to-number (gnus-replace-in-string
782                            (gnus-extra-header header headers)
783                            ".*hits=" "")))
784        (t nil))
785     nil))
786
787 (defun spam-summary-score (headers &optional specific-header)
788   "Score an article for the summary buffer, as fast as possible.
789 With SPECIFIC-HEADER, returns only that header's score.
790 Will not return a nil score."
791   (let (score)
792     (dolist (header 
793              (if specific-header
794                  (list specific-header)
795                (spam-necessary-extra-headers)))
796       (setq score 
797             (spam-extra-header-to-number header headers))
798       (when score 
799         (return)))
800     (or score 0)))
801
802 (defun spam-generic-score (&optional recheck)
803   "Invoke whatever scoring method we can."
804   (interactive "P")
805   (cond
806    ((or spam-use-spamassassin spam-use-spamassassin-headers)
807     (spam-spamassassin-score recheck))
808    ((or spam-use-bsfilter spam-use-bsfilter-headers)
809     (spam-bsfilter-score recheck))
810    (t (spam-bogofilter-score recheck))))
811
812 ;;; Summary entry and exit processing.
813
814 (defun spam-summary-prepare ()
815   (setq spam-old-ham-articles
816         (spam-list-articles gnus-newsgroup-articles 'ham))
817   (setq spam-old-spam-articles
818         (spam-list-articles gnus-newsgroup-articles 'spam))
819   (spam-mark-junk-as-spam-routine))
820
821 ;; The spam processors are invoked for any group, spam or ham or neither
822 (defun spam-summary-prepare-exit ()
823   (unless gnus-group-is-exiting-without-update-p
824     (gnus-message 6 "Exiting summary buffer and applying spam rules")
825
826     ;; first of all, unregister any articles that are no longer ham or spam
827     ;; we have to iterate over the processors, or else we'll be too slow
828     (dolist (classification '(spam ham))
829       (let* ((old-articles (if (eq classification 'spam)
830                                spam-old-spam-articles
831                              spam-old-ham-articles))
832              (new-articles (spam-list-articles
833                             gnus-newsgroup-articles
834                             classification))
835              (changed-articles (spam-set-difference new-articles old-articles)))
836         ;; now that we have the changed articles, we go through the processors
837         (dolist (processor-param spam-list-of-processors)
838           (let ((processor (nth 0 processor-param))
839                 (processor-classification (nth 1 processor-param))
840                 (check (nth 2 processor-param))
841                 unregister-list)
842             (dolist (article changed-articles)
843               (let ((id (spam-fetch-field-message-id-fast article)))
844                 (when (spam-log-unregistration-needed-p
845                        id 'process classification check)
846                   (push article unregister-list))))
847             ;; call spam-register-routine with specific articles to unregister,
848             ;; when there are articles to unregister and the check is enabled
849             (when (and unregister-list (symbol-value check))
850               (spam-register-routine 
851                classification check t unregister-list))))))
852
853     ;; find all the spam processors applicable to this group
854     (dolist (processor-param spam-list-of-processors)
855       (let ((processor (nth 0 processor-param))
856             (classification (nth 1 processor-param))
857             (check (nth 2 processor-param)))
858         (when (and (eq 'spam classification)
859                    (spam-group-processor-p gnus-newsgroup-name processor))
860           (spam-register-routine classification check))))
861
862     (unless (and spam-move-spam-nonspam-groups-only
863                  (spam-group-spam-contents-p gnus-newsgroup-name))
864       (gnus-message 6 "Marking spam as expired and moving it to %s"
865                     (gnus-parameter-spam-process-destination 
866                      gnus-newsgroup-name))
867       (spam-mark-spam-as-expired-and-move-routine
868        (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
869
870     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
871     ;; expire spam, in case the above did not expire them
872     (gnus-message 6 "Marking spam as expired without moving it")
873     (spam-mark-spam-as-expired-and-move-routine nil)
874
875     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
876               (and (spam-group-spam-contents-p gnus-newsgroup-name)
877                    spam-process-ham-in-spam-groups)
878               spam-process-ham-in-nonham-groups)
879       ;; find all the ham processors applicable to this group
880       (dolist (processor-param spam-list-of-processors)
881         (let ((processor (nth 0 processor-param))
882               (classification (nth 1 processor-param))
883               (check (nth 2 processor-param)))
884           (when (and (eq 'ham classification)
885                      (spam-group-processor-p gnus-newsgroup-name processor))
886             (spam-register-routine classification check)))))
887
888     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
889       (gnus-message 6 "Copying ham")
890       (spam-ham-copy-routine
891        (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
892
893     ;; now move all ham articles out of spam groups
894     (when (spam-group-spam-contents-p gnus-newsgroup-name)
895       (gnus-message 6 "Moving ham messages from spam group")
896       (spam-ham-move-routine
897        (gnus-parameter-ham-process-destination gnus-newsgroup-name))))
898
899   (setq spam-old-ham-articles nil)
900   (setq spam-old-spam-articles nil))
901
902 (defun spam-set-difference (list1 list2)
903   "Return a set difference of LIST1 and LIST2.  
904 When either list is nil, the other is returned."
905   (if (and list1 list2)
906       ;; we have two non-nil lists
907       (progn
908         (dolist (item (append list1 list2))
909           (when (and (memq item list1) (memq item list2))
910             (setq list1 (delq item list1))
911             (setq list2 (delq item list2))))
912         (append list1 list2))
913     ;; if either of the lists was nil, return the other one
914     (if list1 list1 list2)))
915
916 (defun spam-mark-junk-as-spam-routine ()
917   ;; check the global list of group names spam-junk-mailgroups and the
918   ;; group parameters
919   (when (spam-group-spam-contents-p gnus-newsgroup-name)
920     (gnus-message 6 "Marking %s articles as spam"
921                   (if spam-mark-only-unseen-as-spam
922                       "unseen"
923                     "unread"))
924     (let ((articles (if spam-mark-only-unseen-as-spam
925                         gnus-newsgroup-unseen
926                       gnus-newsgroup-unreads)))
927       (dolist (article articles)
928         (gnus-summary-mark-article article gnus-spam-mark)))))
929
930 (defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
931   (if (and (car-safe groups) (listp (car-safe groups)))
932       (apply 'spam-mark-spam-as-expired-and-move-routine (car groups))
933     (gnus-summary-kill-process-mark)
934     (let ((articles gnus-newsgroup-articles)
935           (backend-supports-deletions
936            (gnus-check-backend-function
937             'request-move-article gnus-newsgroup-name))
938           article tomove deletep)
939       (dolist (article articles)
940         (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
941           (gnus-summary-mark-article article gnus-expirable-mark)
942           (push article tomove)))
943
944       ;; now do the actual copies
945       (dolist (group groups)
946         (when (and tomove
947                    (stringp group))
948           (dolist (article tomove)
949             (gnus-summary-set-process-mark article))
950           (when tomove
951             (if (or (not backend-supports-deletions)
952                     (> (length groups) 1))
953                 (progn
954                   (gnus-summary-copy-article nil group)
955                   (setq deletep t))
956               (gnus-summary-move-article nil group)))))
957
958       ;; now delete the articles, if there was a copy done, and the
959       ;; backend allows it
960       (when (and deletep backend-supports-deletions)
961         (dolist (article tomove)
962           (gnus-summary-set-process-mark article))
963         (when tomove
964           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
965             (gnus-summary-delete-article nil))))
966
967       (gnus-summary-yank-process-mark))))
968
969 (defun spam-ham-copy-or-move-routine (copy groups)
970   (gnus-summary-kill-process-mark)
971   (let ((todo (spam-list-articles gnus-newsgroup-articles 'ham))
972         (backend-supports-deletions
973          (gnus-check-backend-function
974           'request-move-article gnus-newsgroup-name))
975         (respool-method (gnus-find-method-for-group gnus-newsgroup-name))
976         article mark todo deletep respool)
977
978     (when (member 'respool groups)
979       (setq respool t)                  ; boolean for later
980       (setq groups '("fake"))) ; when respooling, groups are dynamic so fake it
981
982     ;; now do the actual move
983     (dolist (group groups)
984       (when (and todo (stringp group))
985         (dolist (article todo)
986           (when spam-mark-ham-unread-before-move-from-spam-group
987             (gnus-summary-mark-article article gnus-unread-mark))
988           (gnus-summary-set-process-mark article))
989
990         (if respool                        ; respooling is with a "fake" group
991             (let ((spam-split-disabled
992                    (or spam-split-disabled
993                        spam-disable-spam-split-during-ham-respool)))
994               (gnus-summary-respool-article nil respool-method))
995           (if (or (not backend-supports-deletions) ; else, we are not respooling
996                   (> (length groups) 1))
997               (progn                ; if copying, copy and set deletep
998                 (gnus-summary-copy-article nil group)
999                 (setq deletep t))
1000             (gnus-summary-move-article nil group))))) ; else move articles
1001
1002     ;; now delete the articles, unless a) copy is t, and there was a copy done
1003     ;;                                 b) a move was done to a single group
1004     ;;                                 c) backend-supports-deletions is nil
1005     (unless copy
1006       (when (and deletep backend-supports-deletions)
1007         (dolist (article todo)
1008           (gnus-summary-set-process-mark article))
1009         (when todo
1010           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
1011             (gnus-summary-delete-article nil))))))
1012
1013   (gnus-summary-yank-process-mark))
1014
1015 (defun spam-ham-copy-routine (&rest groups)
1016   (if (and (car-safe groups) (listp (car-safe groups)))
1017       (apply 'spam-ham-copy-routine (car groups))
1018     (spam-ham-copy-or-move-routine t groups)))
1019
1020 (defun spam-ham-move-routine (&rest groups)
1021   (if (and (car-safe groups) (listp (car-safe groups)))
1022       (apply 'spam-ham-move-routine (car groups))
1023     (spam-ham-copy-or-move-routine nil groups)))
1024
1025 (defun spam-get-article-as-string (article)
1026   (when (numberp article)
1027     (with-temp-buffer
1028       (gnus-request-article-this-buffer
1029        article
1030        gnus-newsgroup-name)
1031       (buffer-string))))
1032
1033 ;; disabled for now
1034 ;; (defun spam-get-article-as-filename (article)
1035 ;;   (let ((article-filename))
1036 ;;     (when (numberp article)
1037 ;;       (nnml-possibly-change-directory
1038 ;;        (gnus-group-real-name gnus-newsgroup-name))
1039 ;;       (setq article-filename (expand-file-name
1040 ;;                              (int-to-string article) nnml-current-directory)))
1041 ;;     (if (file-exists-p article-filename)
1042 ;;      article-filename
1043 ;;       nil)))
1044
1045 (defun spam-fetch-field-fast (article field &optional prepared-data-header)
1046   "Fetch a field quickly, using the internal gnus-data-list function"
1047   (when (numberp article)
1048     (let* ((data-header (or prepared-data-header
1049                             (spam-fetch-article-header article))))
1050       (if (arrayp data-header)
1051         (cond
1052          ((equal field 'from)
1053           (mail-header-from data-header))
1054          ((equal field 'message-id)
1055           (mail-header-message-id data-header))
1056          ((equal field 'subject)
1057           (mail-header-subject data-header))
1058          ((equal field 'references)
1059           (mail-header-references data-header))
1060          ((equal field 'date)
1061           (mail-header-date data-header))
1062          ((equal field 'xref)
1063           (mail-header-xref data-header))
1064          ((equal field 'extra)
1065           (mail-header-extra data-header))
1066          (t
1067           nil))
1068         (gnus-message 6 "Article %d has a nil data header" article)))))
1069
1070 (defun spam-fetch-field-from-fast (article &optional prepared-data-header)
1071   (spam-fetch-field-fast article 'from prepared-data-header))
1072
1073 (defun spam-fetch-field-subject-fast (article &optional prepared-data-header)
1074   (spam-fetch-field-fast article 'subject prepared-data-header))
1075
1076 (defun spam-fetch-field-message-id-fast (article &optional prepared-data-header)
1077   (spam-fetch-field-fast article 'message-id prepared-data-header))
1078
1079 (defun spam-generate-fake-headers (article)
1080   (let ((dh (spam-fetch-article-header article)))
1081     (if dh
1082         (concat
1083          (format 
1084           (concat "From: %s\nSubject: %s\nMessage-ID: %s\n"
1085                   "Date: %s\nReferences: %s\nXref: %s\n")
1086           (spam-fetch-field-fast article 'from dh)
1087           (spam-fetch-field-fast article 'subject dh)
1088           (spam-fetch-field-fast article 'message-id dh)
1089           (spam-fetch-field-fast article 'date dh)
1090           (spam-fetch-field-fast article 'references dh)
1091           (spam-fetch-field-fast article 'xref dh))
1092          (when (spam-fetch-field-fast article 'extra dh)
1093            (format "%s\n" (spam-fetch-field-fast article 'extra dh))))
1094       (gnus-message
1095        5
1096        "spam-generate-fake-headers: article %d didn't have a valid header"
1097        article))))
1098
1099 (defun spam-fetch-article-header (article)
1100   (save-excursion
1101     (set-buffer gnus-summary-buffer)
1102     (gnus-read-header article)
1103     (nth 3 (assq article gnus-newsgroup-data))))
1104
1105 \f
1106 ;;;; Spam determination.
1107
1108 (defvar spam-list-of-checks
1109   '((spam-use-blacklist                 .       spam-check-blacklist)
1110     (spam-use-regex-headers             .       spam-check-regex-headers)
1111     (spam-use-gmane-xref                .       spam-check-gmane-xref)
1112     (spam-use-regex-body                .       spam-check-regex-body)
1113     (spam-use-whitelist                 .       spam-check-whitelist)
1114     (spam-use-BBDB                      .       spam-check-BBDB)
1115     (spam-use-BBDB-exclusive            .       spam-check-BBDB)
1116     (spam-use-ifile                     .       spam-check-ifile)
1117     (spam-use-spamoracle                .       spam-check-spamoracle)
1118     (spam-use-stat                      .       spam-check-stat)
1119     (spam-use-blackholes                .       spam-check-blackholes)
1120     (spam-use-hashcash                  .       spam-check-hashcash)
1121     (spam-use-spamassassin-headers      .       spam-check-spamassassin-headers)
1122     (spam-use-spamassassin              .       spam-check-spamassassin)
1123     (spam-use-bogofilter-headers        .       spam-check-bogofilter-headers)
1124     (spam-use-bogofilter                .       spam-check-bogofilter)
1125     (spam-use-bsfilter-headers          .       spam-check-bsfilter-headers)
1126     (spam-use-bsfilter                  .       spam-check-bsfilter))
1127   "The spam-list-of-checks list contains pairs associating a
1128 parameter variable with a spam checking function.  If the
1129 parameter variable is true, then the checking function is called,
1130 and its value decides what happens.  Each individual check may
1131 return nil, t, or a mailgroup name.  The value nil means that the
1132 check does not yield a decision, and so, that further checks are
1133 needed.  The value t means that the message is definitely not
1134 spam, and that further spam checks should be inhibited.
1135 Otherwise, a mailgroup name or the symbol 'spam (depending on
1136 spam-split-symbolic-return) is returned where the mail should go,
1137 and further checks are also inhibited.  The usual mailgroup name
1138 is the value of `spam-split-group', meaning that the message is
1139 definitely a spam.")
1140
1141 (defvar spam-list-of-statistical-checks
1142   '(spam-use-ifile
1143     spam-use-regex-body
1144     spam-use-stat
1145     spam-use-bogofilter
1146     spam-use-bsfilter
1147     spam-use-blackholes
1148     spam-use-spamassassin
1149     spam-use-spamoracle)
1150   "The spam-list-of-statistical-checks list contains all the mail
1151 splitters that need to have the full message body available.
1152 Note that you should fetch extra headers if you don't like this,
1153 e.g. fetch the 'Received' header for spam-use-blackholes.")
1154
1155 (defun spam-split (&rest specific-checks)
1156   "Split this message into the `spam' group if it is spam.
1157 This function can be used as an entry in the variable `nnmail-split-fancy',
1158 for example like this: (: spam-split).  It can take checks as
1159 parameters.  A string as a parameter will set the
1160 spam-split-group to that string.
1161
1162 See the Info node `(gnus)Fancy Mail Splitting' for more details."
1163   (interactive)
1164   (setq spam-split-last-successful-check nil)
1165   (unless spam-split-disabled
1166     (let ((spam-split-group-choice spam-split-group))
1167       (dolist (check specific-checks)
1168         (when (stringp check)
1169           (setq spam-split-group-choice check)
1170           (setq specific-checks (delq check specific-checks))))
1171
1172       (let ((spam-split-group spam-split-group-choice))
1173         (save-excursion
1174           (save-restriction
1175             (dolist (check spam-list-of-statistical-checks)
1176               (when (and (symbolp check)
1177                          (or (symbol-value check)
1178                              (memq check specific-checks)))
1179                 (widen)
1180                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
1181                               (symbol-name check))
1182                 (return)))
1183             ;;   (progn (widen) (debug (buffer-string)))
1184             (let ((list-of-checks spam-list-of-checks)
1185                   decision)
1186               (while (and list-of-checks (not decision))
1187                 (let ((pair (pop list-of-checks)))
1188                   (when (or
1189                          ;; either, given specific checks, this is one of them
1190                          (and specific-checks (memq (car pair) specific-checks))
1191                          ;; or, given no specific checks, spam-use-CHECK is set
1192                          (and (null specific-checks) (symbol-value (car pair))))
1193                     (gnus-message 6 "spam-split: calling the %s function"
1194                                   (symbol-name (cdr pair)))
1195                     (setq decision (funcall (cdr pair)))
1196                     ;; if we got a decision at all, save the current check
1197                     (when decision
1198                       (setq spam-split-last-successful-check (car pair)))
1199
1200                     (when (eq decision 'spam)
1201                       (unless spam-split-symbolic-return
1202                         (gnus-error
1203                          5
1204                          (format "spam-split got %s but %s is nil"
1205                                  (symbol-name decision)
1206                                  (symbol-name spam-split-symbolic-return))))))))
1207               (if (eq decision t)
1208                   (if spam-split-symbolic-return-positive 'ham nil)
1209                 decision))))))))
1210
1211 (defun spam-find-spam ()
1212   "This function will detect spam in the current newsgroup using spam-split."
1213   (interactive)
1214
1215   (let* ((group gnus-newsgroup-name)
1216          (autodetect (gnus-parameter-spam-autodetect group))
1217          (methods (gnus-parameter-spam-autodetect-methods group))
1218          (first-method (nth 0 methods))
1219          (articles (if spam-autodetect-recheck-messages
1220                        gnus-newsgroup-articles
1221                      gnus-newsgroup-unseen))
1222          article-cannot-be-faked)
1223
1224     (dolist (check spam-list-of-statistical-checks)
1225       (when (and (symbolp check)
1226                  (memq check methods))
1227         (setq article-cannot-be-faked t)
1228         (return)))
1229
1230     (when (memq 'default methods)
1231       (setq article-cannot-be-faked t))
1232
1233     (when (and autodetect
1234                (not (equal first-method 'none)))
1235     (mapcar
1236      (lambda (article)
1237        (let ((id (spam-fetch-field-message-id-fast article))
1238              (subject (spam-fetch-field-subject-fast article))
1239              (sender (spam-fetch-field-from-fast article))
1240              registry-lookup)
1241          
1242          (unless id
1243            (gnus-message 6 "Article %d has no message ID!" article))
1244          
1245          (when (and id spam-log-to-registry)
1246            (setq registry-lookup (spam-log-registration-type id 'incoming))
1247            (when registry-lookup
1248              (gnus-message
1249               9
1250               "spam-find-spam: message %s was already registered incoming"
1251               id)))
1252
1253          (let* ((spam-split-symbolic-return t)
1254                 (spam-split-symbolic-return-positive t)
1255                 (fake-headers (spam-generate-fake-headers article))
1256                 (split-return
1257                  (or registry-lookup
1258                      (with-temp-buffer
1259                        (if article-cannot-be-faked
1260                            (gnus-request-article-this-buffer
1261                             article
1262                             group)
1263                          ;; else, we fake the article
1264                          (when fake-headers (insert fake-headers)))
1265                        (if (or (null first-method)
1266                                (equal first-method 'default))
1267                            (spam-split)
1268                          (apply 'spam-split methods))))))
1269            (if (equal split-return 'spam)
1270                (gnus-summary-mark-article article gnus-spam-mark))
1271            
1272            (when (and id split-return spam-log-to-registry)
1273              (when (zerop (gnus-registry-group-count id))
1274                (gnus-registry-add-group
1275                 id group subject sender))
1276                
1277              (unless registry-lookup
1278                (spam-log-processing-to-registry
1279                 id
1280                 'incoming
1281                 split-return
1282                 spam-split-last-successful-check
1283                 group))))))
1284     articles))))
1285
1286 (defvar spam-registration-functions
1287   ;; first the ham register, second the spam register function
1288   ;; third the ham unregister, fourth the spam unregister function
1289   '((spam-use-blacklist  nil
1290                          spam-blacklist-register-routine
1291                          nil
1292                          spam-blacklist-unregister-routine)
1293     (spam-use-whitelist  spam-whitelist-register-routine
1294                          nil
1295                          spam-whitelist-unregister-routine
1296                          nil)
1297     (spam-use-ham-copy   nil
1298                          nil
1299                          nil
1300                          nil)
1301     (spam-use-BBDB       spam-BBDB-register-routine
1302                          nil
1303                          spam-BBDB-unregister-routine
1304                          nil)
1305     (spam-use-ifile      spam-ifile-register-ham-routine
1306                          spam-ifile-register-spam-routine
1307                          spam-ifile-unregister-ham-routine
1308                          spam-ifile-unregister-spam-routine)
1309     (spam-use-spamoracle spam-spamoracle-learn-ham
1310                          spam-spamoracle-learn-spam
1311                          spam-spamoracle-unlearn-ham
1312                          spam-spamoracle-unlearn-spam)
1313     (spam-use-stat       spam-stat-register-ham-routine
1314                          spam-stat-register-spam-routine
1315                          spam-stat-unregister-ham-routine
1316                          spam-stat-unregister-spam-routine)
1317     ;; note that spam-use-gmane is not a legitimate check
1318     (spam-use-gmane      nil
1319                          spam-report-gmane-register-routine
1320                          ;; does Gmane support unregistration?
1321                          nil
1322                          nil)
1323     (spam-use-spamassassin spam-spamassassin-register-ham-routine
1324                            spam-spamassassin-register-spam-routine
1325                            spam-spamassassin-unregister-ham-routine
1326                            spam-spamassassin-unregister-spam-routine)
1327     (spam-use-bogofilter spam-bogofilter-register-ham-routine
1328                          spam-bogofilter-register-spam-routine
1329                          spam-bogofilter-unregister-ham-routine
1330                          spam-bogofilter-unregister-spam-routine)
1331     (spam-use-bsfilter   spam-bsfilter-register-ham-routine
1332                          spam-bsfilter-register-spam-routine
1333                          spam-bsfilter-unregister-ham-routine
1334                          spam-bsfilter-unregister-spam-routine))
1335   "The spam-registration-functions list contains pairs
1336 associating a parameter variable with the ham and spam
1337 registration functions, and the ham and spam unregistration
1338 functions")
1339
1340 (defun spam-classification-valid-p (classification)
1341   (or  (eq classification 'spam)
1342        (eq classification 'ham)))
1343
1344 (defun spam-process-type-valid-p (process-type)
1345   (or  (eq process-type 'incoming)
1346        (eq process-type 'process)))
1347
1348 (defun spam-registration-check-valid-p (check)
1349   (assoc check spam-registration-functions))
1350
1351 (defun spam-unregistration-check-valid-p (check)
1352   (assoc check spam-registration-functions))
1353
1354 (defun spam-registration-function (classification check)
1355   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1356     (if (eq classification 'spam)
1357         (nth 1 flist)
1358       (nth 0 flist))))
1359
1360 (defun spam-unregistration-function (classification check)
1361   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1362     (if (eq classification 'spam)
1363         (nth 3 flist)
1364       (nth 2 flist))))
1365
1366 (defun spam-list-articles (articles classification)
1367   (let ((mark-check (if (eq classification 'spam)
1368                         'spam-group-spam-mark-p
1369                       'spam-group-ham-mark-p))
1370         alist mark-cache-yes mark-cache-no)
1371     (dolist (article articles)
1372       (let ((mark (gnus-summary-article-mark article)))
1373         (unless (or (memq mark mark-cache-yes)
1374                     (memq mark mark-cache-no))
1375           (if (funcall mark-check
1376                        gnus-newsgroup-name
1377                        mark)
1378               (push mark mark-cache-yes)
1379             (push mark mark-cache-no)))
1380         (when (memq mark mark-cache-yes)
1381           (push article alist))))
1382     alist))
1383
1384 (defun spam-register-routine (classification
1385                               check
1386                               &optional unregister
1387                               specific-articles)
1388   (when (and (spam-classification-valid-p classification)
1389              (spam-registration-check-valid-p check))
1390     (let* ((register-function
1391             (spam-registration-function classification check))
1392            (unregister-function
1393             (spam-unregistration-function classification check))
1394            (run-function (if unregister
1395                              unregister-function
1396                            register-function))
1397            (log-function (if unregister
1398                              'spam-log-undo-registration
1399                            'spam-log-processing-to-registry))
1400            article articles)
1401
1402       (when run-function
1403         ;; make list of articles, using specific-articles if given
1404         (setq articles (or specific-articles
1405                            (spam-list-articles
1406                             gnus-newsgroup-articles
1407                             classification)))
1408         ;; process them
1409         (gnus-message 5 "%s %d %s articles as %s using backend %s"
1410                       (if unregister "Unregistering" "Registering")
1411                       (length articles)
1412                       (if specific-articles "specific" "")
1413                       (symbol-name classification)
1414                       (symbol-name check))
1415         (funcall run-function articles)
1416         ;; now log all the registrations (or undo them, depending on unregister)
1417         (dolist (article articles)
1418           (funcall log-function
1419                    (spam-fetch-field-message-id-fast article)
1420                    'process
1421                    classification
1422                    check
1423                    gnus-newsgroup-name))))))
1424
1425 ;;; log a ham- or spam-processor invocation to the registry
1426 (defun spam-log-processing-to-registry (id type classification check group)
1427   (when spam-log-to-registry
1428     (if (and (stringp id)
1429              (stringp group)
1430              (spam-process-type-valid-p type)
1431              (spam-classification-valid-p classification)
1432              (spam-registration-check-valid-p check))
1433         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1434               (cell (list classification check group)))
1435           (push cell cell-list)
1436           (gnus-registry-store-extra-entry
1437            id
1438            type
1439            cell-list))
1440
1441       (gnus-message 
1442        5 
1443        (format "%s call with bad ID, type, classification, spam-check, or group"
1444                "spam-log-processing-to-registry")))))
1445
1446 ;;; check if a ham- or spam-processor registration has been done
1447 (defun spam-log-registered-p (id type)
1448   (when spam-log-to-registry
1449     (if (and (stringp id)
1450              (spam-process-type-valid-p type))
1451         (cdr-safe (gnus-registry-fetch-extra id type))
1452       (progn
1453         (gnus-message
1454          5 
1455          (format "%s called with bad ID, type, classification, or spam-check"
1456                  "spam-log-registered-p"))
1457         nil))))
1458
1459 ;;; check what a ham- or spam-processor registration says
1460 ;;; returns nil if conflicting registrations are found
1461 (defun spam-log-registration-type (id type)
1462   (let ((count 0)
1463         decision)
1464     (dolist (reg (spam-log-registered-p id type))
1465       (let ((classification (nth 0 reg)))
1466         (when (spam-classification-valid-p classification)
1467           (when (and decision
1468                      (not (eq classification decision)))
1469             (setq count (+ 1 count)))
1470           (setq decision classification))))
1471     (if (< 0 count)
1472         nil
1473       decision)))
1474
1475
1476 ;;; check if a ham- or spam-processor registration needs to be undone
1477 (defun spam-log-unregistration-needed-p (id type classification check)
1478   (when spam-log-to-registry
1479     (if (and (stringp id)
1480              (spam-process-type-valid-p type)
1481              (spam-classification-valid-p classification)
1482              (spam-registration-check-valid-p check))
1483         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1484               found)
1485           (dolist (cell cell-list)
1486             (unless found
1487               (when (and (eq classification (nth 0 cell))
1488                          (eq check (nth 1 cell)))
1489                 (setq found t))))
1490           found)
1491       (progn
1492         (gnus-message
1493          5 
1494          (format "%s called with bad ID, type, classification, or spam-check"
1495                  "spam-log-unregistration-needed-p"))
1496         nil))))
1497
1498
1499 ;;; undo a ham- or spam-processor registration (the group is not used)
1500 (defun spam-log-undo-registration (id type classification check &optional group)
1501   (when (and spam-log-to-registry
1502              (spam-log-unregistration-needed-p id type classification check))
1503     (if (and (stringp id)
1504              (spam-process-type-valid-p type)
1505              (spam-classification-valid-p classification)
1506              (spam-registration-check-valid-p check))
1507         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1508               new-cell-list found)
1509           (dolist (cell cell-list)
1510             (unless (and (eq classification (nth 0 cell))
1511                          (eq check (nth 1 cell)))
1512               (push cell new-cell-list)))
1513           (gnus-registry-store-extra-entry
1514            id
1515            type
1516            new-cell-list))
1517       (progn
1518         (gnus-message 6 (format "%s call with bad ID, type, spam-check, or group"
1519                                 "spam-log-undo-registration"))
1520         nil))))
1521
1522 ;;; set up IMAP widening if it's necessary
1523 (defun spam-setup-widening ()
1524   (dolist (check spam-list-of-statistical-checks)
1525     (when (symbol-value check)
1526       (setq nnimap-split-download-body-default t))))
1527
1528 \f
1529 ;;;; Gmane xrefs
1530 (defun spam-check-gmane-xref ()
1531   (let ((header (or
1532                  (message-fetch-field "Xref")
1533                  (message-fetch-field "Newsgroups")))
1534         (spam-split-group (if spam-split-symbolic-return
1535                               'spam
1536                             spam-split-group)))
1537     (when header                        ; return nil when no header
1538       (when (string-match spam-gmane-xref-spam-group
1539                           header)
1540           spam-split-group))))
1541
1542 \f
1543 ;;;; Regex body
1544
1545 (defun spam-check-regex-body ()
1546   (let ((spam-regex-headers-ham spam-regex-body-ham)
1547         (spam-regex-headers-spam spam-regex-body-spam))
1548     (spam-check-regex-headers t)))
1549
1550 \f
1551 ;;;; Regex headers
1552
1553 (defun spam-check-regex-headers (&optional body)
1554   (let ((type (if body "body" "header"))
1555         (spam-split-group (if spam-split-symbolic-return
1556                               'spam
1557                             spam-split-group))
1558         ret found)
1559     (dolist (h-regex spam-regex-headers-ham)
1560       (unless found
1561         (goto-char (point-min))
1562         (when (re-search-forward h-regex nil t)
1563           (message "Ham regex %s search positive." type)
1564           (setq found t))))
1565     (dolist (s-regex spam-regex-headers-spam)
1566       (unless found
1567         (goto-char (point-min))
1568         (when (re-search-forward s-regex nil t)
1569           (message "Spam regex %s search positive." type)
1570           (setq found t)
1571           (setq ret spam-split-group))))
1572     ret))
1573
1574 \f
1575 ;;;; Blackholes.
1576
1577 (defun spam-reverse-ip-string (ip)
1578   (when (stringp ip)
1579     (mapconcat 'identity
1580                (nreverse (split-string ip "\\."))
1581                ".")))
1582
1583 (defun spam-check-blackholes ()
1584   "Check the Received headers for blackholed relays."
1585   (let ((headers (message-fetch-field "received"))
1586         (spam-split-group (if spam-split-symbolic-return
1587                               'spam
1588                             spam-split-group))
1589         ips matches)
1590     (when headers
1591       (with-temp-buffer
1592         (insert headers)
1593         (goto-char (point-min))
1594         (gnus-message 6 "Checking headers for relay addresses")
1595         (while (re-search-forward
1596                 "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
1597           (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
1598           (push (spam-reverse-ip-string (match-string 1))
1599                 ips)))
1600       (dolist (server spam-blackhole-servers)
1601         (dolist (ip ips)
1602           (unless (and spam-blackhole-good-server-regex
1603                        ;; match the good-server-regex against the reversed (again) IP string
1604                        (string-match
1605                         spam-blackhole-good-server-regex
1606                         (spam-reverse-ip-string ip)))
1607             (unless matches
1608               (let ((query-string (concat ip "." server)))
1609                 (if spam-use-dig
1610                     (let ((query-result (query-dig query-string)))
1611                       (when query-result
1612                         (gnus-message 6 "(DIG): positive blackhole check '%s'"
1613                                       query-result)
1614                         (push (list ip server query-result)
1615                               matches)))
1616                   ;; else, if not using dig.el
1617                   (when (query-dns query-string)
1618                     (gnus-message 6 "positive blackhole check")
1619                     (push (list ip server (query-dns query-string 'TXT))
1620                           matches)))))))))
1621     (when matches
1622       spam-split-group)))
1623 \f
1624 ;;;; Hashcash.
1625
1626 (condition-case nil
1627     (progn
1628       (require 'hashcash)
1629
1630       (defun spam-check-hashcash ()
1631         "Check the headers for hashcash payments."
1632         (mail-check-payment)))   ;mail-check-payment returns a boolean
1633
1634   (file-error (progn
1635                 (defalias 'mail-check-payment 'ignore)
1636                 (defalias 'spam-check-hashcash 'ignore))))
1637 \f
1638 ;;;; BBDB
1639
1640 ;;; original idea for spam-check-BBDB from Alexander Kotelnikov
1641 ;;; <sacha@giotto.sj.ru>
1642
1643 ;; all this is done inside a condition-case to trap errors
1644
1645 (condition-case nil
1646     (progn
1647       (require 'bbdb)
1648       (require 'bbdb-com)
1649
1650       ;; when the BBDB changes, we want to clear out our cache
1651       (defun spam-clear-cache-BBDB (&rest immaterial)
1652         (spam-clear-cache 'spam-use-BBDB))
1653
1654       (add-hook 'bbdb-change-hook 'spam-clear-cache-BBDB)
1655
1656       (defun spam-enter-ham-BBDB (addresses &optional remove)
1657         "Enter an address into the BBDB; implies ham (non-spam) sender"
1658         (dolist (from addresses)
1659           (when (stringp from)
1660             (let* ((parsed-address (gnus-extract-address-components from))
1661                    (name (or (nth 0 parsed-address) "Ham Sender"))
1662                    (remove-function (if remove
1663                                         'bbdb-delete-record-internal
1664                                       'ignore))
1665                    (net-address (nth 1 parsed-address))
1666                    (record (and net-address
1667                                 (bbdb-search-simple nil net-address))))
1668               (when net-address
1669                 (gnus-message 6 "%s address %s %s BBDB"
1670                               (if remove "Deleting" "Adding")
1671                               from
1672                               (if remove "from" "to"))
1673                 (if record
1674                     (funcall remove-function record)
1675                   (bbdb-create-internal name nil net-address nil nil
1676                                         "ham sender added by spam.el")))))))
1677
1678       (defun spam-BBDB-register-routine (articles &optional unregister)
1679         (let (addresses)
1680           (dolist (article articles)
1681             (when (stringp (spam-fetch-field-from-fast article))
1682               (push (spam-fetch-field-from-fast article) addresses)))
1683           ;; now do the register/unregister action
1684           (spam-enter-ham-BBDB addresses unregister)))
1685
1686       (defun spam-BBDB-unregister-routine (articles)
1687         (spam-BBDB-register-routine articles t))
1688
1689       (defun spam-check-BBDB ()
1690         "Mail from people in the BBDB is classified as ham or non-spam"
1691         (let ((who (message-fetch-field "from"))
1692               (spam-split-group (if spam-split-symbolic-return
1693                                     'spam
1694                                   spam-split-group))
1695               bbdb-cache bbdb-hashtable)
1696           (when spam-cache-lookups
1697             (setq bbdb-cache (gethash 'spam-use-BBDB spam-caches))
1698             (unless bbdb-cache
1699               (setq bbdb-cache
1700                     ;; this is the expanded (bbdb-hashtable) macro
1701                     ;; without the debugging support
1702                     (with-current-buffer (bbdb-buffer)
1703                       (save-excursion
1704                         (save-window-excursion
1705                           (bbdb-records nil t)
1706                           bbdb-hashtable))))
1707               (puthash 'spam-use-BBDB bbdb-cache spam-caches)))
1708           (when who
1709             (setq who (nth 1 (gnus-extract-address-components who)))
1710             (if
1711                 (if spam-cache-lookups
1712                     (symbol-value
1713                      (intern-soft who bbdb-cache))
1714                   (bbdb-search-simple nil who))
1715                 t
1716               (if spam-use-BBDB-exclusive
1717                   spam-split-group
1718                 nil))))))
1719
1720   (file-error (progn
1721                 (defalias 'bbdb-search-simple 'ignore)
1722                 (defalias 'bbdb-records 'ignore)
1723                 (defalias 'bbdb-buffer 'ignore)
1724                 (defalias 'spam-check-BBDB 'ignore)
1725                 (defalias 'spam-BBDB-register-routine 'ignore)
1726                 (defalias 'spam-enter-ham-BBDB 'ignore)
1727                 (defalias 'bbdb-create-internal 'ignore)
1728                 (defalias 'bbdb-delete-record-internal 'ignore)
1729                 (defalias 'bbdb-records 'ignore))))
1730
1731 \f
1732 ;;;; ifile
1733
1734 ;;; check the ifile backend; return nil if the mail was NOT classified
1735 ;;; as spam
1736
1737 (defun spam-get-ifile-database-parameter ()
1738   "Get the command-line parameter for ifile's database from
1739   spam-ifile-database-path."
1740   (if spam-ifile-database-path
1741       (format "--db-file=%s" spam-ifile-database-path)
1742     nil))
1743
1744 (defun spam-check-ifile ()
1745   "Check the ifile backend for the classification of this message."
1746   (let ((article-buffer-name (buffer-name))
1747         (spam-split-group (if spam-split-symbolic-return
1748                               'spam
1749                             spam-split-group))
1750         category return)
1751     (with-temp-buffer
1752       (let ((temp-buffer-name (buffer-name))
1753             (db-param (spam-get-ifile-database-parameter)))
1754         (save-excursion
1755           (set-buffer article-buffer-name)
1756           (apply 'call-process-region
1757                  (point-min) (point-max) spam-ifile-path
1758                  nil temp-buffer-name nil "-c"
1759                  (if db-param `(,db-param "-q") `("-q"))))
1760         ;; check the return now (we're back in the temp buffer)
1761         (goto-char (point-min))
1762         (if (not (eobp))
1763             (setq category (buffer-substring (point) (point-at-eol))))
1764         (when (not (zerop (length category))) ; we need a category here
1765           (if spam-ifile-all-categories
1766               (setq return category)
1767             ;; else, if spam-ifile-all-categories is not set...
1768             (when (string-equal spam-ifile-spam-category category)
1769               (setq return spam-split-group)))))) ; note return is nil otherwise
1770     return))
1771
1772 (defun spam-ifile-register-with-ifile (articles category &optional unregister)
1773   "Register an article, given as a string, with a category.
1774 Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
1775   (let ((category (or category gnus-newsgroup-name))
1776         (add-or-delete-option (if unregister "-d" "-i"))
1777         (db (spam-get-ifile-database-parameter))
1778         parameters)
1779     (with-temp-buffer
1780       (dolist (article articles)
1781         (let ((article-string (spam-get-article-as-string article)))
1782           (when (stringp article-string)
1783             (insert article-string))))
1784       (apply 'call-process-region
1785              (point-min) (point-max) spam-ifile-path
1786              nil nil nil
1787              add-or-delete-option category
1788              (if db `(,db "-h") `("-h"))))))
1789
1790 (defun spam-ifile-register-spam-routine (articles &optional unregister)
1791   (spam-ifile-register-with-ifile articles spam-ifile-spam-category unregister))
1792
1793 (defun spam-ifile-unregister-spam-routine (articles)
1794   (spam-ifile-register-spam-routine articles t))
1795
1796 (defun spam-ifile-register-ham-routine (articles &optional unregister)
1797   (spam-ifile-register-with-ifile articles spam-ifile-ham-category unregister))
1798
1799 (defun spam-ifile-unregister-ham-routine (articles)
1800   (spam-ifile-register-ham-routine articles t))
1801
1802 \f
1803 ;;;; spam-stat
1804
1805 (condition-case nil
1806     (progn
1807       (let ((spam-stat-install-hooks nil))
1808         (require 'spam-stat))
1809
1810       (defun spam-check-stat ()
1811         "Check the spam-stat backend for the classification of this message"
1812         (let ((spam-split-group (if spam-split-symbolic-return
1813                                     'spam
1814                                   spam-split-group))
1815               (spam-stat-split-fancy-spam-group spam-split-group) ; override
1816               (spam-stat-buffer (buffer-name)) ; stat the current buffer
1817               category return)
1818           (spam-stat-split-fancy)))
1819
1820       (defun spam-stat-register-spam-routine (articles &optional unregister)
1821         (dolist (article articles)
1822           (let ((article-string (spam-get-article-as-string article)))
1823             (with-temp-buffer
1824               (insert article-string)
1825               (if unregister
1826                   (spam-stat-buffer-change-to-non-spam)
1827               (spam-stat-buffer-is-spam))))))
1828
1829       (defun spam-stat-unregister-spam-routine (articles)
1830         (spam-stat-register-spam-routine articles t))
1831
1832       (defun spam-stat-register-ham-routine (articles &optional unregister)
1833         (dolist (article articles)
1834           (let ((article-string (spam-get-article-as-string article)))
1835             (with-temp-buffer
1836               (insert article-string)
1837               (if unregister
1838                   (spam-stat-buffer-change-to-spam)
1839               (spam-stat-buffer-is-non-spam))))))
1840
1841       (defun spam-stat-unregister-ham-routine (articles)
1842         (spam-stat-register-ham-routine articles t))
1843
1844       (defun spam-maybe-spam-stat-load ()
1845         (when spam-use-stat (spam-stat-load)))
1846
1847       (defun spam-maybe-spam-stat-save ()
1848         (when spam-use-stat (spam-stat-save))))
1849
1850   (file-error (progn
1851                 (defalias 'spam-stat-load 'ignore)
1852                 (defalias 'spam-stat-save 'ignore)
1853                 (defalias 'spam-maybe-spam-stat-load 'ignore)
1854                 (defalias 'spam-maybe-spam-stat-save 'ignore)
1855                 (defalias 'spam-stat-register-ham-routine 'ignore)
1856                 (defalias 'spam-stat-unregister-ham-routine 'ignore)
1857                 (defalias 'spam-stat-register-spam-routine 'ignore)
1858                 (defalias 'spam-stat-unregister-spam-routine 'ignore)
1859                 (defalias 'spam-stat-buffer-is-spam 'ignore)
1860                 (defalias 'spam-stat-buffer-change-to-spam 'ignore)
1861                 (defalias 'spam-stat-buffer-is-non-spam 'ignore)
1862                 (defalias 'spam-stat-buffer-change-to-non-spam 'ignore)
1863                 (defalias 'spam-stat-split-fancy 'ignore)
1864                 (defalias 'spam-check-stat 'ignore))))
1865
1866 \f
1867
1868 ;;;; Blacklists and whitelists.
1869
1870 (defvar spam-whitelist-cache nil)
1871 (defvar spam-blacklist-cache nil)
1872
1873 (defun spam-kill-whole-line ()
1874   (beginning-of-line)
1875   (let ((kill-whole-line t))
1876     (kill-line)))
1877
1878 ;;; address can be a list, too
1879 (defun spam-enter-whitelist (address &optional remove)
1880   "Enter ADDRESS (list or single) into the whitelist.
1881 With a non-nil REMOVE, remove them."
1882   (interactive "sAddress: ")
1883   (spam-enter-list address spam-whitelist remove)
1884   (setq spam-whitelist-cache nil)
1885   (spam-clear-cache 'spam-use-whitelist))
1886
1887 ;;; address can be a list, too
1888 (defun spam-enter-blacklist (address &optional remove)
1889   "Enter ADDRESS (list or single) into the blacklist.
1890 With a non-nil REMOVE, remove them."
1891   (interactive "sAddress: ")
1892   (spam-enter-list address spam-blacklist remove)
1893   (setq spam-blacklist-cache nil)
1894   (spam-clear-cache 'spam-use-whitelist))
1895
1896 (defun spam-enter-list (addresses file &optional remove)
1897   "Enter ADDRESSES into the given FILE.
1898 Either the whitelist or the blacklist files can be used.  With
1899 REMOVE not nil, remove the ADDRESSES."
1900   (if (stringp addresses)
1901       (spam-enter-list (list addresses) file remove)
1902     ;; else, we have a list of addresses here
1903     (unless (file-exists-p (file-name-directory file))
1904       (make-directory (file-name-directory file) t))
1905     (save-excursion
1906       (set-buffer
1907        (find-file-noselect file))
1908       (dolist (a addresses)
1909         (when (stringp a)
1910           (goto-char (point-min))
1911           (if (re-search-forward (regexp-quote a) nil t)
1912               ;; found the address
1913               (when remove
1914                 (spam-kill-whole-line))
1915             ;; else, the address was not found
1916             (unless remove
1917               (goto-char (point-max))
1918               (unless (bobp)
1919                 (insert "\n"))
1920               (insert a "\n")))))
1921       (save-buffer))))
1922
1923 (defun spam-filelist-build-cache (type)
1924   (let ((cache (if (eq type 'spam-use-blacklist)
1925                    spam-blacklist-cache
1926                  spam-whitelist-cache))
1927         parsed-cache)
1928     (unless (gethash type spam-caches)
1929       (while cache
1930         (let ((address (pop cache)))
1931           (unless (zerop (length address)) ; 0 for a nil address too
1932             (setq address (regexp-quote address))
1933             ;; fix regexp-quote's treatment of user-intended regexes
1934             (while (string-match "\\\\\\*" address)
1935               (setq address (replace-match ".*" t t address))))
1936           (push address parsed-cache)))
1937       (puthash type parsed-cache spam-caches))))
1938
1939 (defun spam-filelist-check-cache (type from)
1940   (when (stringp from)
1941     (spam-filelist-build-cache type)
1942     (let (found)
1943       (dolist (address (gethash type spam-caches))
1944         (when (and address (string-match address from))
1945           (setq found t)
1946           (return)))
1947       found)))
1948
1949 ;;; returns t if the sender is in the whitelist, nil or
1950 ;;; spam-split-group otherwise
1951 (defun spam-check-whitelist ()
1952   ;; FIXME!  Should it detect when file timestamps change?
1953   (let ((spam-split-group (if spam-split-symbolic-return
1954                               'spam
1955                             spam-split-group)))
1956     (unless spam-whitelist-cache
1957       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
1958     (if (spam-from-listed-p 'spam-use-whitelist)
1959         t
1960       (if spam-use-whitelist-exclusive
1961           spam-split-group
1962         nil))))
1963
1964 (defun spam-check-blacklist ()
1965   ;; FIXME!  Should it detect when file timestamps change?
1966   (let ((spam-split-group (if spam-split-symbolic-return
1967                               'spam
1968                             spam-split-group)))
1969     (unless spam-blacklist-cache
1970       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
1971     (and (spam-from-listed-p 'spam-use-blacklist) spam-split-group)))
1972
1973 (defun spam-parse-list (file)
1974   (when (file-readable-p file)
1975     (let (contents address)
1976       (with-temp-buffer
1977         (insert-file-contents file)
1978         (while (not (eobp))
1979           (setq address (buffer-substring (point) (point-at-eol)))
1980           (forward-line 1)
1981           ;; insert the e-mail address if detected, otherwise the raw data
1982           (unless (zerop (length address))
1983             (let ((pure-address (nth 1 (gnus-extract-address-components address))))
1984               (push (or pure-address address) contents)))))
1985       (nreverse contents))))
1986
1987 (defun spam-from-listed-p (type)
1988   (let ((from (message-fetch-field "from"))
1989         found)
1990     (spam-filelist-check-cache type from)))
1991
1992 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
1993   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
1994         (declassification (if blacklist 'ham 'spam))
1995         (enter-function
1996          (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
1997         (remove-function
1998          (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
1999         from addresses unregister-list article-unregister-list)
2000     (dolist (article articles)
2001       (let ((from (spam-fetch-field-from-fast article))
2002             (id (spam-fetch-field-message-id-fast article))
2003             sender-ignored)
2004         (when (stringp from)
2005           (dolist (ignore-regex spam-blacklist-ignored-regexes)
2006             (when (and (not sender-ignored)
2007                        (stringp ignore-regex)
2008                        (string-match ignore-regex from))
2009               (setq sender-ignored t)))
2010           ;; remember the messages we need to unregister, unless remove is set
2011           (when (and
2012                  (null unregister)
2013                  (spam-log-unregistration-needed-p
2014                   id 'process declassification de-symbol))
2015             (push article article-unregister-list)
2016             (push from unregister-list))
2017           (unless sender-ignored
2018             (push from addresses)))))
2019
2020     (if unregister
2021         (funcall enter-function addresses t) ; unregister all these addresses
2022       ;; else, register normally and unregister what we need to
2023       (funcall remove-function unregister-list t)
2024       (dolist (article article-unregister-list)
2025         (spam-log-undo-registration
2026          (spam-fetch-field-message-id-fast article)
2027          'process
2028          declassification
2029          de-symbol))
2030       (funcall enter-function addresses nil))))
2031
2032 (defun spam-blacklist-unregister-routine (articles)
2033   (spam-blacklist-register-routine articles t))
2034
2035 (defun spam-blacklist-register-routine (articles &optional unregister)
2036   (spam-filelist-register-routine articles t unregister))
2037
2038 (defun spam-whitelist-unregister-routine (articles)
2039   (spam-whitelist-register-routine articles t))
2040
2041 (defun spam-whitelist-register-routine (articles &optional unregister)
2042   (spam-filelist-register-routine articles nil unregister))
2043
2044 \f
2045 ;;;; Spam-report glue
2046 (defun spam-report-gmane-register-routine (articles)
2047   (when articles
2048     (apply 'spam-report-gmane articles)))
2049
2050 \f
2051 ;;;; Bogofilter
2052 (defun spam-check-bogofilter-headers (&optional score)
2053   (let ((header (message-fetch-field spam-bogofilter-header))
2054         (spam-split-group (if spam-split-symbolic-return
2055                               'spam
2056                             spam-split-group)))
2057     (when header                        ; return nil when no header
2058       (if score                         ; scoring mode
2059           (if (string-match "spamicity=\\([0-9.]+\\)" header)
2060               (match-string 1 header)
2061             "0")
2062         ;; spam detection mode
2063         (when (string-match spam-bogofilter-bogosity-positive-spam-header
2064                             header)
2065           spam-split-group)))))
2066
2067 ;; return something sensible if the score can't be determined
2068 (defun spam-bogofilter-score (&optional recheck)
2069   "Get the Bogofilter spamicity score"
2070   (interactive "P")
2071   (save-window-excursion
2072     (gnus-summary-show-article t)
2073     (set-buffer gnus-article-buffer)
2074     (let ((score (or (unless recheck
2075                        (spam-check-bogofilter-headers t))
2076                      (spam-check-bogofilter t))))
2077       (gnus-summary-show-article)
2078       (message "Spamicity score %s" score)
2079       (or score "0"))))
2080
2081 (defun spam-check-bogofilter (&optional score)
2082   "Check the Bogofilter backend for the classification of this message"
2083   (let ((article-buffer-name (buffer-name))
2084         (db spam-bogofilter-database-directory)
2085         return)
2086     (with-temp-buffer
2087       (let ((temp-buffer-name (buffer-name)))
2088         (save-excursion
2089           (set-buffer article-buffer-name)
2090           (apply 'call-process-region
2091                  (point-min) (point-max)
2092                  spam-bogofilter-path
2093                  nil temp-buffer-name nil
2094                  (if db `("-d" ,db "-v") `("-v"))))
2095         (setq return (spam-check-bogofilter-headers score))))
2096     return))
2097
2098 (defun spam-bogofilter-register-with-bogofilter (articles
2099                                                  spam
2100                                                  &optional unregister)
2101   "Register an article, given as a string, as spam or non-spam."
2102   (dolist (article articles)
2103     (let ((article-string (spam-get-article-as-string article))
2104           (db spam-bogofilter-database-directory)
2105           (switch (if unregister
2106                       (if spam
2107                           spam-bogofilter-spam-strong-switch
2108                         spam-bogofilter-ham-strong-switch)
2109                     (if spam
2110                         spam-bogofilter-spam-switch
2111                       spam-bogofilter-ham-switch))))
2112       (when (stringp article-string)
2113         (with-temp-buffer
2114           (insert article-string)
2115
2116           (apply 'call-process-region
2117                  (point-min) (point-max)
2118                  spam-bogofilter-path
2119                  nil nil nil switch
2120                  (if db `("-d" ,db "-v") `("-v"))))))))
2121
2122 (defun spam-bogofilter-register-spam-routine (articles &optional unregister)
2123   (spam-bogofilter-register-with-bogofilter articles t unregister))
2124
2125 (defun spam-bogofilter-unregister-spam-routine (articles)
2126   (spam-bogofilter-register-spam-routine articles t))
2127
2128 (defun spam-bogofilter-register-ham-routine (articles &optional unregister)
2129   (spam-bogofilter-register-with-bogofilter articles nil unregister))
2130
2131 (defun spam-bogofilter-unregister-ham-routine (articles)
2132   (spam-bogofilter-register-ham-routine articles t))
2133
2134
2135 \f
2136 ;;;; spamoracle
2137 (defun spam-check-spamoracle ()
2138   "Run spamoracle on an article to determine whether it's spam."
2139   (let ((article-buffer-name (buffer-name))
2140         (spam-split-group (if spam-split-symbolic-return
2141                               'spam
2142                             spam-split-group)))
2143     (with-temp-buffer
2144       (let ((temp-buffer-name (buffer-name)))
2145         (save-excursion
2146           (set-buffer article-buffer-name)
2147           (let ((status
2148                  (apply 'call-process-region
2149                         (point-min) (point-max)
2150                         spam-spamoracle-binary
2151                         nil temp-buffer-name nil
2152                         (if spam-spamoracle-database
2153                             `("-f" ,spam-spamoracle-database "mark")
2154                           '("mark")))))
2155             (if (eq 0 status)
2156                 (progn
2157                   (set-buffer temp-buffer-name)
2158                   (goto-char (point-min))
2159                   (when (re-search-forward "^X-Spam: yes;" nil t)
2160                     spam-split-group))
2161               (error "Error running spamoracle: %s" status))))))))
2162
2163 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
2164   "Run spamoracle in training mode."
2165   (with-temp-buffer
2166     (let ((temp-buffer-name (buffer-name)))
2167       (save-excursion
2168         (goto-char (point-min))
2169         (dolist (article articles)
2170           (insert (spam-get-article-as-string article)))
2171         (let* ((arg (if (spam-xor unregister article-is-spam-p)
2172                         "-spam"
2173                       "-good"))
2174                (status
2175                 (apply 'call-process-region
2176                        (point-min) (point-max)
2177                        spam-spamoracle-binary
2178                        nil temp-buffer-name nil
2179                        (if spam-spamoracle-database
2180                            `("-f" ,spam-spamoracle-database
2181                              "add" ,arg)
2182                          `("add" ,arg)))))
2183           (unless (eq 0 status)
2184             (error "Error running spamoracle: %s" status)))))))
2185
2186 (defun spam-spamoracle-learn-ham (articles &optional unregister)
2187   (spam-spamoracle-learn articles nil unregister))
2188
2189 (defun spam-spamoracle-unlearn-ham (articles &optional unregister)
2190   (spam-spamoracle-learn-ham articles t))
2191
2192 (defun spam-spamoracle-learn-spam (articles &optional unregister)
2193   (spam-spamoracle-learn articles t unregister))
2194
2195 (defun spam-spamoracle-unlearn-spam (articles &optional unregister)
2196   (spam-spamoracle-learn-spam articles t))
2197
2198 \f
2199 ;;;; SpamAssassin
2200 ;;; based mostly on the bogofilter code
2201 (defun spam-check-spamassassin-headers (&optional score)
2202   "Check the SpamAssassin headers for the classification of this message."
2203   (if score                             ; scoring mode
2204       (let ((header (message-fetch-field spam-spamassassin-spam-status-header)))
2205         (when header
2206           (if (string-match "hits=\\(-?[0-9.]+\\)" header)
2207               (match-string 1 header)
2208             "0")))
2209     ;; spam detection mode
2210     (let ((header (message-fetch-field spam-spamassassin-spam-flag-header))
2211           (spam-split-group (if spam-split-symbolic-return
2212                                  'spam
2213                                spam-split-group)))
2214           (when header                  ; return nil when no header
2215             (when (string-match spam-spamassassin-positive-spam-flag-header
2216                                 header)
2217               spam-split-group)))))
2218
2219 (defun spam-check-spamassassin (&optional score)
2220   "Check the SpamAssassin backend for the classification of this message."
2221   (let ((article-buffer-name (buffer-name)))
2222     (with-temp-buffer
2223       (let ((temp-buffer-name (buffer-name)))
2224         (save-excursion
2225           (set-buffer article-buffer-name)
2226           (apply 'call-process-region
2227                  (point-min) (point-max) spam-spamassassin-path
2228                  nil temp-buffer-name nil spam-spamassassin-arguments))
2229         ;; check the return now (we're back in the temp buffer)
2230         (goto-char (point-min))
2231         (spam-check-spamassassin-headers score)))))
2232
2233 ;; return something sensible if the score can't be determined
2234 (defun spam-spamassassin-score (&optional recheck)
2235   "Get the SpamAssassin score"
2236   (interactive "P")
2237   (save-window-excursion
2238     (gnus-summary-show-article t)
2239     (set-buffer gnus-article-buffer)
2240     (let ((score (or (unless recheck
2241                        (spam-check-spamassassin-headers t))
2242                      (spam-check-spamassassin t))))
2243       (gnus-summary-show-article)
2244       (message "SpamAssassin score %s" score)
2245       (or score "0"))))
2246
2247 (defun spam-spamassassin-register-with-sa-learn (articles spam
2248                                                  &optional unregister)
2249   "Register articles with spamassassin's sa-learn as spam or non-spam."
2250   (if articles
2251       (let ((action (if unregister spam-sa-learn-unregister-switch
2252                       (if spam spam-sa-learn-spam-switch
2253                         spam-sa-learn-ham-switch)))
2254             (summary-buffer-name (buffer-name)))
2255         (with-temp-buffer
2256           ;; group the articles into mbox format
2257           (dolist (article articles)
2258             (let (article-string)
2259               (save-excursion
2260                 (set-buffer summary-buffer-name)
2261                 (setq article-string (spam-get-article-as-string article)))
2262               (when (stringp article-string)
2263                 (insert "From \n") ; mbox separator (sa-learn only checks the
2264                                    ; first five chars, so we can get away with
2265                                    ; a bogus line))
2266                 (insert article-string)
2267                 (insert "\n"))))
2268           ;; call sa-learn on all messages at the same time
2269           (apply 'call-process-region
2270                  (point-min) (point-max)
2271                  spam-sa-learn-path
2272                  nil nil nil "--mbox"
2273                  (if spam-sa-learn-rebuild
2274                      (list action)
2275                    `("--no-rebuild" ,action)))))))
2276
2277 (defun spam-spamassassin-register-spam-routine (articles &optional unregister)
2278   (spam-spamassassin-register-with-sa-learn articles t unregister))
2279
2280 (defun spam-spamassassin-register-ham-routine (articles &optional unregister)
2281   (spam-spamassassin-register-with-sa-learn articles nil unregister))
2282
2283 (defun spam-spamassassin-unregister-spam-routine (articles)
2284   (spam-spamassassin-register-with-sa-learn articles t t))
2285
2286 (defun spam-spamassassin-unregister-ham-routine (articles)
2287   (spam-spamassassin-register-with-sa-learn articles nil t))
2288
2289 \f
2290 ;;;; Bsfilter
2291 ;;; based mostly on the bogofilter code
2292 (defun spam-check-bsfilter-headers (&optional score)
2293   (if score
2294       (or (nnmail-fetch-field spam-bsfilter-probability-header)
2295           "0")
2296     (let ((header (nnmail-fetch-field spam-bsfilter-header))
2297           (spam-split-group (if spam-split-symbolic-return
2298                                 'spam
2299                               spam-split-group)))
2300       (when header ; return nil when no header
2301         (when (string-match "YES" header)
2302           spam-split-group)))))
2303
2304 ;; return something sensible if the score can't be determined
2305 (defun spam-bsfilter-score (&optional recheck)
2306   "Get the Bsfilter spamicity score"
2307   (interactive "P")
2308   (save-window-excursion
2309     (gnus-summary-show-article t)
2310     (set-buffer gnus-article-buffer)
2311     (let ((score (or (unless recheck
2312                        (spam-check-bsfilter-headers t))
2313                      (spam-check-bsfilter t))))
2314       (gnus-summary-show-article)
2315       (message "Spamicity score %s" score)
2316       (or score "0"))))
2317
2318 (defun spam-check-bsfilter (&optional score)
2319   "Check the Bsfilter backend for the classification of this message"
2320   (let ((article-buffer-name (buffer-name))
2321         (dir spam-bsfilter-database-directory)
2322         return)
2323     (with-temp-buffer
2324       (let ((temp-buffer-name (buffer-name)))
2325         (save-excursion
2326           (set-buffer article-buffer-name)
2327           (apply 'call-process-region
2328                  (point-min) (point-max)
2329                  spam-bsfilter-path
2330                  nil temp-buffer-name nil
2331                  "--pipe"
2332                  "--insert-flag"
2333                  "--insert-probability"
2334                  (when dir
2335                    (list "--homedir" dir))))
2336         (setq return (spam-check-bsfilter-headers score))))
2337     return))
2338
2339 (defun spam-bsfilter-register-with-bsfilter (articles
2340                                              spam
2341                                              &optional unregister)
2342   "Register an article, given as a string, as spam or non-spam."
2343   (dolist (article articles)
2344     (let ((article-string (spam-get-article-as-string article))
2345           (switch (if unregister
2346                       (if spam
2347                           spam-bsfilter-spam-strong-switch
2348                         spam-bsfilter-ham-strong-switch)
2349                     (if spam
2350                         spam-bsfilter-spam-switch
2351                       spam-bsfilter-ham-switch))))
2352       (when (stringp article-string)
2353         (with-temp-buffer
2354           (insert article-string)
2355           (apply 'call-process-region
2356                  (point-min) (point-max)
2357                  spam-bsfilter-path
2358                  nil nil nil switch
2359                  "--update"
2360                  (when spam-bsfilter-database-directory
2361                    (list "--homedir"
2362                          spam-bsfilter-database-directory))))))))
2363
2364 (defun spam-bsfilter-register-spam-routine (articles &optional unregister)
2365   (spam-bsfilter-register-with-bsfilter articles t unregister))
2366
2367 (defun spam-bsfilter-unregister-spam-routine (articles)
2368   (spam-bsfilter-register-spam-routine articles t))
2369
2370 (defun spam-bsfilter-register-ham-routine (articles &optional unregister)
2371   (spam-bsfilter-register-with-bsfilter articles nil unregister))
2372
2373 (defun spam-bsfilter-unregister-ham-routine (articles)
2374   (spam-bsfilter-register-ham-routine articles t))
2375
2376 \f
2377 ;;;; Hooks
2378
2379 ;;;###autoload
2380 (defun spam-initialize (&rest symbols)
2381   "Install the spam.el hooks and do other initialization.
2382 When SYMBOLS is given, set those variables to t.  This is so you
2383 can call spam-initialize before you set spam-use-* variables on
2384 explicitly, and matters only if you need the extra headers
2385 installed through spam-necessary-extra-headers."
2386   (interactive)
2387
2388   (dolist (var symbols)
2389     (set var t))
2390
2391   (dolist (header (spam-necessary-extra-headers))
2392     (add-to-list 'nnmail-extra-headers header)
2393     (add-to-list 'gnus-extra-headers header))
2394
2395   (setq spam-install-hooks t)
2396   ;; TODO: How do we redo this every time spam-face is customized?
2397   (push '((eq mark gnus-spam-mark) . spam-face)
2398         gnus-summary-highlight)
2399   ;; Add hooks for loading and saving the spam stats
2400   (add-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2401   (add-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2402   (add-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2403   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2404   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2405   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2406   (add-hook 'gnus-summary-prepared-hook 'spam-find-spam))
2407
2408 (defun spam-unload-hook ()
2409   "Uninstall the spam.el hooks"
2410   (interactive)
2411   (remove-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2412   (remove-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2413   (remove-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2414   (remove-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2415   (remove-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2416   (remove-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2417   (remove-hook 'gnus-summary-prepare-hook 'spam-find-spam))
2418
2419 (when spam-install-hooks
2420   (spam-initialize))
2421
2422 (provide 'spam)
2423
2424 ;;; spam.el ends here