a9099ace2ac6a971e5bb5c0fbcbb27f8cfc4960c
[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       (when (< 0 (length (spam-list-articles
865                           gnus-newsgroup-articles
866                           'spam)))
867         (gnus-message 6 "Marking spam as expired and moving it to %s"
868                       (gnus-parameter-spam-process-destination 
869                        gnus-newsgroup-name))
870         (spam-mark-spam-as-expired-and-move-routine
871          (gnus-parameter-spam-process-destination gnus-newsgroup-name))))
872
873     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
874     ;; expire spam, in case the above did not expire them
875     (when (< 0 (length (spam-list-articles
876                         gnus-newsgroup-articles
877                         'spam)))
878       (gnus-message 6 "Marking spam as expired without moving it")
879       (spam-mark-spam-as-expired-and-move-routine nil))
880
881     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
882               (and (spam-group-spam-contents-p gnus-newsgroup-name)
883                    spam-process-ham-in-spam-groups)
884               spam-process-ham-in-nonham-groups)
885       ;; find all the ham processors applicable to this group
886       (dolist (processor-param spam-list-of-processors)
887         (let ((processor (nth 0 processor-param))
888               (classification (nth 1 processor-param))
889               (check (nth 2 processor-param)))
890           (when (and (eq 'ham classification)
891                      (spam-group-processor-p gnus-newsgroup-name processor))
892             (spam-register-routine classification check)))))
893
894     (when (< 0 (length (spam-list-articles
895                         gnus-newsgroup-articles
896                         'ham)))
897       (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
898         (gnus-message 6 "Copying ham")
899         (spam-ham-copy-routine
900          (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
901
902       ;; now move all ham articles out of spam groups
903       (when (spam-group-spam-contents-p gnus-newsgroup-name)
904         (gnus-message 6 "Moving ham messages from spam group")
905         (spam-ham-move-routine
906          (gnus-parameter-ham-process-destination gnus-newsgroup-name)))))
907
908   (setq spam-old-ham-articles nil)
909   (setq spam-old-spam-articles nil))
910
911 (defun spam-set-difference (list1 list2)
912   "Return a set difference of LIST1 and LIST2.  
913 When either list is nil, the other is returned."
914   (if (and list1 list2)
915       ;; we have two non-nil lists
916       (progn
917         (dolist (item (append list1 list2))
918           (when (and (memq item list1) (memq item list2))
919             (setq list1 (delq item list1))
920             (setq list2 (delq item list2))))
921         (append list1 list2))
922     ;; if either of the lists was nil, return the other one
923     (if list1 list1 list2)))
924
925 (defun spam-mark-junk-as-spam-routine ()
926   ;; check the global list of group names spam-junk-mailgroups and the
927   ;; group parameters
928   (when (spam-group-spam-contents-p gnus-newsgroup-name)
929     (gnus-message 6 "Marking %s articles as spam"
930                   (if spam-mark-only-unseen-as-spam
931                       "unseen"
932                     "unread"))
933     (let ((articles (if spam-mark-only-unseen-as-spam
934                         gnus-newsgroup-unseen
935                       gnus-newsgroup-unreads)))
936       (dolist (article articles)
937         (gnus-summary-mark-article article gnus-spam-mark)))))
938
939 (defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
940   (if (and (car-safe groups) (listp (car-safe groups)))
941       (apply 'spam-mark-spam-as-expired-and-move-routine (car groups))
942     (gnus-summary-kill-process-mark)
943     (let ((articles gnus-newsgroup-articles)
944           (backend-supports-deletions
945            (gnus-check-backend-function
946             'request-move-article gnus-newsgroup-name))
947           article tomove deletep)
948       (dolist (article articles)
949         (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
950           (gnus-summary-mark-article article gnus-expirable-mark)
951           (push article tomove)))
952
953       ;; now do the actual copies
954       (dolist (group groups)
955         (when (and tomove
956                    (stringp group))
957           (dolist (article tomove)
958             (gnus-summary-set-process-mark article))
959           (when tomove
960             (if (or (not backend-supports-deletions)
961                     (> (length groups) 1))
962                 (progn
963                   (gnus-summary-copy-article nil group)
964                   (setq deletep t))
965               (gnus-summary-move-article nil group)))))
966
967       ;; now delete the articles, if there was a copy done, and the
968       ;; backend allows it
969       (when (and deletep backend-supports-deletions)
970         (dolist (article tomove)
971           (gnus-summary-set-process-mark article))
972         (when tomove
973           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
974             (gnus-summary-delete-article nil))))
975
976       (gnus-summary-yank-process-mark))))
977
978 (defun spam-ham-copy-or-move-routine (copy groups)
979   (gnus-summary-kill-process-mark)
980   (let ((todo (spam-list-articles gnus-newsgroup-articles 'ham))
981         (backend-supports-deletions
982          (gnus-check-backend-function
983           'request-move-article gnus-newsgroup-name))
984         (respool-method (gnus-find-method-for-group gnus-newsgroup-name))
985         article mark todo deletep respool)
986
987     (when (member 'respool groups)
988       (setq respool t)                  ; boolean for later
989       (setq groups '("fake"))) ; when respooling, groups are dynamic so fake it
990
991     ;; now do the actual move
992     (dolist (group groups)
993       (when (and todo (stringp group))
994         (dolist (article todo)
995           (when spam-mark-ham-unread-before-move-from-spam-group
996             (gnus-summary-mark-article article gnus-unread-mark))
997           (gnus-summary-set-process-mark article))
998
999         (if respool                        ; respooling is with a "fake" group
1000             (let ((spam-split-disabled
1001                    (or spam-split-disabled
1002                        spam-disable-spam-split-during-ham-respool)))
1003               (gnus-summary-respool-article nil respool-method))
1004           (if (or (not backend-supports-deletions) ; else, we are not respooling
1005                   (> (length groups) 1))
1006               (progn                ; if copying, copy and set deletep
1007                 (gnus-summary-copy-article nil group)
1008                 (setq deletep t))
1009             (gnus-summary-move-article nil group))))) ; else move articles
1010
1011     ;; now delete the articles, unless a) copy is t, and there was a copy done
1012     ;;                                 b) a move was done to a single group
1013     ;;                                 c) backend-supports-deletions is nil
1014     (unless copy
1015       (when (and deletep backend-supports-deletions)
1016         (dolist (article todo)
1017           (gnus-summary-set-process-mark article))
1018         (when todo
1019           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
1020             (gnus-summary-delete-article nil))))))
1021
1022   (gnus-summary-yank-process-mark))
1023
1024 (defun spam-ham-copy-routine (&rest groups)
1025   (if (and (car-safe groups) (listp (car-safe groups)))
1026       (apply 'spam-ham-copy-routine (car groups))
1027     (spam-ham-copy-or-move-routine t groups)))
1028
1029 (defun spam-ham-move-routine (&rest groups)
1030   (if (and (car-safe groups) (listp (car-safe groups)))
1031       (apply 'spam-ham-move-routine (car groups))
1032     (spam-ham-copy-or-move-routine nil groups)))
1033
1034 (defun spam-get-article-as-string (article)
1035   (when (numberp article)
1036     (with-temp-buffer
1037       (gnus-request-article-this-buffer
1038        article
1039        gnus-newsgroup-name)
1040       (buffer-string))))
1041
1042 ;; disabled for now
1043 ;; (defun spam-get-article-as-filename (article)
1044 ;;   (let ((article-filename))
1045 ;;     (when (numberp article)
1046 ;;       (nnml-possibly-change-directory
1047 ;;        (gnus-group-real-name gnus-newsgroup-name))
1048 ;;       (setq article-filename (expand-file-name
1049 ;;                              (int-to-string article) nnml-current-directory)))
1050 ;;     (if (file-exists-p article-filename)
1051 ;;      article-filename
1052 ;;       nil)))
1053
1054 (defun spam-fetch-field-fast (article field &optional prepared-data-header)
1055   "Fetch a field quickly, using the internal gnus-data-list function"
1056   (when (numberp article)
1057     (let* ((data-header (or prepared-data-header
1058                             (spam-fetch-article-header article))))
1059       (if (arrayp data-header)
1060         (cond
1061          ((equal field 'from)
1062           (mail-header-from data-header))
1063          ((equal field 'message-id)
1064           (mail-header-message-id data-header))
1065          ((equal field 'subject)
1066           (mail-header-subject data-header))
1067          ((equal field 'references)
1068           (mail-header-references data-header))
1069          ((equal field 'date)
1070           (mail-header-date data-header))
1071          ((equal field 'xref)
1072           (mail-header-xref data-header))
1073          ((equal field 'extra)
1074           (mail-header-extra data-header))
1075          (t
1076           nil))
1077         (gnus-message 6 "Article %d has a nil data header" article)))))
1078
1079 (defun spam-fetch-field-from-fast (article &optional prepared-data-header)
1080   (spam-fetch-field-fast article 'from prepared-data-header))
1081
1082 (defun spam-fetch-field-subject-fast (article &optional prepared-data-header)
1083   (spam-fetch-field-fast article 'subject prepared-data-header))
1084
1085 (defun spam-fetch-field-message-id-fast (article &optional prepared-data-header)
1086   (spam-fetch-field-fast article 'message-id prepared-data-header))
1087
1088 (defun spam-generate-fake-headers (article)
1089   (let ((dh (spam-fetch-article-header article)))
1090     (if dh
1091         (concat
1092          (format 
1093           (concat "From: %s\nSubject: %s\nMessage-ID: %s\n"
1094                   "Date: %s\nReferences: %s\nXref: %s\n")
1095           (spam-fetch-field-fast article 'from dh)
1096           (spam-fetch-field-fast article 'subject dh)
1097           (spam-fetch-field-fast article 'message-id dh)
1098           (spam-fetch-field-fast article 'date dh)
1099           (spam-fetch-field-fast article 'references dh)
1100           (spam-fetch-field-fast article 'xref dh))
1101          (when (spam-fetch-field-fast article 'extra dh)
1102            (format "%s\n" (spam-fetch-field-fast article 'extra dh))))
1103       (gnus-message
1104        5
1105        "spam-generate-fake-headers: article %d didn't have a valid header"
1106        article))))
1107
1108 (defun spam-fetch-article-header (article)
1109   (save-excursion
1110     (set-buffer gnus-summary-buffer)
1111     (gnus-read-header article)
1112     (nth 3 (assq article gnus-newsgroup-data))))
1113
1114 \f
1115 ;;;; Spam determination.
1116
1117 (defvar spam-list-of-checks
1118   '((spam-use-blacklist                 .       spam-check-blacklist)
1119     (spam-use-regex-headers             .       spam-check-regex-headers)
1120     (spam-use-gmane-xref                .       spam-check-gmane-xref)
1121     (spam-use-regex-body                .       spam-check-regex-body)
1122     (spam-use-whitelist                 .       spam-check-whitelist)
1123     (spam-use-BBDB                      .       spam-check-BBDB)
1124     (spam-use-BBDB-exclusive            .       spam-check-BBDB)
1125     (spam-use-ifile                     .       spam-check-ifile)
1126     (spam-use-spamoracle                .       spam-check-spamoracle)
1127     (spam-use-stat                      .       spam-check-stat)
1128     (spam-use-blackholes                .       spam-check-blackholes)
1129     (spam-use-hashcash                  .       spam-check-hashcash)
1130     (spam-use-spamassassin-headers      .       spam-check-spamassassin-headers)
1131     (spam-use-spamassassin              .       spam-check-spamassassin)
1132     (spam-use-bogofilter-headers        .       spam-check-bogofilter-headers)
1133     (spam-use-bogofilter                .       spam-check-bogofilter)
1134     (spam-use-bsfilter-headers          .       spam-check-bsfilter-headers)
1135     (spam-use-bsfilter                  .       spam-check-bsfilter))
1136   "The spam-list-of-checks list contains pairs associating a
1137 parameter variable with a spam checking function.  If the
1138 parameter variable is true, then the checking function is called,
1139 and its value decides what happens.  Each individual check may
1140 return nil, t, or a mailgroup name.  The value nil means that the
1141 check does not yield a decision, and so, that further checks are
1142 needed.  The value t means that the message is definitely not
1143 spam, and that further spam checks should be inhibited.
1144 Otherwise, a mailgroup name or the symbol 'spam (depending on
1145 spam-split-symbolic-return) is returned where the mail should go,
1146 and further checks are also inhibited.  The usual mailgroup name
1147 is the value of `spam-split-group', meaning that the message is
1148 definitely a spam.")
1149
1150 (defvar spam-list-of-statistical-checks
1151   '(spam-use-ifile
1152     spam-use-regex-body
1153     spam-use-stat
1154     spam-use-bogofilter
1155     spam-use-bsfilter
1156     spam-use-blackholes
1157     spam-use-spamassassin
1158     spam-use-spamoracle)
1159   "The spam-list-of-statistical-checks list contains all the mail
1160 splitters that need to have the full message body available.
1161 Note that you should fetch extra headers if you don't like this,
1162 e.g. fetch the 'Received' header for spam-use-blackholes.")
1163
1164 (defun spam-split (&rest specific-checks)
1165   "Split this message into the `spam' group if it is spam.
1166 This function can be used as an entry in the variable `nnmail-split-fancy',
1167 for example like this: (: spam-split).  It can take checks as
1168 parameters.  A string as a parameter will set the
1169 spam-split-group to that string.
1170
1171 See the Info node `(gnus)Fancy Mail Splitting' for more details."
1172   (interactive)
1173   (setq spam-split-last-successful-check nil)
1174   (unless spam-split-disabled
1175     (let ((spam-split-group-choice spam-split-group))
1176       (dolist (check specific-checks)
1177         (when (stringp check)
1178           (setq spam-split-group-choice check)
1179           (setq specific-checks (delq check specific-checks))))
1180
1181       (let ((spam-split-group spam-split-group-choice))
1182         (save-excursion
1183           (save-restriction
1184             (dolist (check spam-list-of-statistical-checks)
1185               (when (and (symbolp check)
1186                          (or (symbol-value check)
1187                              (memq check specific-checks)))
1188                 (widen)
1189                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
1190                               (symbol-name check))
1191                 (return)))
1192             ;;   (progn (widen) (debug (buffer-string)))
1193             (let ((list-of-checks spam-list-of-checks)
1194                   decision)
1195               (while (and list-of-checks (not decision))
1196                 (let ((pair (pop list-of-checks)))
1197                   (when (or
1198                          ;; either, given specific checks, this is one of them
1199                          (and specific-checks (memq (car pair) specific-checks))
1200                          ;; or, given no specific checks, spam-use-CHECK is set
1201                          (and (null specific-checks) (symbol-value (car pair))))
1202                     (gnus-message 6 "spam-split: calling the %s function"
1203                                   (symbol-name (cdr pair)))
1204                     (setq decision (funcall (cdr pair)))
1205                     ;; if we got a decision at all, save the current check
1206                     (when decision
1207                       (setq spam-split-last-successful-check (car pair)))
1208
1209                     (when (eq decision 'spam)
1210                       (unless spam-split-symbolic-return
1211                         (gnus-error
1212                          5
1213                          (format "spam-split got %s but %s is nil"
1214                                  (symbol-name decision)
1215                                  (symbol-name spam-split-symbolic-return))))))))
1216               (if (eq decision t)
1217                   (if spam-split-symbolic-return-positive 'ham nil)
1218                 decision))))))))
1219
1220 (defun spam-find-spam ()
1221   "This function will detect spam in the current newsgroup using spam-split."
1222   (interactive)
1223
1224   (let* ((group gnus-newsgroup-name)
1225          (autodetect (gnus-parameter-spam-autodetect group))
1226          (methods (gnus-parameter-spam-autodetect-methods group))
1227          (first-method (nth 0 methods))
1228          (articles (if spam-autodetect-recheck-messages
1229                        gnus-newsgroup-articles
1230                      gnus-newsgroup-unseen))
1231          article-cannot-be-faked)
1232
1233     (dolist (check spam-list-of-statistical-checks)
1234       (when (and (symbolp check)
1235                  (memq check methods))
1236         (setq article-cannot-be-faked t)
1237         (return)))
1238
1239     (when (memq 'default methods)
1240       (setq article-cannot-be-faked t))
1241
1242     (when (and autodetect
1243                (not (equal first-method 'none)))
1244     (mapcar
1245      (lambda (article)
1246        (let ((id (spam-fetch-field-message-id-fast article))
1247              (subject (spam-fetch-field-subject-fast article))
1248              (sender (spam-fetch-field-from-fast article))
1249              registry-lookup)
1250          
1251          (unless id
1252            (gnus-message 6 "Article %d has no message ID!" article))
1253          
1254          (when (and id spam-log-to-registry)
1255            (setq registry-lookup (spam-log-registration-type id 'incoming))
1256            (when registry-lookup
1257              (gnus-message
1258               9
1259               "spam-find-spam: message %s was already registered incoming"
1260               id)))
1261
1262          (let* ((spam-split-symbolic-return t)
1263                 (spam-split-symbolic-return-positive t)
1264                 (fake-headers (spam-generate-fake-headers article))
1265                 (split-return
1266                  (or registry-lookup
1267                      (with-temp-buffer
1268                        (if article-cannot-be-faked
1269                            (gnus-request-article-this-buffer
1270                             article
1271                             group)
1272                          ;; else, we fake the article
1273                          (when fake-headers (insert fake-headers)))
1274                        (if (or (null first-method)
1275                                (equal first-method 'default))
1276                            (spam-split)
1277                          (apply 'spam-split methods))))))
1278            (if (equal split-return 'spam)
1279                (gnus-summary-mark-article article gnus-spam-mark))
1280            
1281            (when (and id split-return spam-log-to-registry)
1282              (when (zerop (gnus-registry-group-count id))
1283                (gnus-registry-add-group
1284                 id group subject sender))
1285                
1286              (unless registry-lookup
1287                (spam-log-processing-to-registry
1288                 id
1289                 'incoming
1290                 split-return
1291                 spam-split-last-successful-check
1292                 group))))))
1293     articles))))
1294
1295 (defvar spam-registration-functions
1296   ;; first the ham register, second the spam register function
1297   ;; third the ham unregister, fourth the spam unregister function
1298   '((spam-use-blacklist  nil
1299                          spam-blacklist-register-routine
1300                          nil
1301                          spam-blacklist-unregister-routine)
1302     (spam-use-whitelist  spam-whitelist-register-routine
1303                          nil
1304                          spam-whitelist-unregister-routine
1305                          nil)
1306     (spam-use-ham-copy   nil
1307                          nil
1308                          nil
1309                          nil)
1310     (spam-use-BBDB       spam-BBDB-register-routine
1311                          nil
1312                          spam-BBDB-unregister-routine
1313                          nil)
1314     (spam-use-ifile      spam-ifile-register-ham-routine
1315                          spam-ifile-register-spam-routine
1316                          spam-ifile-unregister-ham-routine
1317                          spam-ifile-unregister-spam-routine)
1318     (spam-use-spamoracle spam-spamoracle-learn-ham
1319                          spam-spamoracle-learn-spam
1320                          spam-spamoracle-unlearn-ham
1321                          spam-spamoracle-unlearn-spam)
1322     (spam-use-stat       spam-stat-register-ham-routine
1323                          spam-stat-register-spam-routine
1324                          spam-stat-unregister-ham-routine
1325                          spam-stat-unregister-spam-routine)
1326     ;; note that spam-use-gmane is not a legitimate check
1327     (spam-use-gmane      nil
1328                          spam-report-gmane-register-routine
1329                          ;; does Gmane support unregistration?
1330                          nil
1331                          nil)
1332     (spam-use-spamassassin spam-spamassassin-register-ham-routine
1333                            spam-spamassassin-register-spam-routine
1334                            spam-spamassassin-unregister-ham-routine
1335                            spam-spamassassin-unregister-spam-routine)
1336     (spam-use-bogofilter spam-bogofilter-register-ham-routine
1337                          spam-bogofilter-register-spam-routine
1338                          spam-bogofilter-unregister-ham-routine
1339                          spam-bogofilter-unregister-spam-routine)
1340     (spam-use-bsfilter   spam-bsfilter-register-ham-routine
1341                          spam-bsfilter-register-spam-routine
1342                          spam-bsfilter-unregister-ham-routine
1343                          spam-bsfilter-unregister-spam-routine))
1344   "The spam-registration-functions list contains pairs
1345 associating a parameter variable with the ham and spam
1346 registration functions, and the ham and spam unregistration
1347 functions")
1348
1349 (defun spam-classification-valid-p (classification)
1350   (or  (eq classification 'spam)
1351        (eq classification 'ham)))
1352
1353 (defun spam-process-type-valid-p (process-type)
1354   (or  (eq process-type 'incoming)
1355        (eq process-type 'process)))
1356
1357 (defun spam-registration-check-valid-p (check)
1358   (assoc check spam-registration-functions))
1359
1360 (defun spam-unregistration-check-valid-p (check)
1361   (assoc check spam-registration-functions))
1362
1363 (defun spam-registration-function (classification check)
1364   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1365     (if (eq classification 'spam)
1366         (nth 1 flist)
1367       (nth 0 flist))))
1368
1369 (defun spam-unregistration-function (classification check)
1370   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1371     (if (eq classification 'spam)
1372         (nth 3 flist)
1373       (nth 2 flist))))
1374
1375 (defun spam-list-articles (articles classification)
1376   (let ((mark-check (if (eq classification 'spam)
1377                         'spam-group-spam-mark-p
1378                       'spam-group-ham-mark-p))
1379         alist mark-cache-yes mark-cache-no)
1380     (dolist (article articles)
1381       (let ((mark (gnus-summary-article-mark article)))
1382         (unless (or (memq mark mark-cache-yes)
1383                     (memq mark mark-cache-no))
1384           (if (funcall mark-check
1385                        gnus-newsgroup-name
1386                        mark)
1387               (push mark mark-cache-yes)
1388             (push mark mark-cache-no)))
1389         (when (memq mark mark-cache-yes)
1390           (push article alist))))
1391     alist))
1392
1393 (defun spam-register-routine (classification
1394                               check
1395                               &optional unregister
1396                               specific-articles)
1397   (when (and (spam-classification-valid-p classification)
1398              (spam-registration-check-valid-p check))
1399     (let* ((register-function
1400             (spam-registration-function classification check))
1401            (unregister-function
1402             (spam-unregistration-function classification check))
1403            (run-function (if unregister
1404                              unregister-function
1405                            register-function))
1406            (log-function (if unregister
1407                              'spam-log-undo-registration
1408                            'spam-log-processing-to-registry))
1409            article articles)
1410
1411       (when run-function
1412         ;; make list of articles, using specific-articles if given
1413         (setq articles (or specific-articles
1414                            (spam-list-articles
1415                             gnus-newsgroup-articles
1416                             classification)))
1417         ;; process them
1418         (gnus-message 5 "%s %d %s articles as %s using backend %s"
1419                       (if unregister "Unregistering" "Registering")
1420                       (length articles)
1421                       (if specific-articles "specific" "")
1422                       (symbol-name classification)
1423                       (symbol-name check))
1424         (funcall run-function articles)
1425         ;; now log all the registrations (or undo them, depending on unregister)
1426         (dolist (article articles)
1427           (funcall log-function
1428                    (spam-fetch-field-message-id-fast article)
1429                    'process
1430                    classification
1431                    check
1432                    gnus-newsgroup-name))))))
1433
1434 ;;; log a ham- or spam-processor invocation to the registry
1435 (defun spam-log-processing-to-registry (id type classification check group)
1436   (when spam-log-to-registry
1437     (if (and (stringp id)
1438              (stringp group)
1439              (spam-process-type-valid-p type)
1440              (spam-classification-valid-p classification)
1441              (spam-registration-check-valid-p check))
1442         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1443               (cell (list classification check group)))
1444           (push cell cell-list)
1445           (gnus-registry-store-extra-entry
1446            id
1447            type
1448            cell-list))
1449
1450       (gnus-message 
1451        5 
1452        (format "%s call with bad ID, type, classification, spam-check, or group"
1453                "spam-log-processing-to-registry")))))
1454
1455 ;;; check if a ham- or spam-processor registration has been done
1456 (defun spam-log-registered-p (id type)
1457   (when spam-log-to-registry
1458     (if (and (stringp id)
1459              (spam-process-type-valid-p type))
1460         (cdr-safe (gnus-registry-fetch-extra id type))
1461       (progn
1462         (gnus-message
1463          5 
1464          (format "%s called with bad ID, type, classification, or spam-check"
1465                  "spam-log-registered-p"))
1466         nil))))
1467
1468 ;;; check what a ham- or spam-processor registration says
1469 ;;; returns nil if conflicting registrations are found
1470 (defun spam-log-registration-type (id type)
1471   (let ((count 0)
1472         decision)
1473     (dolist (reg (spam-log-registered-p id type))
1474       (let ((classification (nth 0 reg)))
1475         (when (spam-classification-valid-p classification)
1476           (when (and decision
1477                      (not (eq classification decision)))
1478             (setq count (+ 1 count)))
1479           (setq decision classification))))
1480     (if (< 0 count)
1481         nil
1482       decision)))
1483
1484
1485 ;;; check if a ham- or spam-processor registration needs to be undone
1486 (defun spam-log-unregistration-needed-p (id type classification check)
1487   (when spam-log-to-registry
1488     (if (and (stringp id)
1489              (spam-process-type-valid-p type)
1490              (spam-classification-valid-p classification)
1491              (spam-registration-check-valid-p check))
1492         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1493               found)
1494           (dolist (cell cell-list)
1495             (unless found
1496               (when (and (eq classification (nth 0 cell))
1497                          (eq check (nth 1 cell)))
1498                 (setq found t))))
1499           found)
1500       (progn
1501         (gnus-message
1502          5 
1503          (format "%s called with bad ID, type, classification, or spam-check"
1504                  "spam-log-unregistration-needed-p"))
1505         nil))))
1506
1507
1508 ;;; undo a ham- or spam-processor registration (the group is not used)
1509 (defun spam-log-undo-registration (id type classification check &optional group)
1510   (when (and spam-log-to-registry
1511              (spam-log-unregistration-needed-p id type classification check))
1512     (if (and (stringp id)
1513              (spam-process-type-valid-p type)
1514              (spam-classification-valid-p classification)
1515              (spam-registration-check-valid-p check))
1516         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1517               new-cell-list found)
1518           (dolist (cell cell-list)
1519             (unless (and (eq classification (nth 0 cell))
1520                          (eq check (nth 1 cell)))
1521               (push cell new-cell-list)))
1522           (gnus-registry-store-extra-entry
1523            id
1524            type
1525            new-cell-list))
1526       (progn
1527         (gnus-message 6 (format "%s call with bad ID, type, spam-check, or group"
1528                                 "spam-log-undo-registration"))
1529         nil))))
1530
1531 ;;; set up IMAP widening if it's necessary
1532 (defun spam-setup-widening ()
1533   (dolist (check spam-list-of-statistical-checks)
1534     (when (symbol-value check)
1535       (setq nnimap-split-download-body-default t))))
1536
1537 \f
1538 ;;;; Gmane xrefs
1539 (defun spam-check-gmane-xref ()
1540   (let ((header (or
1541                  (message-fetch-field "Xref")
1542                  (message-fetch-field "Newsgroups")))
1543         (spam-split-group (if spam-split-symbolic-return
1544                               'spam
1545                             spam-split-group)))
1546     (when header                        ; return nil when no header
1547       (when (string-match spam-gmane-xref-spam-group
1548                           header)
1549           spam-split-group))))
1550
1551 \f
1552 ;;;; Regex body
1553
1554 (defun spam-check-regex-body ()
1555   (let ((spam-regex-headers-ham spam-regex-body-ham)
1556         (spam-regex-headers-spam spam-regex-body-spam))
1557     (spam-check-regex-headers t)))
1558
1559 \f
1560 ;;;; Regex headers
1561
1562 (defun spam-check-regex-headers (&optional body)
1563   (let ((type (if body "body" "header"))
1564         (spam-split-group (if spam-split-symbolic-return
1565                               'spam
1566                             spam-split-group))
1567         ret found)
1568     (dolist (h-regex spam-regex-headers-ham)
1569       (unless found
1570         (goto-char (point-min))
1571         (when (re-search-forward h-regex nil t)
1572           (message "Ham regex %s search positive." type)
1573           (setq found t))))
1574     (dolist (s-regex spam-regex-headers-spam)
1575       (unless found
1576         (goto-char (point-min))
1577         (when (re-search-forward s-regex nil t)
1578           (message "Spam regex %s search positive." type)
1579           (setq found t)
1580           (setq ret spam-split-group))))
1581     ret))
1582
1583 \f
1584 ;;;; Blackholes.
1585
1586 (defun spam-reverse-ip-string (ip)
1587   (when (stringp ip)
1588     (mapconcat 'identity
1589                (nreverse (split-string ip "\\."))
1590                ".")))
1591
1592 (defun spam-check-blackholes ()
1593   "Check the Received headers for blackholed relays."
1594   (let ((headers (message-fetch-field "received"))
1595         (spam-split-group (if spam-split-symbolic-return
1596                               'spam
1597                             spam-split-group))
1598         ips matches)
1599     (when headers
1600       (with-temp-buffer
1601         (insert headers)
1602         (goto-char (point-min))
1603         (gnus-message 6 "Checking headers for relay addresses")
1604         (while (re-search-forward
1605                 "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
1606           (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
1607           (push (spam-reverse-ip-string (match-string 1))
1608                 ips)))
1609       (dolist (server spam-blackhole-servers)
1610         (dolist (ip ips)
1611           (unless (and spam-blackhole-good-server-regex
1612                        ;; match the good-server-regex against the reversed (again) IP string
1613                        (string-match
1614                         spam-blackhole-good-server-regex
1615                         (spam-reverse-ip-string ip)))
1616             (unless matches
1617               (let ((query-string (concat ip "." server)))
1618                 (if spam-use-dig
1619                     (let ((query-result (query-dig query-string)))
1620                       (when query-result
1621                         (gnus-message 6 "(DIG): positive blackhole check '%s'"
1622                                       query-result)
1623                         (push (list ip server query-result)
1624                               matches)))
1625                   ;; else, if not using dig.el
1626                   (when (query-dns query-string)
1627                     (gnus-message 6 "positive blackhole check")
1628                     (push (list ip server (query-dns query-string 'TXT))
1629                           matches)))))))))
1630     (when matches
1631       spam-split-group)))
1632 \f
1633 ;;;; Hashcash.
1634
1635 (condition-case nil
1636     (progn
1637       (require 'hashcash)
1638
1639       (defun spam-check-hashcash ()
1640         "Check the headers for hashcash payments."
1641         (mail-check-payment)))   ;mail-check-payment returns a boolean
1642
1643   (file-error (progn
1644                 (defalias 'mail-check-payment 'ignore)
1645                 (defalias 'spam-check-hashcash 'ignore))))
1646 \f
1647 ;;;; BBDB
1648
1649 ;;; original idea for spam-check-BBDB from Alexander Kotelnikov
1650 ;;; <sacha@giotto.sj.ru>
1651
1652 ;; all this is done inside a condition-case to trap errors
1653
1654 (condition-case nil
1655     (progn
1656       (require 'bbdb)
1657       (require 'bbdb-com)
1658
1659       ;; when the BBDB changes, we want to clear out our cache
1660       (defun spam-clear-cache-BBDB (&rest immaterial)
1661         (spam-clear-cache 'spam-use-BBDB))
1662
1663       (add-hook 'bbdb-change-hook 'spam-clear-cache-BBDB)
1664
1665       (defun spam-enter-ham-BBDB (addresses &optional remove)
1666         "Enter an address into the BBDB; implies ham (non-spam) sender"
1667         (dolist (from addresses)
1668           (when (stringp from)
1669             (let* ((parsed-address (gnus-extract-address-components from))
1670                    (name (or (nth 0 parsed-address) "Ham Sender"))
1671                    (remove-function (if remove
1672                                         'bbdb-delete-record-internal
1673                                       'ignore))
1674                    (net-address (nth 1 parsed-address))
1675                    (record (and net-address
1676                                 (bbdb-search-simple nil net-address))))
1677               (when net-address
1678                 (gnus-message 6 "%s address %s %s BBDB"
1679                               (if remove "Deleting" "Adding")
1680                               from
1681                               (if remove "from" "to"))
1682                 (if record
1683                     (funcall remove-function record)
1684                   (bbdb-create-internal name nil net-address nil nil
1685                                         "ham sender added by spam.el")))))))
1686
1687       (defun spam-BBDB-register-routine (articles &optional unregister)
1688         (let (addresses)
1689           (dolist (article articles)
1690             (when (stringp (spam-fetch-field-from-fast article))
1691               (push (spam-fetch-field-from-fast article) addresses)))
1692           ;; now do the register/unregister action
1693           (spam-enter-ham-BBDB addresses unregister)))
1694
1695       (defun spam-BBDB-unregister-routine (articles)
1696         (spam-BBDB-register-routine articles t))
1697
1698       (defun spam-check-BBDB ()
1699         "Mail from people in the BBDB is classified as ham or non-spam"
1700         (let ((who (message-fetch-field "from"))
1701               (spam-split-group (if spam-split-symbolic-return
1702                                     'spam
1703                                   spam-split-group))
1704               bbdb-cache bbdb-hashtable)
1705           (when spam-cache-lookups
1706             (setq bbdb-cache (gethash 'spam-use-BBDB spam-caches))
1707             (unless bbdb-cache
1708               (setq bbdb-cache
1709                     ;; this is the expanded (bbdb-hashtable) macro
1710                     ;; without the debugging support
1711                     (with-current-buffer (bbdb-buffer)
1712                       (save-excursion
1713                         (save-window-excursion
1714                           (bbdb-records nil t)
1715                           bbdb-hashtable))))
1716               (puthash 'spam-use-BBDB bbdb-cache spam-caches)))
1717           (when who
1718             (setq who (nth 1 (gnus-extract-address-components who)))
1719             (if
1720                 (if spam-cache-lookups
1721                     (symbol-value
1722                      (intern-soft who bbdb-cache))
1723                   (bbdb-search-simple nil who))
1724                 t
1725               (if spam-use-BBDB-exclusive
1726                   spam-split-group
1727                 nil))))))
1728
1729   (file-error (progn
1730                 (defalias 'bbdb-search-simple 'ignore)
1731                 (defalias 'bbdb-records 'ignore)
1732                 (defalias 'bbdb-buffer 'ignore)
1733                 (defalias 'spam-check-BBDB 'ignore)
1734                 (defalias 'spam-BBDB-register-routine 'ignore)
1735                 (defalias 'spam-enter-ham-BBDB 'ignore)
1736                 (defalias 'bbdb-create-internal 'ignore)
1737                 (defalias 'bbdb-delete-record-internal 'ignore)
1738                 (defalias 'bbdb-records 'ignore))))
1739
1740 \f
1741 ;;;; ifile
1742
1743 ;;; check the ifile backend; return nil if the mail was NOT classified
1744 ;;; as spam
1745
1746 (defun spam-get-ifile-database-parameter ()
1747   "Get the command-line parameter for ifile's database from
1748   spam-ifile-database-path."
1749   (if spam-ifile-database-path
1750       (format "--db-file=%s" spam-ifile-database-path)
1751     nil))
1752
1753 (defun spam-check-ifile ()
1754   "Check the ifile backend for the classification of this message."
1755   (let ((article-buffer-name (buffer-name))
1756         (spam-split-group (if spam-split-symbolic-return
1757                               'spam
1758                             spam-split-group))
1759         category return)
1760     (with-temp-buffer
1761       (let ((temp-buffer-name (buffer-name))
1762             (db-param (spam-get-ifile-database-parameter)))
1763         (save-excursion
1764           (set-buffer article-buffer-name)
1765           (apply 'call-process-region
1766                  (point-min) (point-max) spam-ifile-path
1767                  nil temp-buffer-name nil "-c"
1768                  (if db-param `(,db-param "-q") `("-q"))))
1769         ;; check the return now (we're back in the temp buffer)
1770         (goto-char (point-min))
1771         (if (not (eobp))
1772             (setq category (buffer-substring (point) (point-at-eol))))
1773         (when (not (zerop (length category))) ; we need a category here
1774           (if spam-ifile-all-categories
1775               (setq return category)
1776             ;; else, if spam-ifile-all-categories is not set...
1777             (when (string-equal spam-ifile-spam-category category)
1778               (setq return spam-split-group)))))) ; note return is nil otherwise
1779     return))
1780
1781 (defun spam-ifile-register-with-ifile (articles category &optional unregister)
1782   "Register an article, given as a string, with a category.
1783 Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
1784   (let ((category (or category gnus-newsgroup-name))
1785         (add-or-delete-option (if unregister "-d" "-i"))
1786         (db (spam-get-ifile-database-parameter))
1787         parameters)
1788     (with-temp-buffer
1789       (dolist (article articles)
1790         (let ((article-string (spam-get-article-as-string article)))
1791           (when (stringp article-string)
1792             (insert article-string))))
1793       (apply 'call-process-region
1794              (point-min) (point-max) spam-ifile-path
1795              nil nil nil
1796              add-or-delete-option category
1797              (if db `(,db "-h") `("-h"))))))
1798
1799 (defun spam-ifile-register-spam-routine (articles &optional unregister)
1800   (spam-ifile-register-with-ifile articles spam-ifile-spam-category unregister))
1801
1802 (defun spam-ifile-unregister-spam-routine (articles)
1803   (spam-ifile-register-spam-routine articles t))
1804
1805 (defun spam-ifile-register-ham-routine (articles &optional unregister)
1806   (spam-ifile-register-with-ifile articles spam-ifile-ham-category unregister))
1807
1808 (defun spam-ifile-unregister-ham-routine (articles)
1809   (spam-ifile-register-ham-routine articles t))
1810
1811 \f
1812 ;;;; spam-stat
1813
1814 (condition-case nil
1815     (progn
1816       (let ((spam-stat-install-hooks nil))
1817         (require 'spam-stat))
1818
1819       (defun spam-check-stat ()
1820         "Check the spam-stat backend for the classification of this message"
1821         (let ((spam-split-group (if spam-split-symbolic-return
1822                                     'spam
1823                                   spam-split-group))
1824               (spam-stat-split-fancy-spam-group spam-split-group) ; override
1825               (spam-stat-buffer (buffer-name)) ; stat the current buffer
1826               category return)
1827           (spam-stat-split-fancy)))
1828
1829       (defun spam-stat-register-spam-routine (articles &optional unregister)
1830         (dolist (article articles)
1831           (let ((article-string (spam-get-article-as-string article)))
1832             (with-temp-buffer
1833               (insert article-string)
1834               (if unregister
1835                   (spam-stat-buffer-change-to-non-spam)
1836               (spam-stat-buffer-is-spam))))))
1837
1838       (defun spam-stat-unregister-spam-routine (articles)
1839         (spam-stat-register-spam-routine articles t))
1840
1841       (defun spam-stat-register-ham-routine (articles &optional unregister)
1842         (dolist (article articles)
1843           (let ((article-string (spam-get-article-as-string article)))
1844             (with-temp-buffer
1845               (insert article-string)
1846               (if unregister
1847                   (spam-stat-buffer-change-to-spam)
1848               (spam-stat-buffer-is-non-spam))))))
1849
1850       (defun spam-stat-unregister-ham-routine (articles)
1851         (spam-stat-register-ham-routine articles t))
1852
1853       (defun spam-maybe-spam-stat-load ()
1854         (when spam-use-stat (spam-stat-load)))
1855
1856       (defun spam-maybe-spam-stat-save ()
1857         (when spam-use-stat (spam-stat-save))))
1858
1859   (file-error (progn
1860                 (defalias 'spam-stat-load 'ignore)
1861                 (defalias 'spam-stat-save 'ignore)
1862                 (defalias 'spam-maybe-spam-stat-load 'ignore)
1863                 (defalias 'spam-maybe-spam-stat-save 'ignore)
1864                 (defalias 'spam-stat-register-ham-routine 'ignore)
1865                 (defalias 'spam-stat-unregister-ham-routine 'ignore)
1866                 (defalias 'spam-stat-register-spam-routine 'ignore)
1867                 (defalias 'spam-stat-unregister-spam-routine 'ignore)
1868                 (defalias 'spam-stat-buffer-is-spam 'ignore)
1869                 (defalias 'spam-stat-buffer-change-to-spam 'ignore)
1870                 (defalias 'spam-stat-buffer-is-non-spam 'ignore)
1871                 (defalias 'spam-stat-buffer-change-to-non-spam 'ignore)
1872                 (defalias 'spam-stat-split-fancy 'ignore)
1873                 (defalias 'spam-check-stat 'ignore))))
1874
1875 \f
1876
1877 ;;;; Blacklists and whitelists.
1878
1879 (defvar spam-whitelist-cache nil)
1880 (defvar spam-blacklist-cache nil)
1881
1882 (defun spam-kill-whole-line ()
1883   (beginning-of-line)
1884   (let ((kill-whole-line t))
1885     (kill-line)))
1886
1887 ;;; address can be a list, too
1888 (defun spam-enter-whitelist (address &optional remove)
1889   "Enter ADDRESS (list or single) into the whitelist.
1890 With a non-nil REMOVE, remove them."
1891   (interactive "sAddress: ")
1892   (spam-enter-list address spam-whitelist remove)
1893   (setq spam-whitelist-cache nil)
1894   (spam-clear-cache 'spam-use-whitelist))
1895
1896 ;;; address can be a list, too
1897 (defun spam-enter-blacklist (address &optional remove)
1898   "Enter ADDRESS (list or single) into the blacklist.
1899 With a non-nil REMOVE, remove them."
1900   (interactive "sAddress: ")
1901   (spam-enter-list address spam-blacklist remove)
1902   (setq spam-blacklist-cache nil)
1903   (spam-clear-cache 'spam-use-whitelist))
1904
1905 (defun spam-enter-list (addresses file &optional remove)
1906   "Enter ADDRESSES into the given FILE.
1907 Either the whitelist or the blacklist files can be used.  With
1908 REMOVE not nil, remove the ADDRESSES."
1909   (if (stringp addresses)
1910       (spam-enter-list (list addresses) file remove)
1911     ;; else, we have a list of addresses here
1912     (unless (file-exists-p (file-name-directory file))
1913       (make-directory (file-name-directory file) t))
1914     (save-excursion
1915       (set-buffer
1916        (find-file-noselect file))
1917       (dolist (a addresses)
1918         (when (stringp a)
1919           (goto-char (point-min))
1920           (if (re-search-forward (regexp-quote a) nil t)
1921               ;; found the address
1922               (when remove
1923                 (spam-kill-whole-line))
1924             ;; else, the address was not found
1925             (unless remove
1926               (goto-char (point-max))
1927               (unless (bobp)
1928                 (insert "\n"))
1929               (insert a "\n")))))
1930       (save-buffer))))
1931
1932 (defun spam-filelist-build-cache (type)
1933   (let ((cache (if (eq type 'spam-use-blacklist)
1934                    spam-blacklist-cache
1935                  spam-whitelist-cache))
1936         parsed-cache)
1937     (unless (gethash type spam-caches)
1938       (while cache
1939         (let ((address (pop cache)))
1940           (unless (zerop (length address)) ; 0 for a nil address too
1941             (setq address (regexp-quote address))
1942             ;; fix regexp-quote's treatment of user-intended regexes
1943             (while (string-match "\\\\\\*" address)
1944               (setq address (replace-match ".*" t t address))))
1945           (push address parsed-cache)))
1946       (puthash type parsed-cache spam-caches))))
1947
1948 (defun spam-filelist-check-cache (type from)
1949   (when (stringp from)
1950     (spam-filelist-build-cache type)
1951     (let (found)
1952       (dolist (address (gethash type spam-caches))
1953         (when (and address (string-match address from))
1954           (setq found t)
1955           (return)))
1956       found)))
1957
1958 ;;; returns t if the sender is in the whitelist, nil or
1959 ;;; spam-split-group otherwise
1960 (defun spam-check-whitelist ()
1961   ;; FIXME!  Should it detect when file timestamps change?
1962   (let ((spam-split-group (if spam-split-symbolic-return
1963                               'spam
1964                             spam-split-group)))
1965     (unless spam-whitelist-cache
1966       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
1967     (if (spam-from-listed-p 'spam-use-whitelist)
1968         t
1969       (if spam-use-whitelist-exclusive
1970           spam-split-group
1971         nil))))
1972
1973 (defun spam-check-blacklist ()
1974   ;; FIXME!  Should it detect when file timestamps change?
1975   (let ((spam-split-group (if spam-split-symbolic-return
1976                               'spam
1977                             spam-split-group)))
1978     (unless spam-blacklist-cache
1979       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
1980     (and (spam-from-listed-p 'spam-use-blacklist) spam-split-group)))
1981
1982 (defun spam-parse-list (file)
1983   (when (file-readable-p file)
1984     (let (contents address)
1985       (with-temp-buffer
1986         (insert-file-contents file)
1987         (while (not (eobp))
1988           (setq address (buffer-substring (point) (point-at-eol)))
1989           (forward-line 1)
1990           ;; insert the e-mail address if detected, otherwise the raw data
1991           (unless (zerop (length address))
1992             (let ((pure-address (nth 1 (gnus-extract-address-components address))))
1993               (push (or pure-address address) contents)))))
1994       (nreverse contents))))
1995
1996 (defun spam-from-listed-p (type)
1997   (let ((from (message-fetch-field "from"))
1998         found)
1999     (spam-filelist-check-cache type from)))
2000
2001 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
2002   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
2003         (declassification (if blacklist 'ham 'spam))
2004         (enter-function
2005          (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
2006         (remove-function
2007          (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
2008         from addresses unregister-list article-unregister-list)
2009     (dolist (article articles)
2010       (let ((from (spam-fetch-field-from-fast article))
2011             (id (spam-fetch-field-message-id-fast article))
2012             sender-ignored)
2013         (when (stringp from)
2014           (dolist (ignore-regex spam-blacklist-ignored-regexes)
2015             (when (and (not sender-ignored)
2016                        (stringp ignore-regex)
2017                        (string-match ignore-regex from))
2018               (setq sender-ignored t)))
2019           ;; remember the messages we need to unregister, unless remove is set
2020           (when (and
2021                  (null unregister)
2022                  (spam-log-unregistration-needed-p
2023                   id 'process declassification de-symbol))
2024             (push article article-unregister-list)
2025             (push from unregister-list))
2026           (unless sender-ignored
2027             (push from addresses)))))
2028
2029     (if unregister
2030         (funcall enter-function addresses t) ; unregister all these addresses
2031       ;; else, register normally and unregister what we need to
2032       (funcall remove-function unregister-list t)
2033       (dolist (article article-unregister-list)
2034         (spam-log-undo-registration
2035          (spam-fetch-field-message-id-fast article)
2036          'process
2037          declassification
2038          de-symbol))
2039       (funcall enter-function addresses nil))))
2040
2041 (defun spam-blacklist-unregister-routine (articles)
2042   (spam-blacklist-register-routine articles t))
2043
2044 (defun spam-blacklist-register-routine (articles &optional unregister)
2045   (spam-filelist-register-routine articles t unregister))
2046
2047 (defun spam-whitelist-unregister-routine (articles)
2048   (spam-whitelist-register-routine articles t))
2049
2050 (defun spam-whitelist-register-routine (articles &optional unregister)
2051   (spam-filelist-register-routine articles nil unregister))
2052
2053 \f
2054 ;;;; Spam-report glue
2055 (defun spam-report-gmane-register-routine (articles)
2056   (when articles
2057     (apply 'spam-report-gmane articles)))
2058
2059 \f
2060 ;;;; Bogofilter
2061 (defun spam-check-bogofilter-headers (&optional score)
2062   (let ((header (message-fetch-field spam-bogofilter-header))
2063         (spam-split-group (if spam-split-symbolic-return
2064                               'spam
2065                             spam-split-group)))
2066     (when header                        ; return nil when no header
2067       (if score                         ; scoring mode
2068           (if (string-match "spamicity=\\([0-9.]+\\)" header)
2069               (match-string 1 header)
2070             "0")
2071         ;; spam detection mode
2072         (when (string-match spam-bogofilter-bogosity-positive-spam-header
2073                             header)
2074           spam-split-group)))))
2075
2076 ;; return something sensible if the score can't be determined
2077 (defun spam-bogofilter-score (&optional recheck)
2078   "Get the Bogofilter spamicity score"
2079   (interactive "P")
2080   (save-window-excursion
2081     (gnus-summary-show-article t)
2082     (set-buffer gnus-article-buffer)
2083     (let ((score (or (unless recheck
2084                        (spam-check-bogofilter-headers t))
2085                      (spam-check-bogofilter t))))
2086       (gnus-summary-show-article)
2087       (message "Spamicity score %s" score)
2088       (or score "0"))))
2089
2090 (defun spam-check-bogofilter (&optional score)
2091   "Check the Bogofilter backend for the classification of this message"
2092   (let ((article-buffer-name (buffer-name))
2093         (db spam-bogofilter-database-directory)
2094         return)
2095     (with-temp-buffer
2096       (let ((temp-buffer-name (buffer-name)))
2097         (save-excursion
2098           (set-buffer article-buffer-name)
2099           (apply 'call-process-region
2100                  (point-min) (point-max)
2101                  spam-bogofilter-path
2102                  nil temp-buffer-name nil
2103                  (if db `("-d" ,db "-v") `("-v"))))
2104         (setq return (spam-check-bogofilter-headers score))))
2105     return))
2106
2107 (defun spam-bogofilter-register-with-bogofilter (articles
2108                                                  spam
2109                                                  &optional unregister)
2110   "Register an article, given as a string, as spam or non-spam."
2111   (dolist (article articles)
2112     (let ((article-string (spam-get-article-as-string article))
2113           (db spam-bogofilter-database-directory)
2114           (switch (if unregister
2115                       (if spam
2116                           spam-bogofilter-spam-strong-switch
2117                         spam-bogofilter-ham-strong-switch)
2118                     (if spam
2119                         spam-bogofilter-spam-switch
2120                       spam-bogofilter-ham-switch))))
2121       (when (stringp article-string)
2122         (with-temp-buffer
2123           (insert article-string)
2124
2125           (apply 'call-process-region
2126                  (point-min) (point-max)
2127                  spam-bogofilter-path
2128                  nil nil nil switch
2129                  (if db `("-d" ,db "-v") `("-v"))))))))
2130
2131 (defun spam-bogofilter-register-spam-routine (articles &optional unregister)
2132   (spam-bogofilter-register-with-bogofilter articles t unregister))
2133
2134 (defun spam-bogofilter-unregister-spam-routine (articles)
2135   (spam-bogofilter-register-spam-routine articles t))
2136
2137 (defun spam-bogofilter-register-ham-routine (articles &optional unregister)
2138   (spam-bogofilter-register-with-bogofilter articles nil unregister))
2139
2140 (defun spam-bogofilter-unregister-ham-routine (articles)
2141   (spam-bogofilter-register-ham-routine articles t))
2142
2143
2144 \f
2145 ;;;; spamoracle
2146 (defun spam-check-spamoracle ()
2147   "Run spamoracle on an article to determine whether it's spam."
2148   (let ((article-buffer-name (buffer-name))
2149         (spam-split-group (if spam-split-symbolic-return
2150                               'spam
2151                             spam-split-group)))
2152     (with-temp-buffer
2153       (let ((temp-buffer-name (buffer-name)))
2154         (save-excursion
2155           (set-buffer article-buffer-name)
2156           (let ((status
2157                  (apply 'call-process-region
2158                         (point-min) (point-max)
2159                         spam-spamoracle-binary
2160                         nil temp-buffer-name nil
2161                         (if spam-spamoracle-database
2162                             `("-f" ,spam-spamoracle-database "mark")
2163                           '("mark")))))
2164             (if (eq 0 status)
2165                 (progn
2166                   (set-buffer temp-buffer-name)
2167                   (goto-char (point-min))
2168                   (when (re-search-forward "^X-Spam: yes;" nil t)
2169                     spam-split-group))
2170               (error "Error running spamoracle: %s" status))))))))
2171
2172 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
2173   "Run spamoracle in training mode."
2174   (with-temp-buffer
2175     (let ((temp-buffer-name (buffer-name)))
2176       (save-excursion
2177         (goto-char (point-min))
2178         (dolist (article articles)
2179           (insert (spam-get-article-as-string article)))
2180         (let* ((arg (if (spam-xor unregister article-is-spam-p)
2181                         "-spam"
2182                       "-good"))
2183                (status
2184                 (apply 'call-process-region
2185                        (point-min) (point-max)
2186                        spam-spamoracle-binary
2187                        nil temp-buffer-name nil
2188                        (if spam-spamoracle-database
2189                            `("-f" ,spam-spamoracle-database
2190                              "add" ,arg)
2191                          `("add" ,arg)))))
2192           (unless (eq 0 status)
2193             (error "Error running spamoracle: %s" status)))))))
2194
2195 (defun spam-spamoracle-learn-ham (articles &optional unregister)
2196   (spam-spamoracle-learn articles nil unregister))
2197
2198 (defun spam-spamoracle-unlearn-ham (articles &optional unregister)
2199   (spam-spamoracle-learn-ham articles t))
2200
2201 (defun spam-spamoracle-learn-spam (articles &optional unregister)
2202   (spam-spamoracle-learn articles t unregister))
2203
2204 (defun spam-spamoracle-unlearn-spam (articles &optional unregister)
2205   (spam-spamoracle-learn-spam articles t))
2206
2207 \f
2208 ;;;; SpamAssassin
2209 ;;; based mostly on the bogofilter code
2210 (defun spam-check-spamassassin-headers (&optional score)
2211   "Check the SpamAssassin headers for the classification of this message."
2212   (if score                             ; scoring mode
2213       (let ((header (message-fetch-field spam-spamassassin-spam-status-header)))
2214         (when header
2215           (if (string-match "hits=\\(-?[0-9.]+\\)" header)
2216               (match-string 1 header)
2217             "0")))
2218     ;; spam detection mode
2219     (let ((header (message-fetch-field spam-spamassassin-spam-flag-header))
2220           (spam-split-group (if spam-split-symbolic-return
2221                                  'spam
2222                                spam-split-group)))
2223           (when header                  ; return nil when no header
2224             (when (string-match spam-spamassassin-positive-spam-flag-header
2225                                 header)
2226               spam-split-group)))))
2227
2228 (defun spam-check-spamassassin (&optional score)
2229   "Check the SpamAssassin backend for the classification of this message."
2230   (let ((article-buffer-name (buffer-name)))
2231     (with-temp-buffer
2232       (let ((temp-buffer-name (buffer-name)))
2233         (save-excursion
2234           (set-buffer article-buffer-name)
2235           (apply 'call-process-region
2236                  (point-min) (point-max) spam-spamassassin-path
2237                  nil temp-buffer-name nil spam-spamassassin-arguments))
2238         ;; check the return now (we're back in the temp buffer)
2239         (goto-char (point-min))
2240         (spam-check-spamassassin-headers score)))))
2241
2242 ;; return something sensible if the score can't be determined
2243 (defun spam-spamassassin-score (&optional recheck)
2244   "Get the SpamAssassin score"
2245   (interactive "P")
2246   (save-window-excursion
2247     (gnus-summary-show-article t)
2248     (set-buffer gnus-article-buffer)
2249     (let ((score (or (unless recheck
2250                        (spam-check-spamassassin-headers t))
2251                      (spam-check-spamassassin t))))
2252       (gnus-summary-show-article)
2253       (message "SpamAssassin score %s" score)
2254       (or score "0"))))
2255
2256 (defun spam-spamassassin-register-with-sa-learn (articles spam
2257                                                  &optional unregister)
2258   "Register articles with spamassassin's sa-learn as spam or non-spam."
2259   (if articles
2260       (let ((action (if unregister spam-sa-learn-unregister-switch
2261                       (if spam spam-sa-learn-spam-switch
2262                         spam-sa-learn-ham-switch)))
2263             (summary-buffer-name (buffer-name)))
2264         (with-temp-buffer
2265           ;; group the articles into mbox format
2266           (dolist (article articles)
2267             (let (article-string)
2268               (save-excursion
2269                 (set-buffer summary-buffer-name)
2270                 (setq article-string (spam-get-article-as-string article)))
2271               (when (stringp article-string)
2272                 (insert "From \n") ; mbox separator (sa-learn only checks the
2273                                    ; first five chars, so we can get away with
2274                                    ; a bogus line))
2275                 (insert article-string)
2276                 (insert "\n"))))
2277           ;; call sa-learn on all messages at the same time
2278           (apply 'call-process-region
2279                  (point-min) (point-max)
2280                  spam-sa-learn-path
2281                  nil nil nil "--mbox"
2282                  (if spam-sa-learn-rebuild
2283                      (list action)
2284                    `("--no-rebuild" ,action)))))))
2285
2286 (defun spam-spamassassin-register-spam-routine (articles &optional unregister)
2287   (spam-spamassassin-register-with-sa-learn articles t unregister))
2288
2289 (defun spam-spamassassin-register-ham-routine (articles &optional unregister)
2290   (spam-spamassassin-register-with-sa-learn articles nil unregister))
2291
2292 (defun spam-spamassassin-unregister-spam-routine (articles)
2293   (spam-spamassassin-register-with-sa-learn articles t t))
2294
2295 (defun spam-spamassassin-unregister-ham-routine (articles)
2296   (spam-spamassassin-register-with-sa-learn articles nil t))
2297
2298 \f
2299 ;;;; Bsfilter
2300 ;;; based mostly on the bogofilter code
2301 (defun spam-check-bsfilter-headers (&optional score)
2302   (if score
2303       (or (nnmail-fetch-field spam-bsfilter-probability-header)
2304           "0")
2305     (let ((header (nnmail-fetch-field spam-bsfilter-header))
2306           (spam-split-group (if spam-split-symbolic-return
2307                                 'spam
2308                               spam-split-group)))
2309       (when header ; return nil when no header
2310         (when (string-match "YES" header)
2311           spam-split-group)))))
2312
2313 ;; return something sensible if the score can't be determined
2314 (defun spam-bsfilter-score (&optional recheck)
2315   "Get the Bsfilter spamicity score"
2316   (interactive "P")
2317   (save-window-excursion
2318     (gnus-summary-show-article t)
2319     (set-buffer gnus-article-buffer)
2320     (let ((score (or (unless recheck
2321                        (spam-check-bsfilter-headers t))
2322                      (spam-check-bsfilter t))))
2323       (gnus-summary-show-article)
2324       (message "Spamicity score %s" score)
2325       (or score "0"))))
2326
2327 (defun spam-check-bsfilter (&optional score)
2328   "Check the Bsfilter backend for the classification of this message"
2329   (let ((article-buffer-name (buffer-name))
2330         (dir spam-bsfilter-database-directory)
2331         return)
2332     (with-temp-buffer
2333       (let ((temp-buffer-name (buffer-name)))
2334         (save-excursion
2335           (set-buffer article-buffer-name)
2336           (apply 'call-process-region
2337                  (point-min) (point-max)
2338                  spam-bsfilter-path
2339                  nil temp-buffer-name nil
2340                  "--pipe"
2341                  "--insert-flag"
2342                  "--insert-probability"
2343                  (when dir
2344                    (list "--homedir" dir))))
2345         (setq return (spam-check-bsfilter-headers score))))
2346     return))
2347
2348 (defun spam-bsfilter-register-with-bsfilter (articles
2349                                              spam
2350                                              &optional unregister)
2351   "Register an article, given as a string, as spam or non-spam."
2352   (dolist (article articles)
2353     (let ((article-string (spam-get-article-as-string article))
2354           (switch (if unregister
2355                       (if spam
2356                           spam-bsfilter-spam-strong-switch
2357                         spam-bsfilter-ham-strong-switch)
2358                     (if spam
2359                         spam-bsfilter-spam-switch
2360                       spam-bsfilter-ham-switch))))
2361       (when (stringp article-string)
2362         (with-temp-buffer
2363           (insert article-string)
2364           (apply 'call-process-region
2365                  (point-min) (point-max)
2366                  spam-bsfilter-path
2367                  nil nil nil switch
2368                  "--update"
2369                  (when spam-bsfilter-database-directory
2370                    (list "--homedir"
2371                          spam-bsfilter-database-directory))))))))
2372
2373 (defun spam-bsfilter-register-spam-routine (articles &optional unregister)
2374   (spam-bsfilter-register-with-bsfilter articles t unregister))
2375
2376 (defun spam-bsfilter-unregister-spam-routine (articles)
2377   (spam-bsfilter-register-spam-routine articles t))
2378
2379 (defun spam-bsfilter-register-ham-routine (articles &optional unregister)
2380   (spam-bsfilter-register-with-bsfilter articles nil unregister))
2381
2382 (defun spam-bsfilter-unregister-ham-routine (articles)
2383   (spam-bsfilter-register-ham-routine articles t))
2384
2385 \f
2386 ;;;; Hooks
2387
2388 ;;;###autoload
2389 (defun spam-initialize (&rest symbols)
2390   "Install the spam.el hooks and do other initialization.
2391 When SYMBOLS is given, set those variables to t.  This is so you
2392 can call spam-initialize before you set spam-use-* variables on
2393 explicitly, and matters only if you need the extra headers
2394 installed through spam-necessary-extra-headers."
2395   (interactive)
2396
2397   (dolist (var symbols)
2398     (set var t))
2399
2400   (dolist (header (spam-necessary-extra-headers))
2401     (add-to-list 'nnmail-extra-headers header)
2402     (add-to-list 'gnus-extra-headers header))
2403
2404   (setq spam-install-hooks t)
2405   ;; TODO: How do we redo this every time spam-face is customized?
2406   (push '((eq mark gnus-spam-mark) . spam-face)
2407         gnus-summary-highlight)
2408   ;; Add hooks for loading and saving the spam stats
2409   (add-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2410   (add-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2411   (add-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2412   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2413   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2414   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2415   (add-hook 'gnus-summary-prepared-hook 'spam-find-spam))
2416
2417 (defun spam-unload-hook ()
2418   "Uninstall the spam.el hooks"
2419   (interactive)
2420   (remove-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2421   (remove-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2422   (remove-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2423   (remove-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2424   (remove-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2425   (remove-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2426   (remove-hook 'gnus-summary-prepare-hook 'spam-find-spam))
2427
2428 (when spam-install-hooks
2429   (spam-initialize))
2430
2431 (provide 'spam)
2432
2433 ;;; spam.el ends here