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