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
 737                classification check t unregister-list))))))
 738
 739     ;; find all the spam processors applicable to this group
 740     (dolist (processor-param spam-list-of-processors)
 741       (let ((processor (nth 0 processor-param))
 742             (classification (nth 1 processor-param))
 743             (check (nth 2 processor-param)))
 744         (when (and (eq 'spam classification)
 745                    (spam-group-processor-p gnus-newsgroup-name processor))
 746           (spam-register-routine classification check))))
 747
 748     (unless (and spam-move-spam-nonspam-groups-only
 749                  (spam-group-spam-contents-p gnus-newsgroup-name))
 750       (gnus-message 5 "Marking spam as expired and moving it to %s"
 751                     (gnus-parameter-spam-process-destination
 752                      gnus-newsgroup-name))
 753       (spam-mark-spam-as-expired-and-move-routine
 754        (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
 755
 756     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
 757     ;; expire spam, in case the above did not expire them
 758     (gnus-message 5 "Marking spam as expired without moving it")
 759     (spam-mark-spam-as-expired-and-move-routine nil)
 760
 761     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
 762               (and (spam-group-spam-contents-p gnus-newsgroup-name)
 763                    spam-process-ham-in-spam-groups)
 764               spam-process-ham-in-nonham-groups)
 765       ;; find all the ham processors applicable to this group
 766       (dolist (processor-param spam-list-of-processors)
 767         (let ((processor (nth 0 processor-param))
 768               (classification (nth 1 processor-param))
 769               (check (nth 2 processor-param)))
 770           (when (and (eq 'ham classification)
 771                      (spam-group-processor-p gnus-newsgroup-name processor))
 772             (spam-register-routine classification check)))))
 773
 774     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
 775       (gnus-message 5 "Copying ham")
 776       (spam-ham-copy-routine
 777        (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
 778
 779     ;; now move all ham articles out of spam groups
 780     (when (spam-group-spam-contents-p gnus-newsgroup-name)
 781       (gnus-message 5 "Moving ham messages from spam group")
 782       (spam-ham-move-routine
 783        (gnus-parameter-ham-process-destination gnus-newsgroup-name))))
 784
 785   (setq spam-old-ham-articles nil)
 786   (setq spam-old-spam-articles nil))
 787
 788 (defun spam-set-difference (list1 list2)
 789   "Return a set difference of LIST1 and LIST2.
 790 When either list is nil, the other is returned."
 791   (if (and list1 list2)
 792       ;; we have two non-nil lists
 793       (progn
 794         (dolist (item (append list1 list2))
 795           (when (and (memq item list1) (memq item list2))
 796             (setq list1 (delq item list1))
 797             (setq list2 (delq item list2))))
 798         (append list1 list2))
 799     ;; if either of the lists was nil, return the other one
 800     (if list1 list1 list2)))
 801
 802 (defun spam-mark-junk-as-spam-routine ()
 803   ;; check the global list of group names spam-junk-mailgroups and the
 804   ;; group parameters
 805   (when (spam-group-spam-contents-p gnus-newsgroup-name)
 806     (gnus-message 5 "Marking %s articles as spam"
 807                   (if spam-mark-only-unseen-as-spam
 808                       "unseen"
 809                     "unread"))
 810     (let ((articles (if spam-mark-only-unseen-as-spam
 811                         gnus-newsgroup-unseen
 812                       gnus-newsgroup-unreads)))
 813       (dolist (article articles)
 814         (gnus-summary-mark-article article gnus-spam-mark)))))
 815
 816 (defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
 817   (if (and (car-safe groups) (listp (car-safe groups)))
 818       (apply 'spam-mark-spam-as-expired-and-move-routine (car groups))
 819     (gnus-summary-kill-process-mark)
 820     (let ((articles gnus-newsgroup-articles)
 821           (backend-supports-deletions
 822            (gnus-check-backend-function
 823             'request-move-article gnus-newsgroup-name))
 824           article tomove deletep)
 825       (dolist (article articles)
 826         (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
 827           (gnus-summary-mark-article article gnus-expirable-mark)
 828           (push article tomove)))
 829
 830       ;; now do the actual copies
 831       (dolist (group groups)
 832         (when (and tomove
 833                    (stringp group))
 834           (dolist (article tomove)
 835             (gnus-summary-set-process-mark article))
 836           (when tomove
 837             (if (or (not backend-supports-deletions)
 838                     (> (length groups) 1))
 839                 (progn
 840                   (gnus-summary-copy-article nil group)
 841                   (setq deletep t))
 842               (gnus-summary-move-article nil group)))))
 843
 844       ;; now delete the articles, if there was a copy done, and the
 845       ;; backend allows it
 846       (when (and deletep backend-supports-deletions)
 847         (dolist (article tomove)
 848           (gnus-summary-set-process-mark article))
 849         (when tomove
 850           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
 851             (gnus-summary-delete-article nil))))
 852
 853       (gnus-summary-yank-process-mark))))
 854
 855 (defun spam-ham-copy-or-move-routine (copy groups)
 856   (gnus-summary-kill-process-mark)
 857   (let ((todo (spam-list-articles gnus-newsgroup-articles 'ham))
 858         (backend-supports-deletions
 859          (gnus-check-backend-function
 860           'request-move-article gnus-newsgroup-name))
 861         (respool-method (gnus-find-method-for-group gnus-newsgroup-name))
 862         article mark todo deletep respool)
 863
 864     (when (member 'respool groups)
 865       (setq respool t)                  ; boolean for later
 866       (setq groups '("fake"))) ; when respooling, groups are dynamic so fake it
 867
 868     ;; now do the actual move
 869     (dolist (group groups)
 870       (when (and todo (stringp group))
 871         (dolist (article todo)
 872           (when spam-mark-ham-unread-before-move-from-spam-group
 873             (gnus-summary-mark-article article gnus-unread-mark))
 874           (gnus-summary-set-process-mark article))
 875
 876         (if respool                        ; respooling is with a "fake" group
 877             (let ((spam-split-disabled
 878                    (or spam-split-disabled
 879                        spam-disable-spam-split-during-ham-respool)))
 880               (gnus-summary-respool-article nil respool-method))
 881           (if (or (not backend-supports-deletions) ; else, we are not respooling
 882                   (> (length groups) 1))
 883               (progn                ; if copying, copy and set deletep
 884                 (gnus-summary-copy-article nil group)
 885                 (setq deletep t))
 886             (gnus-summary-move-article nil group))))) ; else move articles
 887
 888     ;; now delete the articles, unless a) copy is t, and there was a copy done
 889     ;;                                 b) a move was done to a single group
 890     ;;                                 c) backend-supports-deletions is nil
 891     (unless copy
 892       (when (and deletep backend-supports-deletions)
 893         (dolist (article todo)
 894           (gnus-summary-set-process-mark article))
 895         (when todo
 896           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
 897             (gnus-summary-delete-article nil))))))
 898
 899   (gnus-summary-yank-process-mark))
 900
 901 (defun spam-ham-copy-routine (&rest groups)
 902   (if (and (car-safe groups) (listp (car-safe groups)))
 903       (apply 'spam-ham-copy-routine (car groups))
 904     (spam-ham-copy-or-move-routine t groups)))
 905
 906 (defun spam-ham-move-routine (&rest groups)
 907   (if (and (car-safe groups) (listp (car-safe groups)))
 908       (apply 'spam-ham-move-routine (car groups))
 909     (spam-ham-copy-or-move-routine nil groups)))
 910
 911 (defun spam-get-article-as-string (article)
 912   (when (numberp article)
 913     (with-temp-buffer
 914       (gnus-request-article-this-buffer
 915        article
 916        gnus-newsgroup-name)
 917       (buffer-string))))
 918
 919 ;; disabled for now
 920 ;; (defun spam-get-article-as-filename (article)
 921 ;;   (let ((article-filename))
 922 ;;     (when (numberp article)
 923 ;;       (nnml-possibly-change-directory
 924 ;;        (gnus-group-real-name gnus-newsgroup-name))
 925 ;;       (setq article-filename (expand-file-name
 926 ;;                              (int-to-string article) nnml-current-directory)))
 927 ;;     (if (file-exists-p article-filename)
 928 ;;      article-filename
 929 ;;       nil)))
 930
 931 (defun spam-fetch-field-fast (article field &optional prepared-data-header)
 932   "Fetch a field quickly, using the internal gnus-data-list function"
 933   (when (numberp article)
 934     (let* ((data-header (or prepared-data-header
 935                             (spam-fetch-article-header article))))
 936       (if (arrayp data-header)
 937         (cond
 938          ((equal field 'from)
 939           (mail-header-from data-header))
 940          ((equal field 'message-id)
 941           (mail-header-message-id data-header))
 942          ((equal field 'subject)
 943           (mail-header-subject data-header))
 944          ((equal field 'references)
 945           (mail-header-references data-header))
 946          ((equal field 'date)
 947           (mail-header-date data-header))
 948          ((equal field 'xref)
 949           (mail-header-xref data-header))
 950          ((equal field 'extra)
 951           (mail-header-extra data-header))
 952          (t
 953           nil))
 954         (gnus-message 5 "Article %d has a nil data header" article)))))
 955
 956 (defun spam-fetch-field-from-fast (article &optional prepared-data-header)
 957   (spam-fetch-field-fast article 'from prepared-data-header))
 958
 959 (defun spam-fetch-field-subject-fast (article &optional prepared-data-header)
 960   (spam-fetch-field-fast article 'subject prepared-data-header))
 961
 962 (defun spam-fetch-field-message-id-fast (article &optional prepared-data-header)
 963   (spam-fetch-field-fast article 'message-id prepared-data-header))
 964
 965 (defun spam-generate-fake-headers (article)
 966   (let ((dh (spam-fetch-article-header article)))
 967     (if dh
 968         (concat
 969          (format
 970           (concat "From: %s\nSubject: %s\nMessage-ID: %s\n"
 971                   "Date: %s\nReferences: %s\nXref: %s\n")
 972           (spam-fetch-field-fast article 'from dh)
 973           (spam-fetch-field-fast article 'subject dh)
 974           (spam-fetch-field-fast article 'message-id dh)
 975           (spam-fetch-field-fast article 'date dh)
 976           (spam-fetch-field-fast article 'references dh)
 977           (spam-fetch-field-fast article 'xref dh))
 978          (when (spam-fetch-field-fast article 'extra dh)
 979            (format "%s\n" (spam-fetch-field-fast article 'extra dh))))
 980       (gnus-message
 981        5
 982        "spam-generate-fake-headers: article %d didn't have a valid header"
 983        article))))
 984
 985 (defun spam-fetch-article-header (article)
 986   (save-excursion
 987     (set-buffer gnus-summary-buffer)
 988     (gnus-read-header article)
 989     (nth 3 (assq article gnus-newsgroup-data))))
 990
 991 \f
 992 ;;;; Spam determination.
 993
 994 (defvar spam-list-of-checks
 995   '((spam-use-blacklist                 .       spam-check-blacklist)
 996     (spam-use-regex-headers             .       spam-check-regex-headers)
 997     (spam-use-gmane-xref                .       spam-check-gmane-xref)
 998     (spam-use-regex-body                .       spam-check-regex-body)
 999     (spam-use-whitelist                 .       spam-check-whitelist)
1000     (spam-use-BBDB                      .       spam-check-BBDB)
1001     (spam-use-BBDB-exclusive            .       spam-check-BBDB)
1002     (spam-use-ifile                     .       spam-check-ifile)
1003     (spam-use-spamoracle                .       spam-check-spamoracle)
1004     (spam-use-stat                      .       spam-check-stat)
1005     (spam-use-blackholes                .       spam-check-blackholes)
1006     (spam-use-hashcash                  .       spam-check-hashcash)
1007     (spam-use-spamassassin-headers      .       spam-check-spamassassin-headers)
1008     (spam-use-spamassassin              .       spam-check-spamassassin)
1009     (spam-use-bogofilter-headers        .       spam-check-bogofilter-headers)
1010     (spam-use-bogofilter                .       spam-check-bogofilter))
1011   "The spam-list-of-checks list contains pairs associating a
1012 parameter variable with a spam checking function.  If the
1013 parameter variable is true, then the checking function is called,
1014 and its value decides what happens.  Each individual check may
1015 return nil, t, or a mailgroup name.  The value nil means that the
1016 check does not yield a decision, and so, that further checks are
1017 needed.  The value t means that the message is definitely not
1018 spam, and that further spam checks should be inhibited.
1019 Otherwise, a mailgroup name or the symbol 'spam (depending on
1020 spam-split-symbolic-return) is returned where the mail should go,
1021 and further checks are also inhibited.  The usual mailgroup name
1022 is the value of `spam-split-group', meaning that the message is
1023 definitely a spam.")
1024
1025 (defvar spam-list-of-statistical-checks
1026   '(spam-use-ifile
1027     spam-use-regex-body
1028     spam-use-stat
1029     spam-use-bogofilter
1030     spam-use-blackholes
1031     spam-use-spamassassin
1032     spam-use-spamoracle)
1033   "The spam-list-of-statistical-checks list contains all the mail
1034 splitters that need to have the full message body available.
1035 Note that you should fetch extra headers if you don't like this,
1036 e.g. fetch the 'Received' header for spam-use-blackholes.")
1037
1038 (defun spam-split (&rest specific-checks)
1039   "Split this message into the `spam' group if it is spam.
1040 This function can be used as an entry in the variable `nnmail-split-fancy',
1041 for example like this: (: spam-split).  It can take checks as
1042 parameters.  A string as a parameter will set the
1043 spam-split-group to that string.
1044
1045 See the Info node `(gnus)Fancy Mail Splitting' for more details."
1046   (interactive)
1047   (setq spam-split-last-successful-check nil)
1048   (unless spam-split-disabled
1049     (let ((spam-split-group-choice spam-split-group))
1050       (dolist (check specific-checks)
1051         (when (stringp check)
1052           (setq spam-split-group-choice check)
1053           (setq specific-checks (delq check specific-checks))))
1054
1055       (let ((spam-split-group spam-split-group-choice))
1056         (save-excursion
1057           (save-restriction
1058             (dolist (check spam-list-of-statistical-checks)
1059               (when (and (symbolp check)
1060                          (or (symbol-value check)
1061                              (memq check specific-checks)))
1062                 (widen)
1063                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
1064                               (symbol-name check))
1065                 (return)))
1066             ;;   (progn (widen) (debug (buffer-string)))
1067             (let ((list-of-checks spam-list-of-checks)
1068                   decision)
1069               (while (and list-of-checks (not decision))
1070                 (let ((pair (pop list-of-checks)))
1071                   (when (or
1072                          ;; either, given specific checks, this is one of them
1073                          (and specific-checks (memq (car pair) specific-checks))
1074                          ;; or, given no specific checks, spam-use-CHECK is set
1075                          (and (null specific-checks) (symbol-value (car pair))))
1076                     (gnus-message 5 "spam-split: calling the %s function"
1077                                   (symbol-name (cdr pair)))
1078                     (setq decision (funcall (cdr pair)))
1079                     ;; if we got a decision at all, save the current check
1080                     (when decision
1081                       (setq spam-split-last-successful-check (car pair)))
1082
1083                     (when (eq decision 'spam)
1084                       (unless spam-split-symbolic-return
1085                         (gnus-error
1086                          5
1087                          (format "spam-split got %s but %s is nil"
1088                                  (symbol-name decision)
1089                                  (symbol-name spam-split-symbolic-return))))))))
1090               (if (eq decision t)
1091                   (if spam-split-symbolic-return-positive 'ham nil)
1092                 decision))))))))
1093
1094 (defun spam-find-spam ()
1095   "This function will detect spam in the current newsgroup using spam-split."
1096   (interactive)
1097
1098   (let* ((group gnus-newsgroup-name)
1099          (autodetect (gnus-parameter-spam-autodetect group))
1100          (methods (gnus-parameter-spam-autodetect-methods group))
1101          (first-method (nth 0 methods))
1102          (articles (if spam-autodetect-recheck-messages
1103                        gnus-newsgroup-articles
1104                      gnus-newsgroup-unseen))
1105          article-cannot-be-faked)
1106
1107     (dolist (check spam-list-of-statistical-checks)
1108       (when (and (symbolp check)
1109                  (memq check methods))
1110         (setq article-cannot-be-faked t)
1111         (return)))
1112
1113     (when (memq 'default methods)
1114       (setq article-cannot-be-faked t))
1115
1116     (when (and autodetect
1117                (not (equal first-method 'none)))
1118     (mapcar
1119      (lambda (article)
1120        (let ((id (spam-fetch-field-message-id-fast article))
1121              (subject (spam-fetch-field-subject-fast article))
1122              (sender (spam-fetch-field-from-fast article))
1123              registry-lookup)
1124
1125          (unless id
1126            (gnus-message 5 "Article %d has no message ID!" article))
1127
1128          (when (and id spam-log-to-registry)
1129            (setq registry-lookup (spam-log-registration-type id 'incoming))
1130            (when registry-lookup
1131              (gnus-message
1132               9
1133               "spam-find-spam: message %s was already registered incoming"
1134               id)))
1135
1136          (let* ((spam-split-symbolic-return t)
1137                 (spam-split-symbolic-return-positive t)
1138                 (fake-headers (spam-generate-fake-headers article))
1139                 (split-return
1140                  (or registry-lookup
1141                      (with-temp-buffer
1142                        (if article-cannot-be-faked
1143                            (gnus-request-article-this-buffer
1144                             article
1145                             group)
1146                          ;; else, we fake the article
1147                          (when fake-headers (insert fake-headers)))
1148                        (if (or (null first-method)
1149                                (equal first-method 'default))
1150                            (spam-split)
1151                          (apply 'spam-split methods))))))
1152            (if (equal split-return 'spam)
1153                (gnus-summary-mark-article article gnus-spam-mark))
1154
1155            (when (and id split-return spam-log-to-registry)
1156              (when (zerop (gnus-registry-group-count id))
1157                (gnus-registry-add-group
1158                 id group subject sender))
1159
1160              (unless registry-lookup
1161                (spam-log-processing-to-registry
1162                 id
1163                 'incoming
1164                 split-return
1165                 spam-split-last-successful-check
1166                 group))))))
1167     articles))))
1168
1169 (defvar spam-registration-functions
1170   ;; first the ham register, second the spam register function
1171   ;; third the ham unregister, fourth the spam unregister function
1172   '((spam-use-blacklist  nil
1173                          spam-blacklist-register-routine
1174                          nil
1175                          spam-blacklist-unregister-routine)
1176     (spam-use-whitelist  spam-whitelist-register-routine
1177                          nil
1178                          spam-whitelist-unregister-routine
1179                          nil)
1180     (spam-use-ham-copy   nil
1181                          nil
1182                          nil
1183                          nil)
1184     (spam-use-BBDB       spam-BBDB-register-routine
1185                          nil
1186                          spam-BBDB-unregister-routine
1187                          nil)
1188     (spam-use-ifile      spam-ifile-register-ham-routine
1189                          spam-ifile-register-spam-routine
1190                          spam-ifile-unregister-ham-routine
1191                          spam-ifile-unregister-spam-routine)
1192     (spam-use-spamoracle spam-spamoracle-learn-ham
1193                          spam-spamoracle-learn-spam
1194                          spam-spamoracle-unlearn-ham
1195                          spam-spamoracle-unlearn-spam)
1196     (spam-use-stat       spam-stat-register-ham-routine
1197                          spam-stat-register-spam-routine
1198                          spam-stat-unregister-ham-routine
1199                          spam-stat-unregister-spam-routine)
1200     ;; note that spam-use-gmane is not a legitimate check
1201     (spam-use-gmane      nil
1202                          spam-report-gmane-register-routine
1203                          ;; does Gmane support unregistration?
1204                          nil
1205                          nil)
1206     (spam-use-spamassassin spam-spamassassin-register-ham-routine
1207                            spam-spamassassin-register-spam-routine
1208                            spam-spamassassin-unregister-ham-routine
1209                            spam-spamassassin-unregister-spam-routine)
1210     (spam-use-bogofilter spam-bogofilter-register-ham-routine
1211                          spam-bogofilter-register-spam-routine
1212                          spam-bogofilter-unregister-ham-routine
1213                          spam-bogofilter-unregister-spam-routine))
1214   "The spam-registration-functions list contains pairs
1215 associating a parameter variable with the ham and spam
1216 registration functions, and the ham and spam unregistration
1217 functions")
1218
1219 (defun spam-classification-valid-p (classification)
1220   (or  (eq classification 'spam)
1221        (eq classification 'ham)))
1222
1223 (defun spam-process-type-valid-p (process-type)
1224   (or  (eq process-type 'incoming)
1225        (eq process-type 'process)))
1226
1227 (defun spam-registration-check-valid-p (check)
1228   (assoc check spam-registration-functions))
1229
1230 (defun spam-unregistration-check-valid-p (check)
1231   (assoc check spam-registration-functions))
1232
1233 (defun spam-registration-function (classification check)
1234   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1235     (if (eq classification 'spam)
1236         (nth 1 flist)
1237       (nth 0 flist))))
1238
1239 (defun spam-unregistration-function (classification check)
1240   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1241     (if (eq classification 'spam)
1242         (nth 3 flist)
1243       (nth 2 flist))))
1244
1245 (defun spam-list-articles (articles classification)
1246   (let ((mark-check (if (eq classification 'spam)
1247                         'spam-group-spam-mark-p
1248                       'spam-group-ham-mark-p))
1249         alist mark-cache-yes mark-cache-no)
1250     (dolist (article articles)
1251       (let ((mark (gnus-summary-article-mark article)))
1252         (unless (or (memq mark mark-cache-yes)
1253                     (memq mark mark-cache-no))
1254           (if (funcall mark-check
1255                        gnus-newsgroup-name
1256                        mark)
1257               (push mark mark-cache-yes)
1258             (push mark mark-cache-no)))
1259         (when (memq mark mark-cache-yes)
1260           (push article alist))))
1261     alist))
1262
1263 (defun spam-register-routine (classification
1264                               check
1265                               &optional unregister
1266                               specific-articles)
1267   (when (and (spam-classification-valid-p classification)
1268              (spam-registration-check-valid-p check))
1269     (let* ((register-function
1270             (spam-registration-function classification check))
1271            (unregister-function
1272             (spam-unregistration-function classification check))
1273            (run-function (if unregister
1274                              unregister-function
1275                            register-function))
1276            (log-function (if unregister
1277                              'spam-log-undo-registration
1278                            'spam-log-processing-to-registry))
1279            article articles)
1280
1281       (when run-function
1282         ;; make list of articles, using specific-articles if given
1283         (setq articles (or specific-articles
1284                            (spam-list-articles
1285                             gnus-newsgroup-articles
1286                             classification)))
1287         ;; process them
1288         (gnus-message 5 "%s %d %s articles as %s using backend %s"
1289                       (if unregister "Unregistering" "Registering")
1290                       (length articles)
1291                       (if specific-articles "specific" "")
1292                       (symbol-name classification)
1293                       (symbol-name check))
1294         (funcall run-function articles)
1295         ;; now log all the registrations (or undo them, depending on unregister)
1296         (dolist (article articles)
1297           (funcall log-function
1298                    (spam-fetch-field-message-id-fast article)
1299                    'process
1300                    classification
1301                    check
1302                    gnus-newsgroup-name))))))
1303
1304 ;;; log a ham- or spam-processor invocation to the registry
1305 (defun spam-log-processing-to-registry (id type classification check group)
1306   (when spam-log-to-registry
1307     (if (and (stringp id)
1308              (stringp group)
1309              (spam-process-type-valid-p type)
1310              (spam-classification-valid-p classification)
1311              (spam-registration-check-valid-p check))
1312         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1313               (cell (list classification check group)))
1314           (push cell cell-list)
1315           (gnus-registry-store-extra-entry
1316            id
1317            type
1318            cell-list))
1319
1320       (gnus-message
1321        5
1322        (format "%s call with bad ID, type, classification, spam-check, or group"
1323                "spam-log-processing-to-registry")))))
1324
1325 ;;; check if a ham- or spam-processor registration has been done
1326 (defun spam-log-registered-p (id type)
1327   (when spam-log-to-registry
1328     (if (and (stringp id)
1329              (spam-process-type-valid-p type))
1330         (cdr-safe (gnus-registry-fetch-extra id type))
1331       (progn
1332         (gnus-message
1333          5
1334          (format "%s called with bad ID, type, classification, or spam-check"
1335                  "spam-log-registered-p"))
1336         nil))))
1337
1338 ;;; check what a ham- or spam-processor registration says
1339 ;;; returns nil if conflicting registrations are found
1340 (defun spam-log-registration-type (id type)
1341   (let ((count 0)
1342         decision)
1343     (dolist (reg (spam-log-registered-p id type))
1344       (let ((classification (nth 0 reg)))
1345         (when (spam-classification-valid-p classification)
1346           (when (and decision
1347                      (not (eq classification decision)))
1348             (setq count (+ 1 count)))
1349           (setq decision classification))))
1350     (if (< 0 count)
1351         nil
1352       decision)))
1353
1354
1355 ;;; check if a ham- or spam-processor registration needs to be undone
1356 (defun spam-log-unregistration-needed-p (id type classification check)
1357   (when spam-log-to-registry
1358     (if (and (stringp id)
1359              (spam-process-type-valid-p type)
1360              (spam-classification-valid-p classification)
1361              (spam-registration-check-valid-p check))
1362         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1363               found)
1364           (dolist (cell cell-list)
1365             (unless found
1366               (when (and (eq classification (nth 0 cell))
1367                          (eq check (nth 1 cell)))
1368                 (setq found t))))
1369           found)
1370       (progn
1371         (gnus-message
1372          5
1373          (format "%s called with bad ID, type, classification, or spam-check"
1374                  "spam-log-unregistration-needed-p"))
1375         nil))))
1376
1377
1378 ;;; undo a ham- or spam-processor registration (the group is not used)
1379 (defun spam-log-undo-registration (id type classification check &optional group)
1380   (when (and spam-log-to-registry
1381              (spam-log-unregistration-needed-p id type classification check))
1382     (if (and (stringp id)
1383              (spam-process-type-valid-p type)
1384              (spam-classification-valid-p classification)
1385              (spam-registration-check-valid-p check))
1386         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1387               new-cell-list found)
1388           (dolist (cell cell-list)
1389             (unless (and (eq classification (nth 0 cell))
1390                          (eq check (nth 1 cell)))
1391               (push cell new-cell-list)))
1392           (gnus-registry-store-extra-entry
1393            id
1394            type
1395            new-cell-list))
1396       (progn
1397         (gnus-message 5 (format "%s call with bad ID, type, spam-check, or group"
1398                                 "spam-log-undo-registration"))
1399         nil))))
1400
1401 ;;; set up IMAP widening if it's necessary
1402 (defun spam-setup-widening ()
1403   (dolist (check spam-list-of-statistical-checks)
1404     (when (symbol-value check)
1405       (setq nnimap-split-download-body-default t))))
1406
1407 \f
1408 ;;;; Gmane xrefs
1409 (defun spam-check-gmane-xref ()
1410   (let ((header (or
1411                  (message-fetch-field "Xref")
1412                  (message-fetch-field "Newsgroups")))
1413         (spam-split-group (if spam-split-symbolic-return
1414                               'spam
1415                             spam-split-group)))
1416     (when header                        ; return nil when no header
1417       (when (string-match spam-gmane-xref-spam-group
1418                           header)
1419           spam-split-group))))
1420
1421 \f
1422 ;;;; Regex body
1423
1424 (defun spam-check-regex-body ()
1425   (let ((spam-regex-headers-ham spam-regex-body-ham)
1426         (spam-regex-headers-spam spam-regex-body-spam))
1427     (spam-check-regex-headers t)))
1428
1429 \f
1430 ;;;; Regex headers
1431
1432 (defun spam-check-regex-headers (&optional body)
1433   (let ((type (if body "body" "header"))
1434         (spam-split-group (if spam-split-symbolic-return
1435                               'spam
1436                             spam-split-group))
1437         ret found)
1438     (dolist (h-regex spam-regex-headers-ham)
1439       (unless found
1440         (goto-char (point-min))
1441         (when (re-search-forward h-regex nil t)
1442           (message "Ham regex %s search positive." type)
1443           (setq found t))))
1444     (dolist (s-regex spam-regex-headers-spam)
1445       (unless found
1446         (goto-char (point-min))
1447         (when (re-search-forward s-regex nil t)
1448           (message "Spam regex %s search positive." type)
1449           (setq found t)
1450           (setq ret spam-split-group))))
1451     ret))
1452
1453 \f
1454 ;;;; Blackholes.
1455
1456 (defun spam-reverse-ip-string (ip)
1457   (when (stringp ip)
1458     (mapconcat 'identity
1459                (nreverse (split-string ip "\\."))
1460                ".")))
1461
1462 (defun spam-check-blackholes ()
1463   "Check the Received headers for blackholed relays."
1464   (let ((headers (message-fetch-field "received"))
1465         (spam-split-group (if spam-split-symbolic-return
1466                               'spam
1467                             spam-split-group))
1468         ips matches)
1469     (when headers
1470       (with-temp-buffer
1471         (insert headers)
1472         (goto-char (point-min))
1473         (gnus-message 5 "Checking headers for relay addresses")
1474         (while (re-search-forward
1475                 "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
1476           (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
1477           (push (spam-reverse-ip-string (match-string 1))
1478                 ips)))
1479       (dolist (server spam-blackhole-servers)
1480         (dolist (ip ips)
1481           (unless (and spam-blackhole-good-server-regex
1482                        ;; match the good-server-regex against the reversed (again) IP string
1483                        (string-match
1484                         spam-blackhole-good-server-regex
1485                         (spam-reverse-ip-string ip)))
1486             (unless matches
1487               (let ((query-string (concat ip "." server)))
1488                 (if spam-use-dig
1489                     (let ((query-result (query-dig query-string)))
1490                       (when query-result
1491                         (gnus-message 5 "(DIG): positive blackhole check '%s'"
1492                                       query-result)
1493                         (push (list ip server query-result)
1494                               matches)))
1495                   ;; else, if not using dig.el
1496                   (when (query-dns query-string)
1497                     (gnus-message 5 "positive blackhole check")
1498                     (push (list ip server (query-dns query-string 'TXT))
1499                           matches)))))))))
1500     (when matches
1501       spam-split-group)))
1502 \f
1503 ;;;; Hashcash.
1504
1505 (condition-case nil
1506     (progn
1507       (require 'hashcash)
1508
1509       (defun spam-check-hashcash ()
1510         "Check the headers for hashcash payments."
1511         (mail-check-payment)))   ;mail-check-payment returns a boolean
1512
1513   (file-error (progn
1514                 (defalias 'mail-check-payment 'ignore)
1515                 (defalias 'spam-check-hashcash 'ignore))))
1516 \f
1517 ;;;; BBDB
1518
1519 ;;; original idea for spam-check-BBDB from Alexander Kotelnikov
1520 ;;; <sacha@giotto.sj.ru>
1521
1522 ;; all this is done inside a condition-case to trap errors
1523
1524 (condition-case nil
1525     (progn
1526       (require 'bbdb)
1527       (require 'bbdb-com)
1528
1529       ;; when the BBDB changes, we want to clear out our cache
1530       (defun spam-clear-cache-BBDB (&rest immaterial)
1531         (spam-clear-cache 'spam-use-BBDB))
1532
1533       (add-hook 'bbdb-change-hook 'spam-clear-cache-BBDB)
1534
1535       (defun spam-enter-ham-BBDB (addresses &optional remove)
1536         "Enter an address into the BBDB; implies ham (non-spam) sender"
1537         (dolist (from addresses)
1538           (when (stringp from)
1539             (let* ((parsed-address (gnus-extract-address-components from))
1540                    (name (or (nth 0 parsed-address) "Ham Sender"))
1541                    (remove-function (if remove
1542                                         'bbdb-delete-record-internal
1543                                       'ignore))
1544                    (net-address (nth 1 parsed-address))
1545                    (record (and net-address
1546                                 (bbdb-search-simple nil net-address))))
1547               (when net-address
1548                 (gnus-message 5 "%s address %s %s BBDB"
1549                               (if remove "Deleting" "Adding")
1550                               from
1551                               (if remove "from" "to"))
1552                 (if record
1553                     (funcall remove-function record)
1554                   (bbdb-create-internal name nil net-address nil nil
1555                                         "ham sender added by spam.el")))))))
1556
1557       (defun spam-BBDB-register-routine (articles &optional unregister)
1558         (let (addresses)
1559           (dolist (article articles)
1560             (when (stringp (spam-fetch-field-from-fast article))
1561               (push (spam-fetch-field-from-fast article) addresses)))
1562           ;; now do the register/unregister action
1563           (spam-enter-ham-BBDB addresses unregister)))
1564
1565       (defun spam-BBDB-unregister-routine (articles)
1566         (spam-BBDB-register-routine articles t))
1567
1568       (defun spam-check-BBDB ()
1569         "Mail from people in the BBDB is classified as ham or non-spam"
1570         (let ((who (message-fetch-field "from"))
1571               (spam-split-group (if spam-split-symbolic-return
1572                                     'spam
1573                                   spam-split-group))
1574               bbdb-cache bbdb-hashtable)
1575           (when spam-cache-lookups
1576             (setq bbdb-cache (gethash 'spam-use-BBDB spam-caches))
1577             (unless bbdb-cache
1578               (setq bbdb-cache
1579                     ;; this is the expanded (bbdb-hashtable) macro
1580                     ;; without the debugging support
1581                     (with-current-buffer (bbdb-buffer)
1582                       (save-excursion
1583                         (save-window-excursion
1584                           (bbdb-records nil t)
1585                           bbdb-hashtable))))
1586               (puthash 'spam-use-BBDB bbdb-cache spam-caches)))
1587           (when who
1588             (setq who (nth 1 (gnus-extract-address-components who)))
1589             (if
1590                 (if spam-cache-lookups
1591                     (symbol-value
1592                      (intern-soft who bbdb-cache))
1593                   (bbdb-search-simple nil who))
1594                 t
1595               (if spam-use-BBDB-exclusive
1596                   spam-split-group
1597                 nil))))))
1598
1599   (file-error (progn
1600                 (defalias 'bbdb-search-simple 'ignore)
1601                 (defalias 'bbdb-records 'ignore)
1602                 (defalias 'bbdb-buffer 'ignore)
1603                 (defalias 'spam-check-BBDB 'ignore)
1604                 (defalias 'spam-BBDB-register-routine 'ignore)
1605                 (defalias 'spam-enter-ham-BBDB 'ignore)
1606                 (defalias 'bbdb-create-internal 'ignore)
1607                 (defalias 'bbdb-delete-record-internal 'ignore)
1608                 (defalias 'bbdb-records 'ignore))))
1609
1610 \f
1611 ;;;; ifile
1612
1613 ;;; check the ifile backend; return nil if the mail was NOT classified
1614 ;;; as spam
1615
1616 (defun spam-get-ifile-database-parameter ()
1617   "Get the command-line parameter for ifile's database from
1618   spam-ifile-database-path."
1619   (if spam-ifile-database-path
1620       (format "--db-file=%s" spam-ifile-database-path)
1621     nil))
1622
1623 (defun spam-check-ifile ()
1624   "Check the ifile backend for the classification of this message."
1625   (let ((article-buffer-name (buffer-name))
1626         (spam-split-group (if spam-split-symbolic-return
1627                               'spam
1628                             spam-split-group))
1629         category return)
1630     (with-temp-buffer
1631       (let ((temp-buffer-name (buffer-name))
1632             (db-param (spam-get-ifile-database-parameter)))
1633         (save-excursion
1634           (set-buffer article-buffer-name)
1635           (apply 'call-process-region
1636                  (point-min) (point-max) spam-ifile-path
1637                  nil temp-buffer-name nil "-c"
1638                  (if db-param `(,db-param "-q") `("-q"))))
1639         ;; check the return now (we're back in the temp buffer)
1640         (goto-char (point-min))
1641         (if (not (eobp))
1642             (setq category (buffer-substring (point) (point-at-eol))))
1643         (when (not (zerop (length category))) ; we need a category here
1644           (if spam-ifile-all-categories
1645               (setq return category)
1646             ;; else, if spam-ifile-all-categories is not set...
1647             (when (string-equal spam-ifile-spam-category category)
1648               (setq return spam-split-group)))))) ; note return is nil otherwise
1649     return))
1650
1651 (defun spam-ifile-register-with-ifile (articles category &optional unregister)
1652   "Register an article, given as a string, with a category.
1653 Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
1654   (let ((category (or category gnus-newsgroup-name))
1655         (add-or-delete-option (if unregister "-d" "-i"))
1656         (db (spam-get-ifile-database-parameter))
1657         parameters)
1658     (with-temp-buffer
1659       (dolist (article articles)
1660         (let ((article-string (spam-get-article-as-string article)))
1661           (when (stringp article-string)
1662             (insert article-string))))
1663       (apply 'call-process-region
1664              (point-min) (point-max) spam-ifile-path
1665              nil nil nil
1666              add-or-delete-option category
1667              (if db `(,db "-h") `("-h"))))))
1668
1669 (defun spam-ifile-register-spam-routine (articles &optional unregister)
1670   (spam-ifile-register-with-ifile articles spam-ifile-spam-category unregister))
1671
1672 (defun spam-ifile-unregister-spam-routine (articles)
1673   (spam-ifile-register-spam-routine articles t))
1674
1675 (defun spam-ifile-register-ham-routine (articles &optional unregister)
1676   (spam-ifile-register-with-ifile articles spam-ifile-ham-category unregister))
1677
1678 (defun spam-ifile-unregister-ham-routine (articles)
1679   (spam-ifile-register-ham-routine articles t))
1680
1681 \f
1682 ;;;; spam-stat
1683
1684 (condition-case nil
1685     (progn
1686       (let ((spam-stat-install-hooks nil))
1687         (require 'spam-stat))
1688
1689       (defun spam-check-stat ()
1690         "Check the spam-stat backend for the classification of this message"
1691         (let ((spam-split-group (if spam-split-symbolic-return
1692                                     'spam
1693                                   spam-split-group))
1694               (spam-stat-split-fancy-spam-group spam-split-group) ; override
1695               (spam-stat-buffer (buffer-name)) ; stat the current buffer
1696               category return)
1697           (spam-stat-split-fancy)))
1698
1699       (defun spam-stat-register-spam-routine (articles &optional unregister)
1700         (dolist (article articles)
1701           (let ((article-string (spam-get-article-as-string article)))
1702             (with-temp-buffer
1703               (insert article-string)
1704               (if unregister
1705                   (spam-stat-buffer-change-to-non-spam)
1706               (spam-stat-buffer-is-spam))))))
1707
1708       (defun spam-stat-unregister-spam-routine (articles)
1709         (spam-stat-register-spam-routine articles t))
1710
1711       (defun spam-stat-register-ham-routine (articles &optional unregister)
1712         (dolist (article articles)
1713           (let ((article-string (spam-get-article-as-string article)))
1714             (with-temp-buffer
1715               (insert article-string)
1716               (if unregister
1717                   (spam-stat-buffer-change-to-spam)
1718               (spam-stat-buffer-is-non-spam))))))
1719
1720       (defun spam-stat-unregister-ham-routine (articles)
1721         (spam-stat-register-ham-routine articles t))
1722
1723       (defun spam-maybe-spam-stat-load ()
1724         (when spam-use-stat (spam-stat-load)))
1725
1726       (defun spam-maybe-spam-stat-save ()
1727         (when spam-use-stat (spam-stat-save))))
1728
1729   (file-error (progn
1730                 (defalias 'spam-stat-load 'ignore)
1731                 (defalias 'spam-stat-save 'ignore)
1732                 (defalias 'spam-maybe-spam-stat-load 'ignore)
1733                 (defalias 'spam-maybe-spam-stat-save 'ignore)
1734                 (defalias 'spam-stat-register-ham-routine 'ignore)
1735                 (defalias 'spam-stat-unregister-ham-routine 'ignore)
1736                 (defalias 'spam-stat-register-spam-routine 'ignore)
1737                 (defalias 'spam-stat-unregister-spam-routine 'ignore)
1738                 (defalias 'spam-stat-buffer-is-spam 'ignore)
1739                 (defalias 'spam-stat-buffer-change-to-spam 'ignore)
1740                 (defalias 'spam-stat-buffer-is-non-spam 'ignore)
1741                 (defalias 'spam-stat-buffer-change-to-non-spam 'ignore)
1742                 (defalias 'spam-stat-split-fancy 'ignore)
1743                 (defalias 'spam-check-stat 'ignore))))
1744
1745 \f
1746
1747 ;;;; Blacklists and whitelists.
1748
1749 (defvar spam-whitelist-cache nil)
1750 (defvar spam-blacklist-cache nil)
1751
1752 (defun spam-kill-whole-line ()
1753   (beginning-of-line)
1754   (let ((kill-whole-line t))
1755     (kill-line)))
1756
1757 ;;; address can be a list, too
1758 (defun spam-enter-whitelist (address &optional remove)
1759   "Enter ADDRESS (list or single) into the whitelist.
1760 With a non-nil REMOVE, remove them."
1761   (interactive "sAddress: ")
1762   (spam-enter-list address spam-whitelist remove)
1763   (setq spam-whitelist-cache nil)
1764   (spam-clear-cache 'spam-use-whitelist))
1765
1766 ;;; address can be a list, too
1767 (defun spam-enter-blacklist (address &optional remove)
1768   "Enter ADDRESS (list or single) into the blacklist.
1769 With a non-nil REMOVE, remove them."
1770   (interactive "sAddress: ")
1771   (spam-enter-list address spam-blacklist remove)
1772   (setq spam-blacklist-cache nil)
1773   (spam-clear-cache 'spam-use-whitelist))
1774
1775 (defun spam-enter-list (addresses file &optional remove)
1776   "Enter ADDRESSES into the given FILE.
1777 Either the whitelist or the blacklist files can be used.  With
1778 REMOVE not nil, remove the ADDRESSES."
1779   (if (stringp addresses)
1780       (spam-enter-list (list addresses) file remove)
1781     ;; else, we have a list of addresses here
1782     (unless (file-exists-p (file-name-directory file))
1783       (make-directory (file-name-directory file) t))
1784     (save-excursion
1785       (set-buffer
1786        (find-file-noselect file))
1787       (dolist (a addresses)
1788         (when (stringp a)
1789           (goto-char (point-min))
1790           (if (re-search-forward (regexp-quote a) nil t)
1791               ;; found the address
1792               (when remove
1793                 (spam-kill-whole-line))
1794             ;; else, the address was not found
1795             (unless remove
1796               (goto-char (point-max))
1797               (unless (bobp)
1798                 (insert "\n"))
1799               (insert a "\n")))))
1800       (save-buffer))))
1801
1802 (defun spam-filelist-build-cache (type)
1803   (let ((cache (if (eq type 'spam-use-blacklist)
1804                    spam-blacklist-cache
1805                  spam-whitelist-cache))
1806         parsed-cache)
1807     (unless (gethash type spam-caches)
1808       (while cache
1809         (let ((address (pop cache)))
1810           (unless (zerop (length address)) ; 0 for a nil address too
1811             (setq address (regexp-quote address))
1812             ;; fix regexp-quote's treatment of user-intended regexes
1813             (while (string-match "\\\\\\*" address)
1814               (setq address (replace-match ".*" t t address))))
1815           (push address parsed-cache)))
1816       (puthash type parsed-cache spam-caches))))
1817
1818 (defun spam-filelist-check-cache (type from)
1819   (when (stringp from)
1820     (spam-filelist-build-cache type)
1821     (let (found)
1822       (dolist (address (gethash type spam-caches))
1823         (when (and address (string-match address from))
1824           (setq found t)
1825           (return)))
1826       found)))
1827
1828 ;;; returns t if the sender is in the whitelist, nil or
1829 ;;; spam-split-group otherwise
1830 (defun spam-check-whitelist ()
1831   ;; FIXME!  Should it detect when file timestamps change?
1832   (let ((spam-split-group (if spam-split-symbolic-return
1833                               'spam
1834                             spam-split-group)))
1835     (unless spam-whitelist-cache
1836       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
1837     (if (spam-from-listed-p 'spam-use-whitelist)
1838         t
1839       (if spam-use-whitelist-exclusive
1840           spam-split-group
1841         nil))))
1842
1843 (defun spam-check-blacklist ()
1844   ;; FIXME!  Should it detect when file timestamps change?
1845   (let ((spam-split-group (if spam-split-symbolic-return
1846                               'spam
1847                             spam-split-group)))
1848     (unless spam-blacklist-cache
1849       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
1850     (and (spam-from-listed-p 'spam-use-blacklist) spam-split-group)))
1851
1852 (defun spam-parse-list (file)
1853   (when (file-readable-p file)
1854     (let (contents address)
1855       (with-temp-buffer
1856         (insert-file-contents file)
1857         (while (not (eobp))
1858           (setq address (buffer-substring (point) (point-at-eol)))
1859           (forward-line 1)
1860           ;; insert the e-mail address if detected, otherwise the raw data
1861           (unless (zerop (length address))
1862             (let ((pure-address (nth 1 (gnus-extract-address-components address))))
1863               (push (or pure-address address) contents)))))
1864       (nreverse contents))))
1865
1866 (defun spam-from-listed-p (type)
1867   (let ((from (message-fetch-field "from"))
1868         found)
1869     (spam-filelist-check-cache type from)))
1870
1871 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
1872   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
1873         (declassification (if blacklist 'ham 'spam))
1874         (enter-function
1875          (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
1876         (remove-function
1877          (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
1878         from addresses unregister-list article-unregister-list)
1879     (dolist (article articles)
1880       (let ((from (spam-fetch-field-from-fast article))
1881             (id (spam-fetch-field-message-id-fast article))
1882             sender-ignored)
1883         (when (stringp from)
1884           (dolist (ignore-regex spam-blacklist-ignored-regexes)
1885             (when (and (not sender-ignored)
1886                        (stringp ignore-regex)
1887                        (string-match ignore-regex from))
1888               (setq sender-ignored t)))
1889           ;; remember the messages we need to unregister, unless remove is set
1890           (when (and
1891                  (null unregister)
1892                  (spam-log-unregistration-needed-p
1893                   id 'process declassification de-symbol))
1894             (push article article-unregister-list)
1895             (push from unregister-list))
1896           (unless sender-ignored
1897             (push from addresses)))))
1898
1899     (if unregister
1900         (funcall enter-function addresses t) ; unregister all these addresses
1901       ;; else, register normally and unregister what we need to
1902       (funcall remove-function unregister-list t)
1903       (dolist (article article-unregister-list)
1904         (spam-log-undo-registration
1905          (spam-fetch-field-message-id-fast article)
1906          'process
1907          declassification
1908          de-symbol))
1909       (funcall enter-function addresses nil))))
1910
1911 (defun spam-blacklist-unregister-routine (articles)
1912   (spam-blacklist-register-routine articles t))
1913
1914 (defun spam-blacklist-register-routine (articles &optional unregister)
1915   (spam-filelist-register-routine articles t unregister))
1916
1917 (defun spam-whitelist-unregister-routine (articles)
1918   (spam-whitelist-register-routine articles t))
1919
1920 (defun spam-whitelist-register-routine (articles &optional unregister)
1921   (spam-filelist-register-routine articles nil unregister))
1922
1923 \f
1924 ;;;; Spam-report glue
1925 (defun spam-report-gmane-register-routine (articles)
1926   (when articles
1927     (apply 'spam-report-gmane articles)))
1928
1929 \f
1930 ;;;; Bogofilter
1931 (defun spam-check-bogofilter-headers (&optional score)
1932   (let ((header (message-fetch-field spam-bogofilter-header))
1933         (spam-split-group (if spam-split-symbolic-return
1934                               'spam
1935                             spam-split-group)))
1936     (when header                        ; return nil when no header
1937       (if score                         ; scoring mode
1938           (if (string-match "spamicity=\\([0-9.]+\\)" header)
1939               (match-string 1 header)
1940             "0")
1941         ;; spam detection mode
1942         (when (string-match spam-bogofilter-bogosity-positive-spam-header
1943                             header)
1944           spam-split-group)))))
1945
1946 ;; return something sensible if the score can't be determined
1947 (defun spam-bogofilter-score ()
1948   "Get the Bogofilter spamicity score"
1949   (interactive)
1950   (save-window-excursion
1951     (gnus-summary-show-article t)
1952     (set-buffer gnus-article-buffer)
1953     (let ((score (or (spam-check-bogofilter-headers t)
1954                      (spam-check-bogofilter t))))
1955       (gnus-summary-show-article)
1956       (message "Spamicity score %s" score)
1957       (or score "0"))))
1958
1959 (defun spam-check-bogofilter (&optional score)
1960   "Check the Bogofilter backend for the classification of this message"
1961   (let ((article-buffer-name (buffer-name))
1962         (db spam-bogofilter-database-directory)
1963         return)
1964     (with-temp-buffer
1965       (let ((temp-buffer-name (buffer-name)))
1966         (save-excursion
1967           (set-buffer article-buffer-name)
1968           (apply 'call-process-region
1969                  (point-min) (point-max)
1970                  spam-bogofilter-path
1971                  nil temp-buffer-name nil
1972                  (if db `("-d" ,db "-v") `("-v"))))
1973         (setq return (spam-check-bogofilter-headers score))))
1974     return))
1975
1976 (defun spam-bogofilter-register-with-bogofilter (articles
1977                                                  spam
1978                                                  &optional unregister)
1979   "Register an article, given as a string, as spam or non-spam."
1980   (dolist (article articles)
1981     (let ((article-string (spam-get-article-as-string article))
1982           (db spam-bogofilter-database-directory)
1983           (switch (if unregister
1984                       (if spam
1985                           spam-bogofilter-spam-strong-switch
1986                         spam-bogofilter-ham-strong-switch)
1987                     (if spam
1988                         spam-bogofilter-spam-switch
1989                       spam-bogofilter-ham-switch))))
1990       (when (stringp article-string)
1991         (with-temp-buffer
1992           (insert article-string)
1993
1994           (apply 'call-process-region
1995                  (point-min) (point-max)
1996                  spam-bogofilter-path
1997                  nil nil nil switch
1998                  (if db `("-d" ,db "-v") `("-v"))))))))
1999
2000 (defun spam-bogofilter-register-spam-routine (articles &optional unregister)
2001   (spam-bogofilter-register-with-bogofilter articles t unregister))
2002
2003 (defun spam-bogofilter-unregister-spam-routine (articles)
2004   (spam-bogofilter-register-spam-routine articles t))
2005
2006 (defun spam-bogofilter-register-ham-routine (articles &optional unregister)
2007   (spam-bogofilter-register-with-bogofilter articles nil unregister))
2008
2009 (defun spam-bogofilter-unregister-ham-routine (articles)
2010   (spam-bogofilter-register-ham-routine articles t))
2011
2012
2013 \f
2014 ;;;; spamoracle
2015 (defun spam-check-spamoracle ()
2016   "Run spamoracle on an article to determine whether it's spam."
2017   (let ((article-buffer-name (buffer-name))
2018         (spam-split-group (if spam-split-symbolic-return
2019                               'spam
2020                             spam-split-group)))
2021     (with-temp-buffer
2022       (let ((temp-buffer-name (buffer-name)))
2023         (save-excursion
2024           (set-buffer article-buffer-name)
2025           (let ((status
2026                  (apply 'call-process-region
2027                         (point-min) (point-max)
2028                         spam-spamoracle-binary
2029                         nil temp-buffer-name nil
2030                         (if spam-spamoracle-database
2031                             `("-f" ,spam-spamoracle-database "mark")
2032                           '("mark")))))
2033             (if (eq 0 status)
2034                 (progn
2035                   (set-buffer temp-buffer-name)
2036                   (goto-char (point-min))
2037                   (when (re-search-forward "^X-Spam: yes;" nil t)
2038                     spam-split-group))
2039               (error "Error running spamoracle: %s" status))))))))
2040
2041 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
2042   "Run spamoracle in training mode."
2043   (with-temp-buffer
2044     (let ((temp-buffer-name (buffer-name)))
2045       (save-excursion
2046         (goto-char (point-min))
2047         (dolist (article articles)
2048           (insert (spam-get-article-as-string article)))
2049         (let* ((arg (if (spam-xor unregister article-is-spam-p)
2050                         "-spam"
2051                       "-good"))
2052                (status
2053                 (apply 'call-process-region
2054                        (point-min) (point-max)
2055                        spam-spamoracle-binary
2056                        nil temp-buffer-name nil
2057                        (if spam-spamoracle-database
2058                            `("-f" ,spam-spamoracle-database
2059                              "add" ,arg)
2060                          `("add" ,arg)))))
2061           (unless (eq 0 status)
2062             (error "Error running spamoracle: %s" status)))))))
2063
2064 (defun spam-spamoracle-learn-ham (articles &optional unregister)
2065   (spam-spamoracle-learn articles nil unregister))
2066
2067 (defun spam-spamoracle-unlearn-ham (articles &optional unregister)
2068   (spam-spamoracle-learn-ham articles t))
2069
2070 (defun spam-spamoracle-learn-spam (articles &optional unregister)
2071   (spam-spamoracle-learn articles t unregister))
2072
2073 (defun spam-spamoracle-unlearn-spam (articles &optional unregister)
2074   (spam-spamoracle-learn-spam articles t))
2075
2076 \f
2077 ;;;; SpamAssassin
2078 ;;; based mostly on the bogofilter code
2079 (defun spam-check-spamassassin-headers (&optional score)
2080   "Check the SpamAssassin headers for the classification of this message."
2081   (if score                             ; scoring mode
2082       (let ((header (message-fetch-field spam-spamassassin-spam-status-header)))
2083         (when header
2084           (if (string-match "hits=\\(-?[0-9.]+\\)" header)
2085               (match-string 1 header)
2086             "0")))
2087     ;; spam detection mode
2088     (let ((header (message-fetch-field spam-spamassassin-spam-flag-header))
2089           (spam-split-group (if spam-split-symbolic-return
2090                                  'spam
2091                                spam-split-group)))
2092           (when header                  ; return nil when no header
2093             (when (string-match spam-spamassassin-positive-spam-flag-header
2094                                 header)
2095               spam-split-group)))))
2096
2097 (defun spam-check-spamassassin (&optional score)
2098   "Check the SpamAssassin backend for the classification of this message."
2099   (let ((article-buffer-name (buffer-name)))
2100     (with-temp-buffer
2101       (let ((temp-buffer-name (buffer-name)))
2102         (save-excursion
2103           (set-buffer article-buffer-name)
2104           (apply 'call-process-region
2105                  (point-min) (point-max) spam-spamassassin-path
2106                  nil temp-buffer-name nil spam-spamassassin-arguments))
2107         ;; check the return now (we're back in the temp buffer)
2108         (goto-char (point-min))
2109         (spam-check-spamassassin-headers score)))))
2110
2111 ;; return something sensible if the score can't be determined
2112 (defun spam-spamassassin-score ()
2113   "Get the SpamAssassin score"
2114   (interactive)
2115   (save-window-excursion
2116     (gnus-summary-show-article t)
2117     (set-buffer gnus-article-buffer)
2118     (let ((score (or (spam-check-spamassassin-headers t)
2119                      (spam-check-spamassassin t))))
2120       (gnus-summary-show-article)
2121       (message "SpamAssassin score %s" score)
2122       (or score "0"))))
2123
2124 (defun spam-spamassassin-register-with-sa-learn (articles spam
2125                                                  &optional unregister)
2126   "Register articles with spamassassin's sa-learn as spam or non-spam."
2127   (if articles
2128       (let ((action (if unregister spam-sa-learn-unregister-switch
2129                       (if spam spam-sa-learn-spam-switch
2130                         spam-sa-learn-ham-switch)))
2131             (summary-buffer-name (buffer-name)))
2132         (with-temp-buffer
2133           ;; group the articles into mbox format
2134           (dolist (article articles)
2135             (let (article-string)
2136               (save-excursion
2137                 (set-buffer summary-buffer-name)
2138                 (setq article-string (spam-get-article-as-string article)))
2139               (when (stringp article-string)
2140                 (insert "From \n") ; mbox separator (sa-learn only checks the
2141                                    ; first five chars, so we can get away with
2142                                    ; a bogus line))
2143                 (insert article-string)
2144                 (insert "\n"))))
2145           ;; call sa-learn on all messages at the same time
2146           (apply 'call-process-region
2147                  (point-min) (point-max)
2148                  spam-sa-learn-path
2149                  nil nil nil "--mbox"
2150                  (if spam-sa-learn-rebuild
2151                      (list action)
2152                    `("--no-rebuild" ,action)))))))
2153
2154 (defun spam-spamassassin-register-spam-routine (articles &optional unregister)
2155   (spam-spamassassin-register-with-sa-learn articles t unregister))
2156
2157 (defun spam-spamassassin-register-ham-routine (articles &optional unregister)
2158   (spam-spamassassin-register-with-sa-learn articles nil unregister))
2159
2160 (defun spam-spamassassin-unregister-spam-routine (articles)
2161   (spam-spamassassin-register-with-sa-learn articles t t))
2162
2163 (defun spam-spamassassin-unregister-ham-routine (articles)
2164   (spam-spamassassin-register-with-sa-learn articles nil t))
2165 \f
2166 ;;;; Hooks
2167
2168 ;;;###autoload
2169 (defun spam-initialize ()
2170   "Install the spam.el hooks and do other initialization"
2171   (interactive)
2172   (setq spam-install-hooks t)
2173   ;; TODO: How do we redo this every time spam-face is customized?
2174   (push '((eq mark gnus-spam-mark) . spam-face)
2175         gnus-summary-highlight)
2176   ;; Add hooks for loading and saving the spam stats
2177   (add-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2178   (add-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2179   (add-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2180   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2181   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2182   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2183   (add-hook 'gnus-summary-prepared-hook 'spam-find-spam))
2184
2185 (defun spam-unload-hook ()
2186   "Uninstall the spam.el hooks"
2187   (interactive)
2188   (remove-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2189   (remove-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2190   (remove-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2191   (remove-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2192   (remove-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2193   (remove-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2194   (remove-hook 'gnus-summary-prepare-hook 'spam-find-spam))
2195
2196 (when spam-install-hooks
2197   (spam-initialize))
2198
2199 (provide 'spam)
2200
2201 ;;; spam.el ends here