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 "~/News/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-blacklist nil
 166   "Whether the blacklist should be used by `spam-split'."
 167   :type 'boolean
 168   :group 'spam)
 169
 170 (defcustom spam-blacklist-ignored-regexes nil
 171   "Regular expressions that the blacklist should ignore."
 172   :type '(repeat (regexp :tag "Regular expression to ignore when blacklisting"))
 173   :group 'spam)
 174
 175 (defcustom spam-use-whitelist nil
 176   "Whether the whitelist should be used by `spam-split'."
 177   :type 'boolean
 178   :group 'spam)
 179
 180 (defcustom spam-use-whitelist-exclusive nil
 181   "Whether whitelist-exclusive should be used by `spam-split'.
 182 Exclusive whitelisting means that all messages from senders not in the whitelist
 183 are considered spam."
 184   :type 'boolean
 185   :group 'spam)
 186
 187 (defcustom spam-use-blackholes nil
 188   "Whether blackholes should be used by `spam-split'."
 189   :type 'boolean
 190   :group 'spam)
 191
 192 (defcustom spam-use-hashcash nil
 193   "Whether hashcash payments should be detected by `spam-split'."
 194   :type 'boolean
 195   :group 'spam)
 196
 197 (defcustom spam-use-regex-headers nil
 198   "Whether a header regular expression match should be used by `spam-split'.
 199 Also see the variables `spam-regex-headers-spam' and `spam-regex-headers-ham'."
 200   :type 'boolean
 201   :group 'spam)
 202
 203 (defcustom spam-use-regex-body nil
 204   "Whether a body regular expression match should be used by `spam-split'.
 205 Also see the variables `spam-regex-body-spam' and `spam-regex-body-ham'."
 206   :type 'boolean
 207   :group 'spam)
 208
 209 (defcustom spam-use-bogofilter-headers nil
 210   "Whether bogofilter headers should be used by `spam-split'.
 211 Enable this if you pre-process messages with Bogofilter BEFORE Gnus sees them."
 212   :type 'boolean
 213   :group 'spam)
 214
 215 (defcustom spam-use-bogofilter nil
 216   "Whether bogofilter should be invoked by `spam-split'.
 217 Enable this if you want Gnus to invoke Bogofilter on new messages."
 218   :type 'boolean
 219   :group 'spam)
 220
 221 (defcustom spam-use-BBDB nil
 222   "Whether BBDB should be used by `spam-split'."
 223   :type 'boolean
 224   :group 'spam)
 225
 226 (defcustom spam-use-BBDB-exclusive nil
 227   "Whether BBDB-exclusive should be used by `spam-split'.
 228 Exclusive BBDB means that all messages from senders not in the BBDB are
 229 considered spam."
 230   :type 'boolean
 231   :group 'spam)
 232
 233 (defcustom spam-use-ifile nil
 234   "Whether ifile should be used by `spam-split'."
 235   :type 'boolean
 236   :group 'spam)
 237
 238 (defcustom spam-use-stat nil
 239   "Whether `spam-stat' should be used by `spam-split'."
 240   :type 'boolean
 241   :group 'spam)
 242
 243 (defcustom spam-use-spamoracle nil
 244   "Whether spamoracle should be used by `spam-split'."
 245   :type 'boolean
 246   :group 'spam)
 247
 248 (defcustom spam-use-spamassassin nil
 249   "Whether spamassassin should be invoked by `spam-split'.
 250 Enable this if you want Gnus to invoke SpamAssassin on new messages."
 251   :type 'boolean
 252   :group 'spam)
 253
 254 (defcustom spam-use-spamassassin-headers nil
 255   "Whether spamassassin headers should be checked by `spam-split'.
 256 Enable this if you pre-process messages with SpamAssassin BEFORE Gnus sees
 257 them."
 258   :type 'boolean
 259   :group 'spam)
 260
 261 (defcustom spam-install-hooks (or
 262                                spam-use-dig
 263                                spam-use-blacklist
 264                                spam-use-whitelist
 265                                spam-use-whitelist-exclusive
 266                                spam-use-blackholes
 267                                spam-use-hashcash
 268                                spam-use-regex-headers
 269                                spam-use-regex-body
 270                                spam-use-bogofilter
 271                                spam-use-bogofilter-headers
 272                                spam-use-spamassassin
 273                                spam-use-spamassassin-headers
 274                                spam-use-BBDB
 275                                spam-use-BBDB-exclusive
 276                                spam-use-ifile
 277                                spam-use-stat
 278                                spam-use-spamoracle)
 279   "Whether the spam hooks should be installed.
 280 Default to t if one of the spam-use-* variables is set."
 281   :group 'spam
 282   :type 'boolean)
 283
 284 (defcustom spam-split-group "spam"
 285   "Group name where incoming spam should be put by `spam-split'."
 286   :type 'string
 287   :group 'spam)
 288
 289 ;;; TODO: deprecate this variable, it's confusing since it's a list of strings,
 290 ;;; not regular expressions
 291 (defcustom spam-junk-mailgroups (cons
 292                                  spam-split-group
 293                                  '("mail.junk" "poste.pourriel"))
 294   "Mailgroups with spam contents.
 295 All unmarked article in such group receive the spam mark on group entry."
 296   :type '(repeat (string :tag "Group"))
 297   :group 'spam)
 298
 299 (defcustom spam-blackhole-servers '("bl.spamcop.net" "relays.ordb.org"
 300                                     "dev.null.dk" "relays.visi.com")
 301   "List of blackhole servers."
 302   :type '(repeat (string :tag "Server"))
 303   :group 'spam)
 304
 305 (defcustom spam-blackhole-good-server-regex nil
 306   "String matching IP addresses that should not be checked in the blackholes."
 307   :type '(radio (const nil)
 308                 (regexp :format "%t: %v\n" :size 0))
 309   :group 'spam)
 310
 311 (defcustom spam-face 'gnus-splash-face
 312   "Face for spam-marked articles."
 313   :type 'face
 314   :group 'spam)
 315
 316 (defcustom spam-regex-headers-spam '("^X-Spam-Flag: YES")
 317   "Regular expression for positive header spam matches."
 318   :type '(repeat (regexp :tag "Regular expression to match spam header"))
 319   :group 'spam)
 320
 321 (defcustom spam-regex-headers-ham '("^X-Spam-Flag: NO")
 322   "Regular expression for positive header ham matches."
 323   :type '(repeat (regexp :tag "Regular expression to match ham header"))
 324   :group 'spam)
 325
 326 (defcustom spam-regex-body-spam '()
 327   "Regular expression for positive body spam matches."
 328   :type '(repeat (regexp :tag "Regular expression to match spam body"))
 329   :group 'spam)
 330
 331 (defcustom spam-regex-body-ham '()
 332   "Regular expression for positive body ham matches."
 333   :type '(repeat (regexp :tag "Regular expression to match ham body"))
 334   :group 'spam)
 335
 336 (defgroup spam-ifile nil
 337   "Spam ifile configuration."
 338   :group 'spam)
 339
 340 (defcustom spam-ifile-path (executable-find "ifile")
 341   "File path of the ifile executable program."
 342   :type '(choice (file :tag "Location of ifile")
 343                  (const :tag "ifile is not installed"))
 344   :group 'spam-ifile)
 345
 346 (defcustom spam-ifile-database-path nil
 347   "File path of the ifile database."
 348   :type '(choice (file :tag "Location of the ifile database")
 349                  (const :tag "Use the default"))
 350   :group 'spam-ifile)
 351
 352 (defcustom spam-ifile-spam-category "spam"
 353   "Name of the spam ifile category."
 354   :type 'string
 355   :group 'spam-ifile)
 356
 357 (defcustom spam-ifile-ham-category nil
 358   "Name of the ham ifile category.
 359 If nil, the current group name will be used."
 360   :type '(choice (string :tag "Use a fixed category")
 361                  (const :tag "Use the current group name"))
 362   :group 'spam-ifile)
 363
 364 (defcustom spam-ifile-all-categories nil
 365   "Whether the ifile check will return all categories, or just spam.
 366 Set this to t if you want to use the `spam-split' invocation of ifile as
 367 your main source of newsgroup names."
 368   :type 'boolean
 369   :group 'spam-ifile)
 370
 371 (defgroup spam-bogofilter nil
 372   "Spam bogofilter configuration."
 373   :group 'spam)
 374
 375 (defcustom spam-bogofilter-path (executable-find "bogofilter")
 376   "File path of the Bogofilter executable program."
 377   :type '(choice (file :tag "Location of bogofilter")
 378                  (const :tag "Bogofilter is not installed"))
 379   :group 'spam-bogofilter)
 380
 381 (defcustom spam-bogofilter-header "X-Bogosity"
 382   "The header that Bogofilter inserts in messages."
 383   :type 'string
 384   :group 'spam-bogofilter)
 385
 386 (defcustom spam-bogofilter-spam-switch "-s"
 387   "The switch that Bogofilter uses to register spam messages."
 388   :type 'string
 389   :group 'spam-bogofilter)
 390
 391 (defcustom spam-bogofilter-ham-switch "-n"
 392   "The switch that Bogofilter uses to register ham messages."
 393   :type 'string
 394   :group 'spam-bogofilter)
 395
 396 (defcustom spam-bogofilter-spam-strong-switch "-S"
 397   "The switch that Bogofilter uses to unregister ham messages."
 398   :type 'string
 399   :group 'spam-bogofilter)
 400
 401 (defcustom spam-bogofilter-ham-strong-switch "-N"
 402   "The switch that Bogofilter uses to unregister spam messages."
 403   :type 'string
 404   :group 'spam-bogofilter)
 405
 406 (defcustom spam-bogofilter-bogosity-positive-spam-header "^\\(Yes\\|Spam\\)"
 407   "The regex on `spam-bogofilter-header' for positive spam identification."
 408   :type 'regexp
 409   :group 'spam-bogofilter)
 410
 411 (defcustom spam-bogofilter-database-directory nil
 412   "Directory path of the Bogofilter databases."
 413   :type '(choice (directory
 414                   :tag "Location of the Bogofilter database directory")
 415                  (const :tag "Use the default"))
 416   :group 'spam-bogofilter)
 417
 418 (defgroup spam-spamoracle nil
 419   "Spam spamoracle configuration."
 420   :group 'spam)
 421
 422 (defcustom spam-spamoracle-database nil
 423   "Location of spamoracle database file.
 424 When nil, use the default spamoracle database."
 425   :type '(choice (directory :tag "Location of spamoracle database file.")
 426                  (const :tag "Use the default"))
 427   :group 'spam-spamoracle)
 428
 429 (defcustom spam-spamoracle-binary (executable-find "spamoracle")
 430   "Location of the spamoracle binary."
 431   :type '(choice (directory :tag "Location of the spamoracle binary")
 432                  (const :tag "Use the default"))
 433   :group 'spam-spamoracle)
 434
 435 (defgroup spam-spamassassin nil
 436   "Spam SpamAssassin configuration."
 437   :group 'spam)
 438
 439 (defcustom spam-spamassassin-path (executable-find "spamassassin")
 440   "File path of the spamassassin executable program.
 441 Hint: set this to \"spamc\" if you have spamd running.  See the spamc and
 442 spamd man pages for more information on these programs."
 443   :type '(choice (file :tag "Location of spamc")
 444                  (const :tag "spamassassin is not installed"))
 445   :group 'spam-spamassassin)
 446
 447 (defcustom spam-spamassassin-arguments ()
 448   "Arguments to pass to the spamassassin executable.
 449 This must be a list.  For example, `(\"-C\" \"configfile\")'."
 450   :type '(restricted-sexp :match-alternatives (listp))
 451   :group 'spam-spamassassin)
 452
 453 (defcustom spam-spamassassin-spam-flag-header "X-Spam-Flag"
 454   "The header inserted by SpamAssassin to flag spam."
 455   :type 'string
 456   :group 'spam-spamassassin)
 457
 458 (defcustom spam-spamassassin-positive-spam-flag-header "YES"
 459   "The regex on `spam-spamassassin-spam-flag-header' for positive spam
 460 identification"
 461   :type 'string
 462   :group 'spam-spamassassin)
 463
 464 (defcustom spam-spamassassin-spam-status-header "X-Spam-Status"
 465   "The header inserted by SpamAssassin, giving extended scoring information"
 466   :type 'string
 467   :group 'spam-spamassassin)
 468
 469 (defcustom spam-sa-learn-path (executable-find "sa-learn")
 470   "File path of the sa-learn executable program."
 471   :type '(choice (file :tag "Location of spamassassin")
 472                  (const :tag "spamassassin is not installed"))
 473   :group 'spam-spamassassin)
 474
 475 (defcustom spam-sa-learn-rebuild t
 476   "Whether sa-learn should rebuild the database every time it is called
 477 Enable this if you want sa-learn to rebuild the database automatically.  Doing
 478 this will slightly increase the running time of the spam registration process.
 479 If you choose not to do this, you will have to run \"sa-learn --rebuild\" in
 480 order for SpamAssassin to recognize the new registered spam."
 481   :type 'boolean
 482   :group 'spam-spamassassin)
 483
 484 (defcustom spam-sa-learn-spam-switch "--spam"
 485   "The switch that sa-learn uses to register spam messages"
 486   :type 'string
 487   :group 'spam-spamassassin)
 488
 489 (defcustom spam-sa-learn-ham-switch "--ham"
 490   "The switch that sa-learn uses to register ham messages"
 491   :type 'string
 492   :group 'spam-spamassassin)
 493
 494 (defcustom spam-sa-learn-unregister-switch "--forget"
 495   "The switch that sa-learn uses to unregister messages messages"
 496   :type 'string
 497   :group 'spam-spamassassin)
 498
 499 ;;; Key bindings for spam control.
 500
 501 (gnus-define-keys gnus-summary-mode-map
 502   "St" spam-bogofilter-score
 503   "Sx" gnus-summary-mark-as-spam
 504   "Mst" spam-bogofilter-score
 505   "Msx" gnus-summary-mark-as-spam
 506   "\M-d" gnus-summary-mark-as-spam)
 507
 508 (defvar spam-cache-lookups t
 509   "Whether spam.el will try to cache lookups using `spam-caches'.")
 510
 511 (defvar spam-caches (make-hash-table
 512                      :size 10
 513                      :test 'equal)
 514   "Cache of spam detection entries.")
 515
 516 (defvar spam-old-ham-articles nil
 517   "List of old ham articles, generated when a group is entered.")
 518
 519 (defvar spam-old-spam-articles nil
 520   "List of old spam articles, generated when a group is entered.")
 521
 522 (defvar spam-split-disabled nil
 523   "If non-nil, `spam-split' is disabled, and always returns nil.")
 524
 525 (defvar spam-split-last-successful-check nil
 526   "Internal variable.
 527 `spam-split' will set this to nil or a spam-use-XYZ check if it
 528 finds ham or spam.")
 529
 530 ;; convenience functions
 531 (defun spam-clear-cache (symbol)
 532   "Clear the spam-caches entry for a check."
 533   (remhash symbol spam-caches))
 534
 535 (defun spam-xor (a b)
 536   "Logical A xor B."
 537   (and (or a b) (not (and a b))))
 538
 539 (defun spam-group-ham-mark-p (group mark &optional spam)
 540   "Checks if MARK is considered a ham mark in GROUP."
 541   (when (stringp group)
 542     (let* ((marks (spam-group-ham-marks group spam))
 543            (marks (if (symbolp mark)
 544                       marks
 545                     (mapcar 'symbol-value marks))))
 546       (memq mark marks))))
 547
 548 (defun spam-group-spam-mark-p (group mark)
 549   "Checks if MARK is considered a spam mark in GROUP."
 550   (spam-group-ham-mark-p group mark t))
 551
 552 (defun spam-group-ham-marks (group &optional spam)
 553   "In GROUP, get all the ham marks."
 554   (when (stringp group)
 555     (let* ((marks (if spam
 556                       (gnus-parameter-spam-marks group)
 557                     (gnus-parameter-ham-marks group)))
 558            (marks (car marks))
 559            (marks (if (listp (car marks)) (car marks) marks)))
 560       marks)))
 561
 562 (defun spam-group-spam-marks (group)
 563   "In GROUP, get all the spam marks."
 564   (spam-group-ham-marks group t))
 565
 566 (defun spam-group-spam-contents-p (group)
 567   "Is GROUP a spam group?"
 568   (if (stringp group)
 569       (or (member group spam-junk-mailgroups)
 570           (memq 'gnus-group-spam-classification-spam
 571                 (gnus-parameter-spam-contents group)))
 572     nil))
 573
 574 (defun spam-group-ham-contents-p (group)
 575   "Is GROUP a ham group?"
 576   (if (stringp group)
 577       (memq 'gnus-group-spam-classification-ham
 578             (gnus-parameter-spam-contents group))
 579     nil))
 580
 581 (defvar spam-list-of-processors
 582   '((gnus-group-spam-exit-processor-report-gmane spam spam-use-gmane)
 583     (gnus-group-spam-exit-processor-bogofilter   spam spam-use-bogofilter)
 584     (gnus-group-spam-exit-processor-blacklist    spam spam-use-blacklist)
 585     (gnus-group-spam-exit-processor-ifile        spam spam-use-ifile)
 586     (gnus-group-spam-exit-processor-stat         spam spam-use-stat)
 587     (gnus-group-spam-exit-processor-spamoracle   spam spam-use-spamoracle)
 588     (gnus-group-spam-exit-processor-spamassassin spam spam-use-spam-spamassassin)
 589     (gnus-group-ham-exit-processor-ifile         ham spam-use-ifile)
 590     (gnus-group-ham-exit-processor-bogofilter    ham spam-use-bogofilter)
 591     (gnus-group-ham-exit-processor-stat          ham spam-use-stat)
 592     (gnus-group-ham-exit-processor-whitelist     ham spam-use-whitelist)
 593     (gnus-group-ham-exit-processor-BBDB          ham spam-use-BBDB)
 594     (gnus-group-ham-exit-processor-copy          ham spam-use-ham-copy)
 595     (gnus-group-ham-exit-processor-spamassassin  ham spam-use-ham-spamassassin)
 596     (gnus-group-ham-exit-processor-spamoracle    ham spam-use-spamoracle))
 597   "The `spam-list-of-processors' list.
 598 This list contains pairs associating a ham/spam exit processor
 599 variable with a classification and a spam-use-* variable.")
 600
 601 (defun spam-group-processor-p (group processor)
 602   (if (and (stringp group)
 603            (symbolp processor))
 604       (or (member processor (nth 0 (gnus-parameter-spam-process group)))
 605           (spam-group-processor-multiple-p
 606            group
 607            (cdr-safe (assoc processor spam-list-of-processors))))
 608     nil))
 609
 610 (defun spam-group-processor-multiple-p (group processor-info)
 611   (let* ((classification (nth 0 processor-info))
 612          (check (nth 1 processor-info))
 613          (parameters (nth 0 (gnus-parameter-spam-process group)))
 614          found)
 615     (dolist (parameter parameters)
 616       (when (and (null found)
 617                  (listp parameter)
 618                  (eq classification (nth 0 parameter))
 619                  (eq check (nth 1 parameter)))
 620         (setq found t)))
 621     found))
 622
 623 (defun spam-group-spam-processor-report-gmane-p (group)
 624   (spam-group-processor-p group 'gnus-group-spam-exit-processor-report-gmane))
 625
 626 (defun spam-group-spam-processor-bogofilter-p (group)
 627   (spam-group-processor-p group 'gnus-group-spam-exit-processor-bogofilter))
 628
 629 (defun spam-group-spam-processor-blacklist-p (group)
 630   (spam-group-processor-p group 'gnus-group-spam-exit-processor-blacklist))
 631
 632 (defun spam-group-spam-processor-ifile-p (group)
 633   (spam-group-processor-p group 'gnus-group-spam-exit-processor-ifile))
 634
 635 (defun spam-group-ham-processor-ifile-p (group)
 636   (spam-group-processor-p group 'gnus-group-ham-exit-processor-ifile))
 637
 638 (defun spam-group-spam-processor-spamoracle-p (group)
 639   (spam-group-processor-p group 'gnus-group-spam-exit-processor-spamoracle))
 640
 641 (defun spam-group-ham-processor-bogofilter-p (group)
 642   (spam-group-processor-p group 'gnus-group-ham-exit-processor-bogofilter))
 643
 644 (defun spam-group-spam-processor-stat-p (group)
 645   (spam-group-processor-p group 'gnus-group-spam-exit-processor-stat))
 646
 647 (defun spam-group-ham-processor-stat-p (group)
 648   (spam-group-processor-p group 'gnus-group-ham-exit-processor-stat))
 649
 650 (defun spam-group-ham-processor-whitelist-p (group)
 651   (spam-group-processor-p group 'gnus-group-ham-exit-processor-whitelist))
 652
 653 (defun spam-group-ham-processor-BBDB-p (group)
 654   (spam-group-processor-p group 'gnus-group-ham-exit-processor-BBDB))
 655
 656 (defun spam-group-ham-processor-copy-p (group)
 657   (spam-group-processor-p group 'gnus-group-ham-exit-processor-copy))
 658
 659 (defun spam-group-ham-processor-spamoracle-p (group)
 660   (spam-group-processor-p group 'gnus-group-ham-exit-processor-spamoracle))
 661
 662 (defun spam-report-articles-gmane (n)
 663   "Report the current message as spam.
 664 Respects the process/prefix convention."
 665   (interactive "P")
 666   (dolist (article (gnus-summary-work-articles n))
 667     (gnus-summary-remove-process-mark article)
 668     (spam-report-gmane article)))
 669
 670 ;;; Summary entry and exit processing.
 671
 672 (defun spam-summary-prepare ()
 673   (setq spam-old-ham-articles
 674         (spam-list-articles gnus-newsgroup-articles 'ham))
 675   (setq spam-old-spam-articles
 676         (spam-list-articles gnus-newsgroup-articles 'spam))
 677   (spam-mark-junk-as-spam-routine))
 678
 679 ;; The spam processors are invoked for any group, spam or ham or neither
 680 (defun spam-summary-prepare-exit ()
 681   (unless gnus-group-is-exiting-without-update-p
 682     (gnus-message 6 "Exiting summary buffer and applying spam rules")
 683
 684     ;; first of all, unregister any articles that are no longer ham or spam
 685     ;; we have to iterate over the processors, or else we'll be too slow
 686     (dolist (classification '(spam ham))
 687       (let* ((old-articles (if (eq classification 'spam)
 688                                spam-old-spam-articles
 689                              spam-old-ham-articles))
 690              (new-articles (spam-list-articles
 691                             gnus-newsgroup-articles
 692                             classification))
 693              (changed-articles (gnus-set-difference old-articles new-articles)))
 694         ;; now that we have the changed articles, we go through the processors
 695         (dolist (processor-param spam-list-of-processors)
 696           (let ((processor (nth 0 processor-param))
 697                 (processor-classification (nth 1 processor-param))
 698                 (check (nth 2 processor-param))
 699                 unregister-list)
 700             (dolist (article changed-articles)
 701               (let ((id (spam-fetch-field-message-id-fast article)))
 702                 (when (spam-log-unregistration-needed-p
 703                        id 'process classification check)
 704                   (push article unregister-list))))
 705             ;; call spam-register-routine with specific articles to unregister,
 706             ;; when there are articles to unregister and the check is enabled
 707             (when (and unregister-list (symbol-value check))
 708               (spam-register-routine classification check t unregister-list))))))
 709
 710     ;; find all the spam processors applicable to this group
 711     (dolist (processor-param spam-list-of-processors)
 712       (let ((processor (nth 0 processor-param))
 713             (classification (nth 1 processor-param))
 714             (check (nth 2 processor-param)))
 715         (when (and (eq 'spam classification)
 716                    (spam-group-processor-p gnus-newsgroup-name processor))
 717           (spam-register-routine classification check))))
 718
 719     (if spam-move-spam-nonspam-groups-only
 720         (when (not (spam-group-spam-contents-p gnus-newsgroup-name))
 721           (spam-mark-spam-as-expired-and-move-routine
 722            (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
 723       (gnus-message 5 "Marking spam as expired and moving it to %s"
 724                     gnus-newsgroup-name)
 725       (spam-mark-spam-as-expired-and-move-routine
 726        (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
 727
 728     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
 729     ;; expire spam, in case the above did not expire them
 730     (gnus-message 5 "Marking spam as expired without moving it")
 731     (spam-mark-spam-as-expired-and-move-routine nil)
 732
 733     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
 734               (and (spam-group-spam-contents-p gnus-newsgroup-name)
 735                    spam-process-ham-in-spam-groups)
 736               spam-process-ham-in-nonham-groups)
 737       ;; find all the ham processors applicable to this group
 738       (dolist (processor-param spam-list-of-processors)
 739         (let ((processor (nth 0 processor-param))
 740               (classification (nth 1 processor-param))
 741               (check (nth 2 processor-param)))
 742           (when (and (eq 'ham classification)
 743                      (spam-group-processor-p gnus-newsgroup-name processor))
 744             (spam-register-routine classification check)))))
 745
 746     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
 747       (gnus-message 5 "Copying ham")
 748       (spam-ham-copy-routine
 749        (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
 750
 751     ;; now move all ham articles out of spam groups
 752     (when (spam-group-spam-contents-p gnus-newsgroup-name)
 753       (gnus-message 5 "Moving ham messages from spam group")
 754       (spam-ham-move-routine
 755        (gnus-parameter-ham-process-destination gnus-newsgroup-name))))
 756
 757   (setq spam-old-ham-articles nil)
 758   (setq spam-old-spam-articles nil))
 759
 760 (defun spam-mark-junk-as-spam-routine ()
 761   ;; check the global list of group names spam-junk-mailgroups and the
 762   ;; group parameters
 763   (when (spam-group-spam-contents-p gnus-newsgroup-name)
 764     (gnus-message 5 "Marking %s articles as spam"
 765                   (if spam-mark-only-unseen-as-spam
 766                       "unseen"
 767                     "unread"))
 768     (let ((articles (if spam-mark-only-unseen-as-spam
 769                         gnus-newsgroup-unseen
 770                       gnus-newsgroup-unreads)))
 771       (dolist (article articles)
 772         (gnus-summary-mark-article article gnus-spam-mark)))))
 773
 774 (defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
 775   (if (and (car-safe groups) (listp (car-safe groups)))
 776       (apply 'spam-mark-spam-as-expired-and-move-routine (car groups))
 777     (gnus-summary-kill-process-mark)
 778     (let ((articles gnus-newsgroup-articles)
 779           (backend-supports-deletions
 780            (gnus-check-backend-function
 781             'request-move-article gnus-newsgroup-name))
 782           article tomove deletep)
 783       (dolist (article articles)
 784         (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
 785           (gnus-summary-mark-article article gnus-expirable-mark)
 786           (push article tomove)))
 787
 788       ;; now do the actual copies
 789       (dolist (group groups)
 790         (when (and tomove
 791                    (stringp group))
 792           (dolist (article tomove)
 793             (gnus-summary-set-process-mark article))
 794           (when tomove
 795             (if (or (not backend-supports-deletions)
 796                     (> (length groups) 1))
 797                 (progn
 798                   (gnus-summary-copy-article nil group)
 799                   (setq deletep t))
 800               (gnus-summary-move-article nil group)))))
 801
 802       ;; now delete the articles, if there was a copy done, and the
 803       ;; backend allows it
 804       (when (and deletep backend-supports-deletions)
 805         (dolist (article tomove)
 806           (gnus-summary-set-process-mark article))
 807         (when tomove
 808           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
 809             (gnus-summary-delete-article nil))))
 810
 811       (gnus-summary-yank-process-mark))))
 812
 813 (defun spam-ham-copy-or-move-routine (copy groups)
 814   (gnus-summary-kill-process-mark)
 815   (let ((todo (spam-list-articles gnus-newsgroup-articles 'ham))
 816         (backend-supports-deletions
 817          (gnus-check-backend-function
 818           'request-move-article gnus-newsgroup-name))
 819         (respool-method (gnus-find-method-for-group gnus-newsgroup-name))
 820         article mark todo deletep respool)
 821
 822     (when (member 'respool groups)
 823       (setq respool t)                  ; boolean for later
 824       (setq groups '("fake"))) ; when respooling, groups are dynamic so fake it
 825
 826     ;; now do the actual move
 827     (dolist (group groups)
 828       (when (and todo (stringp group))
 829         (dolist (article todo)
 830           (when spam-mark-ham-unread-before-move-from-spam-group
 831             (gnus-summary-mark-article article gnus-unread-mark))
 832           (gnus-summary-set-process-mark article))
 833
 834         (if respool                        ; respooling is with a "fake" group
 835             (let ((spam-split-disabled
 836                    (or spam-split-disabled
 837                        spam-disable-spam-split-during-ham-respool)))
 838               (gnus-summary-respool-article nil respool-method))
 839           (if (or (not backend-supports-deletions) ; else, we are not respooling
 840                   (> (length groups) 1))
 841               (progn                ; if copying, copy and set deletep
 842                 (gnus-summary-copy-article nil group)
 843                 (setq deletep t))
 844             (gnus-summary-move-article nil group))))) ; else move articles
 845
 846     ;; now delete the articles, unless a) copy is t, and there was a copy done
 847     ;;                                 b) a move was done to a single group
 848     ;;                                 c) backend-supports-deletions is nil
 849     (unless copy
 850       (when (and deletep backend-supports-deletions)
 851         (dolist (article todo)
 852           (gnus-summary-set-process-mark article))
 853         (when todo
 854           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
 855             (gnus-summary-delete-article nil))))))
 856
 857   (gnus-summary-yank-process-mark))
 858
 859 (defun spam-ham-copy-routine (&rest groups)
 860   (if (and (car-safe groups) (listp (car-safe groups)))
 861       (apply 'spam-ham-copy-routine (car groups))
 862     (spam-ham-copy-or-move-routine t groups)))
 863
 864 (defun spam-ham-move-routine (&rest groups)
 865   (if (and (car-safe groups) (listp (car-safe groups)))
 866       (apply 'spam-ham-move-routine (car groups))
 867     (spam-ham-copy-or-move-routine nil groups)))
 868
 869 (defun spam-get-article-as-string (article)
 870   (when (numberp article)
 871     (with-temp-buffer
 872       (gnus-request-article-this-buffer
 873        article
 874        gnus-newsgroup-name)
 875       (buffer-string))))
 876
 877 ;; disabled for now
 878 ;; (defun spam-get-article-as-filename (article)
 879 ;;   (let ((article-filename))
 880 ;;     (when (numberp article)
 881 ;;       (nnml-possibly-change-directory
 882 ;;        (gnus-group-real-name gnus-newsgroup-name))
 883 ;;       (setq article-filename (expand-file-name
 884 ;;                              (int-to-string article) nnml-current-directory)))
 885 ;;     (if (file-exists-p article-filename)
 886 ;;      article-filename
 887 ;;       nil)))
 888
 889 (defun spam-fetch-field-fast (article field &optional prepared-data-header)
 890   "Fetch a field quickly, using the internal gnus-data-list function"
 891   (when (numberp article)
 892     (let* ((data-header (or prepared-data-header
 893                             (spam-fetch-article-header article))))
 894       (if (arrayp data-header)
 895         (cond
 896          ((equal field 'from)
 897           (mail-header-from data-header))
 898          ((equal field 'message-id)
 899           (mail-header-message-id data-header))
 900          ((equal field 'subject)
 901           (mail-header-subject data-header))
 902          ((equal field 'references)
 903           (mail-header-references data-header))
 904          ((equal field 'date)
 905           (mail-header-date data-header))
 906          ((equal field 'xref)
 907           (mail-header-xref data-header))
 908          ((equal field 'extra)
 909           (mail-header-extra data-header))
 910          (t
 911           nil))
 912         (gnus-error 5 "Article %d has a nil data header" article)))))
 913
 914 (defun spam-fetch-field-from-fast (article &optional prepared-data-header)
 915   (spam-fetch-field-fast article 'from prepared-data-header))
 916
 917 (defun spam-fetch-field-subject-fast (article &optional prepared-data-header)
 918   (spam-fetch-field-fast article 'subject prepared-data-header))
 919
 920 (defun spam-fetch-field-message-id-fast (article &optional prepared-data-header)
 921   (spam-fetch-field-fast article 'message-id prepared-data-header))
 922
 923 (defun spam-generate-fake-headers (article)
 924   (let ((dh (spam-fetch-article-header article)))
 925     (if dh
 926         (concat
 927          (format
 928           (concat "From: %s\nSubject: %s\nMessage-ID: %s\n"
 929                   "Date: %s\nReferences: %s\nXref: %s\n")
 930           (spam-fetch-field-fast article 'from dh)
 931           (spam-fetch-field-fast article 'subject dh)
 932           (spam-fetch-field-fast article 'message-id dh)
 933           (spam-fetch-field-fast article 'date dh)
 934           (spam-fetch-field-fast article 'references dh)
 935           (spam-fetch-field-fast article 'xref dh))
 936          (when (spam-fetch-field-fast article 'extra dh)
 937            (format "%s\n" (spam-fetch-field-fast article 'extra dh))))
 938       (gnus-error
 939        5
 940        "spam-generate-fake-headers: article %d didn't have a valid header"
 941        article))))
 942
 943 (defun spam-fetch-article-header (article)
 944   (save-excursion
 945     (set-buffer gnus-summary-buffer)
 946     (nth 3 (assq article gnus-newsgroup-data))))
 947
 948 \f
 949 ;;;; Spam determination.
 950
 951 (defvar spam-list-of-checks
 952   '((spam-use-blacklist                 .       spam-check-blacklist)
 953     (spam-use-regex-headers             .       spam-check-regex-headers)
 954     (spam-use-regex-body                .       spam-check-regex-body)
 955     (spam-use-whitelist                 .       spam-check-whitelist)
 956     (spam-use-BBDB                      .       spam-check-BBDB)
 957     (spam-use-ifile                     .       spam-check-ifile)
 958     (spam-use-spamoracle                .       spam-check-spamoracle)
 959     (spam-use-stat                      .       spam-check-stat)
 960     (spam-use-blackholes                .       spam-check-blackholes)
 961     (spam-use-hashcash                  .       spam-check-hashcash)
 962     (spam-use-spamassassin-headers      .       spam-check-spamassassin-headers)
 963     (spam-use-spamassassin              .       spam-check-spamassassin)
 964     (spam-use-bogofilter-headers        .       spam-check-bogofilter-headers)
 965     (spam-use-bogofilter                .       spam-check-bogofilter))
 966   "The spam-list-of-checks list contains pairs associating a
 967 parameter variable with a spam checking function.  If the
 968 parameter variable is true, then the checking function is called,
 969 and its value decides what happens.  Each individual check may
 970 return nil, t, or a mailgroup name.  The value nil means that the
 971 check does not yield a decision, and so, that further checks are
 972 needed.  The value t means that the message is definitely not
 973 spam, and that further spam checks should be inhibited.
 974 Otherwise, a mailgroup name or the symbol 'spam (depending on
 975 spam-split-symbolic-return) is returned where the mail should go,
 976 and further checks are also inhibited.  The usual mailgroup name
 977 is the value of `spam-split-group', meaning that the message is
 978 definitely a spam.")
 979
 980 (defvar spam-list-of-statistical-checks
 981   '(spam-use-ifile
 982     spam-use-regex-body
 983     spam-use-stat
 984     spam-use-bogofilter
 985     spam-use-blackholes
 986     spam-use-spamassassin
 987     spam-use-spamoracle)
 988   "The spam-list-of-statistical-checks list contains all the mail
 989 splitters that need to have the full message body available.
 990 Note that you should fetch extra headers if you don't like this,
 991 e.g. fetch the 'Received' header for spam-use-blackholes.")
 992
 993 (defun spam-split (&rest specific-checks)
 994   "Split this message into the `spam' group if it is spam.
 995 This function can be used as an entry in the variable `nnmail-split-fancy',
 996 for example like this: (: spam-split).  It can take checks as
 997 parameters.  A string as a parameter will set the
 998 spam-split-group to that string.
 999
1000 See the Info node `(gnus)Fancy Mail Splitting' for more details."
1001   (interactive)
1002   (setq spam-split-last-successful-check nil)
1003   (unless spam-split-disabled
1004     (let ((spam-split-group-choice spam-split-group))
1005       (dolist (check specific-checks)
1006         (when (stringp check)
1007           (setq spam-split-group-choice check)
1008           (setq specific-checks (delq check specific-checks))))
1009
1010       (let ((spam-split-group spam-split-group-choice))
1011         (save-excursion
1012           (save-restriction
1013             (dolist (check spam-list-of-statistical-checks)
1014               (when (and (symbolp check)
1015                          (or (symbol-value check)
1016                              (memq check specific-checks)))
1017                 (widen)
1018                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
1019                               (symbol-name check))
1020                 (return)))
1021             ;;   (progn (widen) (debug (buffer-string)))
1022             (let ((list-of-checks spam-list-of-checks)
1023                   decision)
1024               (while (and list-of-checks (not decision))
1025                 (let ((pair (pop list-of-checks)))
1026                   (when (or
1027                          ;; either, given specific checks, this is one of them
1028                          (and specific-checks (memq (car pair) specific-checks))
1029                          ;; or, given no specific checks, spam-use-CHECK is set
1030                          (and (null specific-checks) (symbol-value (car pair))))
1031                     (gnus-message 5 "spam-split: calling the %s function"
1032                                   (symbol-name (cdr pair)))
1033                     (setq decision (funcall (cdr pair)))
1034                     ;; if we got a decision at all, save the current check
1035                     (when decision
1036                       (setq spam-split-last-successful-check (car pair)))
1037
1038                     (when (eq decision 'spam)
1039                       (unless spam-split-symbolic-return
1040                         (gnus-error
1041                          5
1042                          (format "spam-split got %s but %s is nil"
1043                                  (symbol-name decision)
1044                                  (symbol-name spam-split-symbolic-return))))))))
1045               (if (eq decision t)
1046                   (if spam-split-symbolic-return-positive 'ham nil)
1047                 decision))))))))
1048
1049 (defun spam-find-spam ()
1050   "This function will detect spam in the current newsgroup using spam-split."
1051   (interactive)
1052
1053   (let* ((group gnus-newsgroup-name)
1054          (autodetect (gnus-parameter-spam-autodetect group))
1055          (methods (gnus-parameter-spam-autodetect-methods group))
1056          (first-method (nth 0 methods))
1057          (articles (if spam-autodetect-recheck-messages
1058                        gnus-newsgroup-articles
1059                      gnus-newsgroup-unseen))
1060          article-cannot-be-faked)
1061
1062     (dolist (check spam-list-of-statistical-checks)
1063       (when (and (symbolp check)
1064                  (memq check methods))
1065         (setq article-cannot-be-faked t)
1066         (return)))
1067
1068     (when (memq 'default methods)
1069       (setq article-cannot-be-faked t))
1070
1071     (when (and autodetect
1072                (not (equal first-method 'none)))
1073     (mapcar
1074      (lambda (article)
1075        (let ((id (spam-fetch-field-message-id-fast article))
1076              (subject (spam-fetch-field-subject-fast article))
1077              (sender (spam-fetch-field-from-fast article))
1078              registry-lookup)
1079
1080          (unless id
1081            (gnus-error 5 "Article %d has no message ID!" article))
1082
1083          (when (and id spam-log-to-registry)
1084            (setq registry-lookup (spam-log-registration-type id 'incoming))
1085            (when registry-lookup
1086              (gnus-message
1087               9
1088               "spam-find-spam: message %s was already registered incoming"
1089               id)))
1090
1091          (let* ((spam-split-symbolic-return t)
1092                 (spam-split-symbolic-return-positive t)
1093                 (fake-headers (spam-generate-fake-headers article))
1094                 (split-return
1095                  (or registry-lookup
1096                      (with-temp-buffer
1097                        (if article-cannot-be-faked
1098                            (gnus-request-article-this-buffer
1099                             article
1100                             group)
1101                          ;; else, we fake the article
1102                          (when fake-headers (insert fake-headers)))
1103                        (if (or (null first-method)
1104                                (equal first-method 'default))
1105                            (spam-split)
1106                          (apply 'spam-split methods))))))
1107            (if (equal split-return 'spam)
1108                (gnus-summary-mark-article article gnus-spam-mark))
1109
1110            (when (and id split-return spam-log-to-registry)
1111              (when (zerop (gnus-registry-group-count id))
1112                (gnus-registry-add-group
1113                 id group subject sender))
1114
1115              (unless registry-lookup
1116                (spam-log-processing-to-registry
1117                 id
1118                 'incoming
1119                 split-return
1120                 spam-split-last-successful-check
1121                 group))))))
1122     articles))))
1123
1124 (defvar spam-registration-functions
1125   ;; first the ham register, second the spam register function
1126   ;; third the ham unregister, fourth the spam unregister function
1127   '((spam-use-blacklist  nil
1128                          spam-blacklist-register-routine
1129                          nil
1130                          spam-blacklist-unregister-routine)
1131     (spam-use-whitelist  spam-whitelist-register-routine
1132                          nil
1133                          spam-whitelist-unregister-routine
1134                          nil)
1135     (spam-use-BBDB       spam-BBDB-register-routine
1136                          nil
1137                          spam-BBDB-unregister-routine
1138                          nil)
1139     (spam-use-ifile      spam-ifile-register-ham-routine
1140                          spam-ifile-register-spam-routine
1141                          spam-ifile-unregister-ham-routine
1142                          spam-ifile-unregister-spam-routine)
1143     (spam-use-spamoracle spam-spamoracle-learn-ham
1144                          spam-spamoracle-learn-spam
1145                          spam-spamoracle-unlearn-ham
1146                          spam-spamoracle-unlearn-spam)
1147     (spam-use-stat       spam-stat-register-ham-routine
1148                          spam-stat-register-spam-routine
1149                          spam-stat-unregister-ham-routine
1150                          spam-stat-unregister-spam-routine)
1151     ;; note that spam-use-gmane is not a legitimate check
1152     (spam-use-gmane      nil
1153                          spam-report-gmane-register-routine
1154                          ;; does Gmane support unregistration?
1155                          nil
1156                          nil)
1157     (spam-use-spamassassin spam-spamassassin-register-ham-routine
1158                            spam-spamassassin-register-spam-routine
1159                            spam-spamassassin-unregister-ham-routine
1160                            spam-spamassassin-unregister-spam-routine)
1161     (spam-use-bogofilter spam-bogofilter-register-ham-routine
1162                          spam-bogofilter-register-spam-routine
1163                          spam-bogofilter-unregister-ham-routine
1164                          spam-bogofilter-unregister-spam-routine))
1165   "The spam-registration-functions list contains pairs
1166 associating a parameter variable with the ham and spam
1167 registration functions, and the ham and spam unregistration
1168 functions")
1169
1170 (defun spam-classification-valid-p (classification)
1171   (or  (eq classification 'spam)
1172        (eq classification 'ham)))
1173
1174 (defun spam-process-type-valid-p (process-type)
1175   (or  (eq process-type 'incoming)
1176        (eq process-type 'process)))
1177
1178 (defun spam-registration-check-valid-p (check)
1179   (assoc check spam-registration-functions))
1180
1181 (defun spam-unregistration-check-valid-p (check)
1182   (assoc check spam-registration-functions))
1183
1184 (defun spam-registration-function (classification check)
1185   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1186     (if (eq classification 'spam)
1187         (nth 1 flist)
1188       (nth 0 flist))))
1189
1190 (defun spam-unregistration-function (classification check)
1191   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1192     (if (eq classification 'spam)
1193         (nth 3 flist)
1194       (nth 2 flist))))
1195
1196 (defun spam-list-articles (articles classification)
1197   (let ((mark-check (if (eq classification 'spam)
1198                         'spam-group-spam-mark-p
1199                       'spam-group-ham-mark-p))
1200         list mark-cache-yes mark-cache-no)
1201     (dolist (article articles)
1202       (let ((mark (gnus-summary-article-mark article)))
1203         (unless (memq mark mark-cache-no)
1204           (if (memq mark mark-cache-yes)
1205               (push article list)
1206             ;; else, we have to actually check the mark
1207             (if (funcall mark-check
1208                          gnus-newsgroup-name
1209                          mark)
1210                 (progn
1211                   (push article list)
1212                   (push mark mark-cache-yes))
1213               (push mark mark-cache-no))))))
1214     list))
1215
1216 (defun spam-register-routine (classification
1217                               check
1218                               &optional unregister
1219                               specific-articles)
1220   (when (and (spam-classification-valid-p classification)
1221              (spam-registration-check-valid-p check))
1222     (let* ((register-function
1223             (spam-registration-function classification check))
1224            (unregister-function
1225             (spam-unregistration-function classification check))
1226            (run-function (if unregister
1227                              unregister-function
1228                            register-function))
1229            (log-function (if unregister
1230                              'spam-log-undo-registration
1231                            'spam-log-processing-to-registry))
1232            article articles)
1233
1234       (when run-function
1235         ;; make list of articles, using specific-articles if given
1236         (setq articles (or specific-articles
1237                            (spam-list-articles
1238                             gnus-newsgroup-articles
1239                             classification)))
1240         ;; process them
1241         (gnus-message 5 "%s %d %s articles with classification %s, check %s"
1242                       (if unregister "Unregistering" "Registering")
1243                       (length articles)
1244                       (if specific-articles "specific" "")
1245                       (symbol-name classification)
1246                       (symbol-name check))
1247         (funcall run-function articles)
1248         ;; now log all the registrations (or undo them, depending on unregister)
1249         (dolist (article articles)
1250           (funcall log-function
1251                    (spam-fetch-field-message-id-fast article)
1252                    'process
1253                    classification
1254                    check
1255                    gnus-newsgroup-name))))))
1256
1257 ;;; log a ham- or spam-processor invocation to the registry
1258 (defun spam-log-processing-to-registry (id type classification check group)
1259   (when spam-log-to-registry
1260     (if (and (stringp id)
1261              (stringp group)
1262              (spam-process-type-valid-p type)
1263              (spam-classification-valid-p classification)
1264              (spam-registration-check-valid-p check))
1265         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1266               (cell (list classification check group)))
1267           (push cell cell-list)
1268           (gnus-registry-store-extra-entry
1269            id
1270            type
1271            cell-list))
1272
1273       (gnus-error 5 (format "%s called with bad ID, type, classification, check, or group"
1274                             "spam-log-processing-to-registry")))))
1275
1276 ;;; check if a ham- or spam-processor registration has been done
1277 (defun spam-log-registered-p (id type)
1278   (when spam-log-to-registry
1279     (if (and (stringp id)
1280              (spam-process-type-valid-p type))
1281         (cdr-safe (gnus-registry-fetch-extra id type))
1282       (progn
1283         (gnus-error 5 (format "%s called with bad ID, type, classification, or check"
1284                               "spam-log-registered-p"))
1285         nil))))
1286
1287 ;;; check what a ham- or spam-processor registration says
1288 ;;; returns nil if conflicting registrations are found
1289 (defun spam-log-registration-type (id type)
1290   (let ((count 0)
1291         decision)
1292     (dolist (reg (spam-log-registered-p id type))
1293       (let ((classification (nth 0 reg)))
1294         (when (spam-classification-valid-p classification)
1295           (when (and decision
1296                      (not (eq classification decision)))
1297             (setq count (+ 1 count)))
1298           (setq decision classification))))
1299     (if (< 0 count)
1300         nil
1301       decision)))
1302
1303 ;;; check if a ham- or spam-processor registration needs to be undone
1304 (defun spam-log-unregistration-needed-p (id type classification check)
1305   (when spam-log-to-registry
1306     (if (and (stringp id)
1307              (spam-process-type-valid-p type)
1308              (spam-classification-valid-p classification)
1309              (spam-registration-check-valid-p check))
1310         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1311               found)
1312           (dolist (cell cell-list)
1313             (unless found
1314               (when (and (eq classification (nth 0 cell))
1315                          (eq check (nth 1 cell)))
1316                 (setq found t))))
1317           found)
1318       (progn
1319         (gnus-error 5 (format "%s called with bad ID, type, classification, or check"
1320                               "spam-log-unregistration-needed-p"))
1321         nil))))
1322
1323
1324 ;;; undo a ham- or spam-processor registration (the group is not used)
1325 (defun spam-log-undo-registration (id type classification check &optional group)
1326   (when (and spam-log-to-registry
1327              (spam-log-unregistration-needed-p id type classification check))
1328     (if (and (stringp id)
1329              (spam-process-type-valid-p type)
1330              (spam-classification-valid-p classification)
1331              (spam-registration-check-valid-p check))
1332         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1333               new-cell-list found)
1334           (dolist (cell cell-list)
1335             (unless (and (eq classification (nth 0 cell))
1336                          (eq check (nth 1 cell)))
1337               (push cell new-cell-list)))
1338           (gnus-registry-store-extra-entry
1339            id
1340            type
1341            new-cell-list))
1342       (progn
1343         (gnus-error 5 (format "%s called with bad ID, type, check, or group"
1344                               "spam-log-undo-registration"))
1345         nil))))
1346
1347 ;;; set up IMAP widening if it's necessary
1348 (defun spam-setup-widening ()
1349   (dolist (check spam-list-of-statistical-checks)
1350     (when (symbol-value check)
1351       (setq nnimap-split-download-body-default t))))
1352
1353 \f
1354 ;;;; Regex body
1355
1356 (defun spam-check-regex-body ()
1357   (let ((spam-regex-headers-ham spam-regex-body-ham)
1358         (spam-regex-headers-spam spam-regex-body-spam))
1359     (spam-check-regex-headers t)))
1360
1361 \f
1362 ;;;; Regex headers
1363
1364 (defun spam-check-regex-headers (&optional body)
1365   (let ((type (if body "body" "header"))
1366         (spam-split-group (if spam-split-symbolic-return
1367                               'spam
1368                             spam-split-group))
1369         ret found)
1370     (dolist (h-regex spam-regex-headers-ham)
1371       (unless found
1372         (goto-char (point-min))
1373         (when (re-search-forward h-regex nil t)
1374           (message "Ham regex %s search positive." type)
1375           (setq found t))))
1376     (dolist (s-regex spam-regex-headers-spam)
1377       (unless found
1378         (goto-char (point-min))
1379         (when (re-search-forward s-regex nil t)
1380           (message "Spam regex %s search positive." type)
1381           (setq found t)
1382           (setq ret spam-split-group))))
1383     ret))
1384
1385 \f
1386 ;;;; Blackholes.
1387
1388 (defun spam-reverse-ip-string (ip)
1389   (when (stringp ip)
1390     (mapconcat 'identity
1391                (nreverse (split-string ip "\\."))
1392                ".")))
1393
1394 (defun spam-check-blackholes ()
1395   "Check the Received headers for blackholed relays."
1396   (let ((headers (message-fetch-field "received"))
1397         (spam-split-group (if spam-split-symbolic-return
1398                               'spam
1399                             spam-split-group))
1400         ips matches)
1401     (when headers
1402       (with-temp-buffer
1403         (insert headers)
1404         (goto-char (point-min))
1405         (gnus-message 5 "Checking headers for relay addresses")
1406         (while (re-search-forward
1407                 "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
1408           (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
1409           (push (spam-reverse-ip-string (match-string 1))
1410                 ips)))
1411       (dolist (server spam-blackhole-servers)
1412         (dolist (ip ips)
1413           (unless (and spam-blackhole-good-server-regex
1414                        ;; match the good-server-regex against the reversed (again) IP string
1415                        (string-match
1416                         spam-blackhole-good-server-regex
1417                         (spam-reverse-ip-string ip)))
1418             (unless matches
1419               (let ((query-string (concat ip "." server)))
1420                 (if spam-use-dig
1421                     (let ((query-result (query-dig query-string)))
1422                       (when query-result
1423                         (gnus-message 5 "(DIG): positive blackhole check '%s'"
1424                                       query-result)
1425                         (push (list ip server query-result)
1426                               matches)))
1427                   ;; else, if not using dig.el
1428                   (when (query-dns query-string)
1429                     (gnus-message 5 "positive blackhole check")
1430                     (push (list ip server (query-dns query-string 'TXT))
1431                           matches)))))))))
1432     (when matches
1433       spam-split-group)))
1434 \f
1435 ;;;; Hashcash.
1436
1437 (condition-case nil
1438     (progn
1439       (require 'hashcash)
1440
1441       (defun spam-check-hashcash ()
1442         "Check the headers for hashcash payments."
1443         (mail-check-payment)))   ;mail-check-payment returns a boolean
1444
1445   (file-error (progn
1446                 (defalias 'mail-check-payment 'ignore)
1447                 (defalias 'spam-check-hashcash 'ignore))))
1448 \f
1449 ;;;; BBDB
1450
1451 ;;; original idea for spam-check-BBDB from Alexander Kotelnikov
1452 ;;; <sacha@giotto.sj.ru>
1453
1454 ;; all this is done inside a condition-case to trap errors
1455
1456 (condition-case nil
1457     (progn
1458       (require 'bbdb)
1459       (require 'bbdb-com)
1460
1461       ;; when the BBDB changes, we want to clear out our cache
1462       (defun spam-clear-cache-BBDB (&rest immaterial)
1463         (spam-clear-cache 'spam-use-BBDB))
1464
1465       (add-hook 'bbdb-change-hook 'spam-clear-cache-BBDB)
1466
1467       (defun spam-enter-ham-BBDB (addresses &optional remove)
1468         "Enter an address into the BBDB; implies ham (non-spam) sender"
1469         (dolist (from addresses)
1470           (when (stringp from)
1471             (let* ((parsed-address (gnus-extract-address-components from))
1472                    (name (or (nth 0 parsed-address) "Ham Sender"))
1473                    (remove-function (if remove
1474                                         'bbdb-delete-record-internal
1475                                       'ignore))
1476                    (net-address (nth 1 parsed-address))
1477                    (record (and net-address
1478                                 (bbdb-search-simple nil net-address))))
1479               (when net-address
1480                 (gnus-message 5 "%s address %s %s BBDB"
1481                               (if remove "Deleting" "Adding")
1482                               from
1483                               (if remove "from" "to"))
1484                 (if record
1485                     (funcall remove-function record)
1486                   (bbdb-create-internal name nil net-address nil nil
1487                                         "ham sender added by spam.el")))))))
1488
1489       (defun spam-BBDB-register-routine (articles &optional unregister)
1490         (let (addresses)
1491           (dolist (article articles)
1492             (when (stringp (spam-fetch-field-from-fast article))
1493               (push (spam-fetch-field-from-fast article) addresses)))
1494           ;; now do the register/unregister action
1495           (spam-enter-ham-BBDB addresses unregister)))
1496
1497       (defun spam-BBDB-unregister-routine (articles)
1498         (spam-BBDB-register-routine articles t))
1499
1500       (defun spam-check-BBDB ()
1501         "Mail from people in the BBDB is classified as ham or non-spam"
1502         (let ((who (message-fetch-field "from"))
1503               (spam-split-group (if spam-split-symbolic-return
1504                                     'spam
1505                                   spam-split-group))
1506               bbdb-cache bbdb-hashtable)
1507           (when spam-cache-lookups
1508             (setq bbdb-cache (gethash 'spam-use-BBDB spam-caches))
1509             (unless bbdb-cache
1510               (setq bbdb-cache
1511                     ;; this is the expanded (bbdb-hashtable) macro
1512                     ;; without the debugging support
1513                     (with-current-buffer (bbdb-buffer)
1514                       (save-excursion
1515                         (save-window-excursion
1516                           (bbdb-records nil t)
1517                           bbdb-hashtable))))
1518               (puthash 'spam-use-BBDB bbdb-cache spam-caches)))
1519           (when who
1520             (setq who (nth 1 (gnus-extract-address-components who)))
1521             (if
1522                 (if spam-cache-lookups
1523                     (symbol-value
1524                      (intern-soft who bbdb-cache))
1525                   (bbdb-search-simple nil who))
1526                 t
1527               (if spam-use-BBDB-exclusive
1528                   spam-split-group
1529                 nil))))))
1530
1531   (file-error (progn
1532                 (defalias 'bbdb-search-simple 'ignore)
1533                 (defalias 'bbdb-records 'ignore)
1534                 (defalias 'bbdb-buffer 'ignore)
1535                 (defalias 'spam-check-BBDB 'ignore)
1536                 (defalias 'spam-BBDB-register-routine 'ignore)
1537                 (defalias 'spam-enter-ham-BBDB 'ignore)
1538                 (defalias 'bbdb-create-internal 'ignore)
1539                 (defalias 'bbdb-delete-record-internal 'ignore)
1540                 (defalias 'bbdb-records 'ignore))))
1541
1542 \f
1543 ;;;; ifile
1544
1545 ;;; check the ifile backend; return nil if the mail was NOT classified
1546 ;;; as spam
1547
1548 (defun spam-get-ifile-database-parameter ()
1549   "Get the command-line parameter for ifile's database from
1550   spam-ifile-database-path."
1551   (if spam-ifile-database-path
1552       (format "--db-file=%s" spam-ifile-database-path)
1553     nil))
1554
1555 (defun spam-check-ifile ()
1556   "Check the ifile backend for the classification of this message."
1557   (let ((article-buffer-name (buffer-name))
1558         (spam-split-group (if spam-split-symbolic-return
1559                               'spam
1560                             spam-split-group))
1561         category return)
1562     (with-temp-buffer
1563       (let ((temp-buffer-name (buffer-name))
1564             (db-param (spam-get-ifile-database-parameter)))
1565         (save-excursion
1566           (set-buffer article-buffer-name)
1567           (apply 'call-process-region
1568                  (point-min) (point-max) spam-ifile-path
1569                  nil temp-buffer-name nil "-c"
1570                  (if db-param `(,db-param "-q") `("-q"))))
1571         ;; check the return now (we're back in the temp buffer)
1572         (goto-char (point-min))
1573         (if (not (eobp))
1574             (setq category (buffer-substring (point) (point-at-eol))))
1575         (when (not (zerop (length category))) ; we need a category here
1576           (if spam-ifile-all-categories
1577               (setq return category)
1578             ;; else, if spam-ifile-all-categories is not set...
1579             (when (string-equal spam-ifile-spam-category category)
1580               (setq return spam-split-group)))))) ; note return is nil otherwise
1581     return))
1582
1583 (defun spam-ifile-register-with-ifile (articles category &optional unregister)
1584   "Register an article, given as a string, with a category.
1585 Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
1586   (let ((category (or category gnus-newsgroup-name))
1587         (add-or-delete-option (if unregister "-d" "-i"))
1588         (db (spam-get-ifile-database-parameter))
1589         parameters)
1590     (with-temp-buffer
1591       (dolist (article articles)
1592         (let ((article-string (spam-get-article-as-string article)))
1593           (when (stringp article-string)
1594             (insert article-string))))
1595       (apply 'call-process-region
1596              (point-min) (point-max) spam-ifile-path
1597              nil nil nil
1598              add-or-delete-option category
1599              (if db `(,db "-h") `("-h"))))))
1600
1601 (defun spam-ifile-register-spam-routine (articles &optional unregister)
1602   (spam-ifile-register-with-ifile articles spam-ifile-spam-category unregister))
1603
1604 (defun spam-ifile-unregister-spam-routine (articles)
1605   (spam-ifile-register-spam-routine articles t))
1606
1607 (defun spam-ifile-register-ham-routine (articles &optional unregister)
1608   (spam-ifile-register-with-ifile articles spam-ifile-ham-category unregister))
1609
1610 (defun spam-ifile-unregister-ham-routine (articles)
1611   (spam-ifile-register-ham-routine articles t))
1612
1613 \f
1614 ;;;; spam-stat
1615
1616 (condition-case nil
1617     (progn
1618       (let ((spam-stat-install-hooks nil))
1619         (require 'spam-stat))
1620
1621       (defun spam-check-stat ()
1622         "Check the spam-stat backend for the classification of this message"
1623         (let ((spam-split-group (if spam-split-symbolic-return
1624                                     'spam
1625                                   spam-split-group))
1626               (spam-stat-split-fancy-spam-group spam-split-group) ; override
1627               (spam-stat-buffer (buffer-name)) ; stat the current buffer
1628               category return)
1629           (spam-stat-split-fancy)))
1630
1631       (defun spam-stat-register-spam-routine (articles &optional unregister)
1632         (dolist (article articles)
1633           (let ((article-string (spam-get-article-as-string article)))
1634             (with-temp-buffer
1635               (insert article-string)
1636               (if unregister
1637                   (spam-stat-buffer-change-to-non-spam)
1638               (spam-stat-buffer-is-spam))))))
1639
1640       (defun spam-stat-unregister-spam-routine (articles)
1641         (spam-stat-register-spam-routine articles t))
1642
1643       (defun spam-stat-register-ham-routine (articles &optional unregister)
1644         (dolist (article articles)
1645           (let ((article-string (spam-get-article-as-string article)))
1646             (with-temp-buffer
1647               (insert article-string)
1648               (if unregister
1649                   (spam-stat-buffer-change-to-spam)
1650               (spam-stat-buffer-is-non-spam))))))
1651
1652       (defun spam-stat-unregister-ham-routine (articles)
1653         (spam-stat-register-ham-routine articles t))
1654
1655       (defun spam-maybe-spam-stat-load ()
1656         (when spam-use-stat (spam-stat-load)))
1657
1658       (defun spam-maybe-spam-stat-save ()
1659         (when spam-use-stat (spam-stat-save))))
1660
1661   (file-error (progn
1662                 (defalias 'spam-stat-load 'ignore)
1663                 (defalias 'spam-stat-save 'ignore)
1664                 (defalias 'spam-maybe-spam-stat-load 'ignore)
1665                 (defalias 'spam-maybe-spam-stat-save 'ignore)
1666                 (defalias 'spam-stat-register-ham-routine 'ignore)
1667                 (defalias 'spam-stat-unregister-ham-routine 'ignore)
1668                 (defalias 'spam-stat-register-spam-routine 'ignore)
1669                 (defalias 'spam-stat-unregister-spam-routine 'ignore)
1670                 (defalias 'spam-stat-buffer-is-spam 'ignore)
1671                 (defalias 'spam-stat-buffer-change-to-spam 'ignore)
1672                 (defalias 'spam-stat-buffer-is-non-spam 'ignore)
1673                 (defalias 'spam-stat-buffer-change-to-non-spam 'ignore)
1674                 (defalias 'spam-stat-split-fancy 'ignore)
1675                 (defalias 'spam-check-stat 'ignore))))
1676
1677 \f
1678
1679 ;;;; Blacklists and whitelists.
1680
1681 (defvar spam-whitelist-cache nil)
1682 (defvar spam-blacklist-cache nil)
1683
1684 (defun spam-kill-whole-line ()
1685   (beginning-of-line)
1686   (let ((kill-whole-line t))
1687     (kill-line)))
1688
1689 ;;; address can be a list, too
1690 (defun spam-enter-whitelist (address &optional remove)
1691   "Enter ADDRESS (list or single) into the whitelist.
1692 With a non-nil REMOVE, remove them."
1693   (interactive "sAddress: ")
1694   (spam-enter-list address spam-whitelist remove)
1695   (setq spam-whitelist-cache nil)
1696   (spam-clear-cache 'spam-use-whitelist))
1697
1698 ;;; address can be a list, too
1699 (defun spam-enter-blacklist (address &optional remove)
1700   "Enter ADDRESS (list or single) into the blacklist.
1701 With a non-nil REMOVE, remove them."
1702   (interactive "sAddress: ")
1703   (spam-enter-list address spam-blacklist remove)
1704   (setq spam-blacklist-cache nil)
1705   (spam-clear-cache 'spam-use-whitelist))
1706
1707 (defun spam-enter-list (addresses file &optional remove)
1708   "Enter ADDRESSES into the given FILE.
1709 Either the whitelist or the blacklist files can be used.  With
1710 REMOVE not nil, remove the ADDRESSES."
1711   (if (stringp addresses)
1712       (spam-enter-list (list addresses) file remove)
1713     ;; else, we have a list of addresses here
1714     (unless (file-exists-p (file-name-directory file))
1715       (make-directory (file-name-directory file) t))
1716     (save-excursion
1717       (set-buffer
1718        (find-file-noselect file))
1719       (dolist (a addresses)
1720         (when (stringp a)
1721           (goto-char (point-min))
1722           (if (re-search-forward (regexp-quote a) nil t)
1723               ;; found the address
1724               (when remove
1725                 (spam-kill-whole-line))
1726             ;; else, the address was not found
1727             (unless remove
1728               (goto-char (point-max))
1729               (unless (bobp)
1730                 (insert "\n"))
1731               (insert a "\n")))))
1732       (save-buffer))))
1733
1734 (defun spam-filelist-build-cache (type)
1735   (let ((cache (if (eq type 'spam-use-blacklist)
1736                    spam-blacklist-cache
1737                  spam-whitelist-cache))
1738         parsed-cache)
1739     (unless (gethash type spam-caches)
1740       (while cache
1741         (let ((address (pop cache)))
1742           (unless (zerop (length address)) ; 0 for a nil address too
1743             (setq address (regexp-quote address))
1744             ;; fix regexp-quote's treatment of user-intended regexes
1745             (while (string-match "\\\\\\*" address)
1746               (setq address (replace-match ".*" t t address))))
1747           (push address parsed-cache)))
1748       (puthash type parsed-cache spam-caches))))
1749
1750 (defun spam-filelist-check-cache (type from)
1751   (when (stringp from)
1752     (spam-filelist-build-cache type)
1753     (let (found)
1754       (dolist (address (gethash type spam-caches))
1755         (when (and address (string-match address from))
1756           (setq found t)
1757           (return)))
1758       found)))
1759
1760 ;;; returns t if the sender is in the whitelist, nil or
1761 ;;; spam-split-group otherwise
1762 (defun spam-check-whitelist ()
1763   ;; FIXME!  Should it detect when file timestamps change?
1764   (let ((spam-split-group (if spam-split-symbolic-return
1765                               'spam
1766                             spam-split-group)))
1767     (unless spam-whitelist-cache
1768       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
1769     (if (spam-from-listed-p 'spam-use-whitelist)
1770         t
1771       (if spam-use-whitelist-exclusive
1772           spam-split-group
1773         nil))))
1774
1775 (defun spam-check-blacklist ()
1776   ;; FIXME!  Should it detect when file timestamps change?
1777   (let ((spam-split-group (if spam-split-symbolic-return
1778                               'spam
1779                             spam-split-group)))
1780     (unless spam-blacklist-cache
1781       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
1782     (and (spam-from-listed-p 'spam-use-blacklist) spam-split-group)))
1783
1784 (defun spam-parse-list (file)
1785   (when (file-readable-p file)
1786     (let (contents address)
1787       (with-temp-buffer
1788         (insert-file-contents file)
1789         (while (not (eobp))
1790           (setq address (buffer-substring (point) (point-at-eol)))
1791           (forward-line 1)
1792           ;; insert the e-mail address if detected, otherwise the raw data
1793           (unless (zerop (length address))
1794             (let ((pure-address (nth 1 (gnus-extract-address-components address))))
1795               (push (or pure-address address) contents)))))
1796       (nreverse contents))))
1797
1798 (defun spam-from-listed-p (type)
1799   (let ((from (message-fetch-field "from"))
1800         found)
1801     (spam-filelist-check-cache type from)))
1802
1803 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
1804   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
1805         (declassification (if blacklist 'ham 'spam))
1806         (enter-function
1807          (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
1808         (remove-function
1809          (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
1810         from addresses unregister-list)
1811     (dolist (article articles)
1812       (let ((from (spam-fetch-field-from-fast article))
1813             (id (spam-fetch-field-message-id-fast article))
1814             sender-ignored)
1815         (when (stringp from)
1816           (dolist (ignore-regex spam-blacklist-ignored-regexes)
1817             (when (and (not sender-ignored)
1818                        (stringp ignore-regex)
1819                        (string-match ignore-regex from))
1820               (setq sender-ignored t)))
1821           ;; remember the messages we need to unregister, unless remove is set
1822           (when (and
1823                  (null unregister)
1824                  (spam-log-unregistration-needed-p
1825                   id 'process declassification de-symbol))
1826             (push from unregister-list))
1827           (unless sender-ignored
1828             (push from addresses)))))
1829
1830     (if unregister
1831         (funcall enter-function addresses t) ; unregister all these addresses
1832       ;; else, register normally and unregister what we need to
1833       (funcall remove-function unregister-list t)
1834       (dolist (article unregister-list)
1835         (spam-log-undo-registration
1836          (spam-fetch-field-message-id-fast article)
1837          'process
1838          declassification
1839          de-symbol))
1840       (funcall enter-function addresses nil))))
1841
1842 (defun spam-blacklist-unregister-routine (articles)
1843   (spam-blacklist-register-routine articles t))
1844
1845 (defun spam-blacklist-register-routine (articles &optional unregister)
1846   (spam-filelist-register-routine articles t unregister))
1847
1848 (defun spam-whitelist-unregister-routine (articles)
1849   (spam-whitelist-register-routine articles t))
1850
1851 (defun spam-whitelist-register-routine (articles &optional unregister)
1852   (spam-filelist-register-routine articles nil unregister))
1853
1854 \f
1855 ;;;; Spam-report glue
1856 (defun spam-report-gmane-register-routine (articles)
1857   (when articles
1858     (apply 'spam-report-gmane articles)))
1859
1860 \f
1861 ;;;; Bogofilter
1862 (defun spam-check-bogofilter-headers (&optional score)
1863   (let ((header (message-fetch-field spam-bogofilter-header))
1864         (spam-split-group (if spam-split-symbolic-return
1865                               'spam
1866                             spam-split-group)))
1867     (when header                        ; return nil when no header
1868       (if score                         ; scoring mode
1869           (if (string-match "spamicity=\\([0-9.]+\\)" header)
1870               (match-string 1 header)
1871             "0")
1872         ;; spam detection mode
1873         (when (string-match spam-bogofilter-bogosity-positive-spam-header
1874                             header)
1875           spam-split-group)))))
1876
1877 ;; return something sensible if the score can't be determined
1878 (defun spam-bogofilter-score ()
1879   "Get the Bogofilter spamicity score"
1880   (interactive)
1881   (save-window-excursion
1882     (gnus-summary-show-article t)
1883     (set-buffer gnus-article-buffer)
1884     (let ((score (or (spam-check-bogofilter-headers t)
1885                      (spam-check-bogofilter t))))
1886       (gnus-summary-show-article)
1887       (message "Spamicity score %s" score)
1888       (or score "0"))))
1889
1890 (defun spam-check-bogofilter (&optional score)
1891   "Check the Bogofilter backend for the classification of this message"
1892   (let ((article-buffer-name (buffer-name))
1893         (db spam-bogofilter-database-directory)
1894         return)
1895     (with-temp-buffer
1896       (let ((temp-buffer-name (buffer-name)))
1897         (save-excursion
1898           (set-buffer article-buffer-name)
1899           (apply 'call-process-region
1900                  (point-min) (point-max)
1901                  spam-bogofilter-path
1902                  nil temp-buffer-name nil
1903                  (if db `("-d" ,db "-v") `("-v"))))
1904         (setq return (spam-check-bogofilter-headers score))))
1905     return))
1906
1907 (defun spam-bogofilter-register-with-bogofilter (articles
1908                                                  spam
1909                                                  &optional unregister)
1910   "Register an article, given as a string, as spam or non-spam."
1911   (dolist (article articles)
1912     (let ((article-string (spam-get-article-as-string article))
1913           (db spam-bogofilter-database-directory)
1914           (switch (if unregister
1915                       (if spam
1916                           spam-bogofilter-spam-strong-switch
1917                         spam-bogofilter-ham-strong-switch)
1918                     (if spam
1919                         spam-bogofilter-spam-switch
1920                       spam-bogofilter-ham-switch))))
1921       (when (stringp article-string)
1922         (with-temp-buffer
1923           (insert article-string)
1924
1925           (apply 'call-process-region
1926                  (point-min) (point-max)
1927                  spam-bogofilter-path
1928                  nil nil nil switch
1929                  (if db `("-d" ,db "-v") `("-v"))))))))
1930
1931 (defun spam-bogofilter-register-spam-routine (articles &optional unregister)
1932   (spam-bogofilter-register-with-bogofilter articles t unregister))
1933
1934 (defun spam-bogofilter-unregister-spam-routine (articles)
1935   (spam-bogofilter-register-spam-routine articles t))
1936
1937 (defun spam-bogofilter-register-ham-routine (articles &optional unregister)
1938   (spam-bogofilter-register-with-bogofilter articles nil unregister))
1939
1940 (defun spam-bogofilter-unregister-ham-routine (articles)
1941   (spam-bogofilter-register-ham-routine articles t))
1942
1943
1944 \f
1945 ;;;; spamoracle
1946 (defun spam-check-spamoracle ()
1947   "Run spamoracle on an article to determine whether it's spam."
1948   (let ((article-buffer-name (buffer-name))
1949         (spam-split-group (if spam-split-symbolic-return
1950                               'spam
1951                             spam-split-group)))
1952     (with-temp-buffer
1953       (let ((temp-buffer-name (buffer-name)))
1954         (save-excursion
1955           (set-buffer article-buffer-name)
1956           (let ((status
1957                  (apply 'call-process-region
1958                         (point-min) (point-max)
1959                         spam-spamoracle-binary
1960                         nil temp-buffer-name nil
1961                         (if spam-spamoracle-database
1962                             `("-f" ,spam-spamoracle-database "mark")
1963                           '("mark")))))
1964             (if (eq 0 status)
1965                 (progn
1966                   (set-buffer temp-buffer-name)
1967                   (goto-char (point-min))
1968                   (when (re-search-forward "^X-Spam: yes;" nil t)
1969                     spam-split-group))
1970               (error "Error running spamoracle" status))))))))
1971
1972 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
1973   "Run spamoracle in training mode."
1974   (with-temp-buffer
1975     (let ((temp-buffer-name (buffer-name)))
1976       (save-excursion
1977         (goto-char (point-min))
1978         (dolist (article articles)
1979           (insert (spam-get-article-as-string article)))
1980         (let* ((arg (if (spam-xor unregister article-is-spam-p)
1981                         "-spam"
1982                       "-good"))
1983                (status
1984                 (apply 'call-process-region
1985                        (point-min) (point-max)
1986                        spam-spamoracle-binary
1987                        nil temp-buffer-name nil
1988                        (if spam-spamoracle-database
1989                            `("-f" ,spam-spamoracle-database
1990                              "add" ,arg)
1991                          `("add" ,arg)))))
1992           (when (not (eq 0 status))
1993             (error "Error running spamoracle" status)))))))
1994
1995 (defun spam-spamoracle-learn-ham (articles &optional unregister)
1996   (spam-spamoracle-learn articles nil unregister))
1997
1998 (defun spam-spamoracle-unlearn-ham (articles &optional unregister)
1999   (spam-spamoracle-learn-ham articles t))
2000
2001 (defun spam-spamoracle-learn-spam (articles &optional unregister)
2002   (spam-spamoracle-learn articles t unregister))
2003
2004 (defun spam-spamoracle-unlearn-spam (articles &optional unregister)
2005   (spam-spamoracle-learn-spam articles t))
2006
2007 \f
2008 ;;;; SpamAssassin
2009 ;;; based mostly on the bogofilter code
2010 (defun spam-check-spamassassin-headers (&optional score)
2011   "Check the SpamAssassin headers for the classification of this message."
2012   (if score                             ; scoring mode
2013       (let ((header (message-fetch-field spam-spamassassin-spam-status-header)))
2014         (when header
2015           (if (string-match "hits=\\(-?[0-9.]+\\)" header)
2016               (match-string 1 header)
2017             "0")))
2018     ;; spam detection mode
2019     (let ((header (message-fetch-field spam-spamassassin-spam-flag-header))
2020           (spam-split-group (if spam-split-symbolic-return
2021                                  'spam
2022                                spam-split-group)))
2023           (when header                  ; return nil when no header
2024             (when (string-match spam-spamassassin-positive-spam-flag-header
2025                                 header)
2026               spam-split-group)))))
2027
2028 (defun spam-check-spamassassin (&optional score)
2029   "Check the SpamAssassin backend for the classification of this message."
2030   (let ((article-buffer-name (buffer-name)))
2031     (with-temp-buffer
2032       (let ((temp-buffer-name (buffer-name)))
2033         (save-excursion
2034           (set-buffer article-buffer-name)
2035           (apply 'call-process-region
2036                  (point-min) (point-max) spam-spamassassin-path
2037                  nil temp-buffer-name nil spam-spamassassin-arguments))
2038         ;; check the return now (we're back in the temp buffer)
2039         (goto-char (point-min))
2040         (spam-check-spamassassin-headers score)))))
2041
2042 ;; return something sensible if the score can't be determined
2043 (defun spam-spamassassin-score ()
2044   "Get the SpamAssassin score"
2045   (interactive)
2046   (save-window-excursion
2047     (gnus-summary-show-article t)
2048     (set-buffer gnus-article-buffer)
2049     (let ((score (or (spam-check-spamassassin-headers t)
2050                      (spam-check-spamassassin t))))
2051       (gnus-summary-show-article)
2052       (message "SpamAssassin score %s" score)
2053       (or score "0"))))
2054
2055 (defun spam-spamassassin-register-with-sa-learn (articles spam
2056                                                  &optional unregister)
2057   "Register articles with spamassassin's sa-learn as spam or non-spam."
2058   (if articles
2059       (let ((action (if unregister spam-sa-learn-unregister-switch
2060                       (if spam spam-sa-learn-spam-switch
2061                         spam-sa-learn-ham-switch)))
2062             (summary-buffer-name (buffer-name)))
2063         (with-temp-buffer
2064           ;; group the articles into mbox format
2065           (dolist (article articles)
2066             (let (article-string)
2067               (save-excursion
2068                 (set-buffer summary-buffer-name)
2069                 (setq article-string (spam-get-article-as-string article)))
2070               (when (stringp article-string)
2071                 (insert "From \n") ; mbox separator (sa-learn only checks the
2072                                    ; first five chars, so we can get away with
2073                                    ; a bogus line))
2074                 (insert article-string)
2075                 (insert "\n"))))
2076           ;; call sa-learn on all messages at the same time
2077           (apply 'call-process-region
2078                  (point-min) (point-max)
2079                  spam-sa-learn-path
2080                  nil nil nil "--mbox"
2081                  (if spam-sa-learn-rebuild
2082                      (list action)
2083                    `("--no-rebuild" ,action)))))))
2084
2085 (defun spam-spamassassin-register-spam-routine (articles &optional unregister)
2086   (spam-spamassassin-register-with-sa-learn articles t unregister))
2087
2088 (defun spam-spamassassin-register-ham-routine (articles &optional unregister)
2089   (spam-spamassassin-register-with-sa-learn articles nil unregister))
2090
2091 (defun spam-assassin-register-spam-routine (articles)
2092   (spam-spamassassin-register-with-sa-learn articles t t))
2093
2094 (defun spam-assassin-register-ham-routine (articles)
2095   (spam-spamassassin-register-with-sa-learn articles nil t))
2096 \f
2097 ;;;; Hooks
2098
2099 ;;;###autoload
2100 (defun spam-initialize ()
2101   "Install the spam.el hooks and do other initialization"
2102   (interactive)
2103   (setq spam-install-hooks t)
2104   ;; TODO: How do we redo this every time spam-face is customized?
2105   (push '((eq mark gnus-spam-mark) . spam-face)
2106         gnus-summary-highlight)
2107   ;; Add hooks for loading and saving the spam stats
2108   (add-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2109   (add-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2110   (add-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2111   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2112   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2113   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2114   (add-hook 'gnus-summary-prepared-hook 'spam-find-spam))
2115
2116 (defun spam-unload-hook ()
2117   "Uninstall the spam.el hooks"
2118   (interactive)
2119   (remove-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2120   (remove-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2121   (remove-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2122   (remove-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2123   (remove-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2124   (remove-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2125   (remove-hook 'gnus-summary-prepare-hook 'spam-find-spam))
2126
2127 (when spam-install-hooks
2128   (spam-initialize))
2129
2130 (provide 'spam)
2131
2132 ;;; spam.el ends here