cgit.sxemacs.org Git - gnus/blob - lisp/spam.el

   1 ;;; spam.el --- Identifying spam
   2 ;; Copyright (C) 2002, 2003, 2004 Free Software Foundation, Inc.
   3
   4 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
   5 ;; Keywords: network
   6
   7 ;; This file is part of GNU Emacs.
   8
   9 ;; GNU Emacs is free software; you can redistribute it and/or modify
  10 ;; it under the terms of the GNU General Public License as published by
  11 ;; the Free Software Foundation; either version 2, or (at your option)
  12 ;; any later version.
  13
  14 ;; GNU Emacs is distributed in the hope that it will be useful,
  15 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17 ;; GNU General Public License for more details.
  18
  19 ;; You should have received a copy of the GNU General Public License
  20 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  21 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  22 ;; Boston, MA 02111-1307, USA.
  23
  24 ;;; Commentary:
  25
  26 ;;; This module addresses a few aspects of spam control under Gnus.  Page
  27 ;;; breaks are used for grouping declarations and documentation relating to
  28 ;;; each particular aspect.
  29
  30 ;;; The integration with Gnus is not yet complete.  See various `FIXME'
  31 ;;; comments, below, for supplementary explanations or discussions.
  32
  33 ;;; Several TODO items are marked as such
  34
  35 ;; TODO: spam scores, detection of spam in newsgroups, cross-server splitting,
  36 ;; remote processing, training through files
  37
  38 ;;; Code:
  39
  40 (eval-when-compile (require 'cl))
  41
  42 (require 'gnus-sum)
  43
  44 (require 'gnus-uu)                      ; because of key prefix issues
  45 ;;; for the definitions of group content classification and spam processors
  46 (require 'gnus)
  47 (require 'message)              ;for the message-fetch-field functions
  48
  49 ;; for nnimap-split-download-body-default
  50 (eval-when-compile (require 'nnimap))
  51
  52 ;; autoload query-dig
  53 (eval-and-compile
  54   (autoload 'query-dig "dig"))
  55
  56 ;; autoload spam-report
  57 (eval-and-compile
  58   (autoload 'spam-report-gmane "spam-report"))
  59
  60 ;; autoload gnus-registry
  61 (eval-and-compile
  62   (autoload 'gnus-registry-group-count "gnus-registry")
  63   (autoload 'gnus-registry-add-group "gnus-registry")
  64   (autoload 'gnus-registry-store-extra-entry "gnus-registry")
  65   (autoload 'gnus-registry-fetch-extra "gnus-registry"))
  66
  67 ;; autoload query-dns
  68 (eval-and-compile
  69   (autoload 'query-dns "dns"))
  70
  71 ;;; Main parameters.
  72
  73 (defgroup spam nil
  74   "Spam configuration.")
  75
  76 (defcustom spam-directory (nnheader-concat gnus-directory "spam/")
  77   "Directory for spam whitelists and blacklists."
  78   :type 'directory
  79   :group 'spam)
  80
  81 (defcustom spam-move-spam-nonspam-groups-only t
  82   "Whether spam should be moved in non-spam groups only.
  83 When t, only ham and unclassified groups will have their spam moved
  84 to the spam-process-destination.  When nil, spam will also be moved from
  85 spam groups."
  86   :type 'boolean
  87   :group 'spam)
  88
  89 (defcustom spam-process-ham-in-nonham-groups nil
  90   "Whether ham should be processed in non-ham groups."
  91   :type 'boolean
  92   :group 'spam)
  93
  94 (defcustom spam-log-to-registry nil
  95   "Whether spam/ham processing should be logged in the registry."
  96   :type 'boolean
  97   :group 'spam)
  98
  99 (defcustom spam-split-symbolic-return nil
 100   "Whether `spam-split' should work with symbols or group names."
 101   :type 'boolean
 102   :group 'spam)
 103
 104 (defcustom spam-split-symbolic-return-positive nil
 105   "Whether `spam-split' should ALWAYS work with symbols or group names.
 106 Do not set this if you use `spam-split' in a fancy split
 107   method."
 108   :type 'boolean
 109   :group 'spam)
 110
 111 (defcustom spam-process-ham-in-spam-groups nil
 112   "Whether ham should be processed in spam groups."
 113   :type 'boolean
 114   :group 'spam)
 115
 116 (defcustom spam-mark-only-unseen-as-spam t
 117   "Whether only unseen articles should be marked as spam in spam groups.
 118 When nil, all unread articles in a spam group are marked as
 119 spam.  Set this if you want to leave an article unread in a spam group
 120 without losing it to the automatic spam-marking process."
 121   :type 'boolean
 122   :group 'spam)
 123
 124 (defcustom spam-mark-ham-unread-before-move-from-spam-group nil
 125   "Whether ham should be marked unread before it's moved.
 126 The article is moved out of a spam group according to ham-process-destination.
 127 This variable is an official entry in the international Longest Variable Name
 128 Competition."
 129   :type 'boolean
 130   :group 'spam)
 131
 132 (defcustom spam-disable-spam-split-during-ham-respool nil
 133   "Whether `spam-split' should be ignored while resplitting ham.
 134 This is useful to prevent ham from ending up in the same spam
 135 group after the resplit.  Don't set this to t if you have `spam-split' as the
 136 last rule in your split configuration."
 137   :type 'boolean
 138   :group 'spam)
 139
 140 (defcustom spam-autodetect-recheck-messages nil
 141   "Should spam.el recheck all meessages when autodetecting?
 142 Normally this is nil, so only unseen messages will be checked."
 143   :type 'boolean
 144   :group 'spam)
 145
 146 (defcustom spam-whitelist (expand-file-name "whitelist" spam-directory)
 147   "The location of the whitelist.
 148 The file format is one regular expression per line.
 149 The regular expression is matched against the address."
 150   :type 'file
 151   :group 'spam)
 152
 153 (defcustom spam-blacklist (expand-file-name "blacklist" spam-directory)
 154   "The location of the blacklist.
 155 The file format is one regular expression per line.
 156 The regular expression is matched against the address."
 157   :type 'file
 158   :group 'spam)
 159
 160 (defcustom spam-use-dig t
 161   "Whether `query-dig' should be used instead of `query-dns'."
 162   :type 'boolean
 163   :group 'spam)
 164
 165 (defcustom spam-use-gmane-xref nil
 166   "Whether the Gmane spam xref should be used by `spam-split'."
 167   :type 'boolean
 168   :group 'spam)
 169
 170 (defcustom spam-use-blacklist nil
 171   "Whether the blacklist should be used by `spam-split'."
 172   :type 'boolean
 173   :group 'spam)
 174
 175 (defcustom spam-blacklist-ignored-regexes nil
 176   "Regular expressions that the blacklist should ignore."
 177   :type '(repeat (regexp :tag "Regular expression to ignore when blacklisting"))
 178   :group 'spam)
 179
 180 (defcustom spam-use-whitelist nil
 181   "Whether the whitelist should be used by `spam-split'."
 182   :type 'boolean
 183   :group 'spam)
 184
 185 (defcustom spam-use-whitelist-exclusive nil
 186   "Whether whitelist-exclusive should be used by `spam-split'.
 187 Exclusive whitelisting means that all messages from senders not in the whitelist
 188 are considered spam."
 189   :type 'boolean
 190   :group 'spam)
 191
 192 (defcustom spam-use-blackholes nil
 193   "Whether blackholes should be used by `spam-split'."
 194   :type 'boolean
 195   :group 'spam)
 196
 197 (defcustom spam-use-hashcash nil
 198   "Whether hashcash payments should be detected by `spam-split'."
 199   :type 'boolean
 200   :group 'spam)
 201
 202 (defcustom spam-use-regex-headers nil
 203   "Whether a header regular expression match should be used by `spam-split'.
 204 Also see the variables `spam-regex-headers-spam' and `spam-regex-headers-ham'."
 205   :type 'boolean
 206   :group 'spam)
 207
 208 (defcustom spam-use-regex-body nil
 209   "Whether a body regular expression match should be used by `spam-split'.
 210 Also see the variables `spam-regex-body-spam' and `spam-regex-body-ham'."
 211   :type 'boolean
 212   :group 'spam)
 213
 214 (defcustom spam-use-bogofilter-headers nil
 215   "Whether bogofilter headers should be used by `spam-split'.
 216 Enable this if you pre-process messages with Bogofilter BEFORE Gnus sees them."
 217   :type 'boolean
 218   :group 'spam)
 219
 220 (defcustom spam-use-bogofilter nil
 221   "Whether bogofilter should be invoked by `spam-split'.
 222 Enable this if you want Gnus to invoke Bogofilter on new messages."
 223   :type 'boolean
 224   :group 'spam)
 225
 226 (defcustom spam-use-BBDB nil
 227   "Whether BBDB should be used by `spam-split'."
 228   :type 'boolean
 229   :group 'spam)
 230
 231 (defcustom spam-use-BBDB-exclusive nil
 232   "Whether BBDB-exclusive should be used by `spam-split'.
 233 Exclusive BBDB means that all messages from senders not in the BBDB are
 234 considered spam."
 235   :type 'boolean
 236   :group 'spam)
 237
 238 (defcustom spam-use-ifile nil
 239   "Whether ifile should be used by `spam-split'."
 240   :type 'boolean
 241   :group 'spam)
 242
 243 (defcustom spam-use-stat nil
 244   "Whether `spam-stat' should be used by `spam-split'."
 245   :type 'boolean
 246   :group 'spam)
 247
 248 (defcustom spam-use-spamoracle nil
 249   "Whether spamoracle should be used by `spam-split'."
 250   :type 'boolean
 251   :group 'spam)
 252
 253 (defcustom spam-use-spamassassin nil
 254   "Whether spamassassin should be invoked by `spam-split'.
 255 Enable this if you want Gnus to invoke SpamAssassin on new messages."
 256   :type 'boolean
 257   :group 'spam)
 258
 259 (defcustom spam-use-spamassassin-headers nil
 260   "Whether spamassassin headers should be checked by `spam-split'.
 261 Enable this if you pre-process messages with SpamAssassin BEFORE Gnus sees
 262 them."
 263   :type 'boolean
 264   :group 'spam)
 265
 266 (defcustom spam-install-hooks (or
 267                                spam-use-dig
 268                                spam-use-gmane-xref
 269                                spam-use-blacklist
 270                                spam-use-whitelist
 271                                spam-use-whitelist-exclusive
 272                                spam-use-blackholes
 273                                spam-use-hashcash
 274                                spam-use-regex-headers
 275                                spam-use-regex-body
 276                                spam-use-bogofilter
 277                                spam-use-bogofilter-headers
 278                                spam-use-spamassassin
 279                                spam-use-spamassassin-headers
 280                                spam-use-BBDB
 281                                spam-use-BBDB-exclusive
 282                                spam-use-ifile
 283                                spam-use-stat
 284                                spam-use-spamoracle)
 285   "Whether the spam hooks should be installed.
 286 Default to t if one of the spam-use-* variables is set."
 287   :group 'spam
 288   :type 'boolean)
 289
 290 (defcustom spam-split-group "spam"
 291   "Group name where incoming spam should be put by `spam-split'."
 292   :type 'string
 293   :group 'spam)
 294
 295 ;;; TODO: deprecate this variable, it's confusing since it's a list of strings,
 296 ;;; not regular expressions
 297 (defcustom spam-junk-mailgroups (cons
 298                                  spam-split-group
 299                                  '("mail.junk" "poste.pourriel"))
 300   "Mailgroups with spam contents.
 301 All unmarked article in such group receive the spam mark on group entry."
 302   :type '(repeat (string :tag "Group"))
 303   :group 'spam)
 304
 305
 306 (defcustom spam-gmane-xref-spam-group "gmane.spam.detected"
 307   "The group where spam xrefs can be found on Gmane.
 308 Only meaningful if you enable `spam-use-gmane-xref'."
 309   :type 'string
 310   :group 'spam)
 311
 312 (defcustom spam-blackhole-servers '("bl.spamcop.net" "relays.ordb.org"
 313                                     "dev.null.dk" "relays.visi.com")
 314   "List of blackhole servers.
 315 Only meaningful if you enable `spam-use-blackholes'."
 316   :type '(repeat (string :tag "Server"))
 317   :group 'spam)
 318
 319 (defcustom spam-blackhole-good-server-regex nil
 320   "String matching IP addresses that should not be checked in the blackholes.
 321 Only meaningful if you enable `spam-use-blackholes'."
 322   :type '(radio (const nil)
 323                 (regexp :format "%t: %v\n" :size 0))
 324   :group 'spam)
 325
 326 (defcustom spam-face 'gnus-splash-face
 327   "Face for spam-marked articles."
 328   :type 'face
 329   :group 'spam)
 330
 331 (defcustom spam-regex-headers-spam '("^X-Spam-Flag: YES")
 332   "Regular expression for positive header spam matches.
 333 Only meaningful if you enable `spam-use-regex-headers'."
 334   :type '(repeat (regexp :tag "Regular expression to match spam header"))
 335   :group 'spam)
 336
 337 (defcustom spam-regex-headers-ham '("^X-Spam-Flag: NO")
 338   "Regular expression for positive header ham matches.
 339 Only meaningful if you enable `spam-use-regex-headers'."
 340   :type '(repeat (regexp :tag "Regular expression to match ham header"))
 341   :group 'spam)
 342
 343 (defcustom spam-regex-body-spam '()
 344   "Regular expression for positive body spam matches.
 345 Only meaningful if you enable `spam-use-regex-body'."
 346   :type '(repeat (regexp :tag "Regular expression to match spam body"))
 347   :group 'spam)
 348
 349 (defcustom spam-regex-body-ham '()
 350   "Regular expression for positive body ham matches.
 351 Only meaningful if you enable `spam-use-regex-body'."
 352   :type '(repeat (regexp :tag "Regular expression to match ham body"))
 353   :group 'spam)
 354
 355 (defgroup spam-ifile nil
 356   "Spam ifile configuration."
 357   :group 'spam)
 358
 359 (defcustom spam-ifile-path (executable-find "ifile")
 360   "File path of the ifile executable program."
 361   :type '(choice (file :tag "Location of ifile")
 362                  (const :tag "ifile is not installed"))
 363   :group 'spam-ifile)
 364
 365 (defcustom spam-ifile-database-path nil
 366   "File path of the ifile database."
 367   :type '(choice (file :tag "Location of the ifile database")
 368                  (const :tag "Use the default"))
 369   :group 'spam-ifile)
 370
 371 (defcustom spam-ifile-spam-category "spam"
 372   "Name of the spam ifile category."
 373   :type 'string
 374   :group 'spam-ifile)
 375
 376 (defcustom spam-ifile-ham-category nil
 377   "Name of the ham ifile category.
 378 If nil, the current group name will be used."
 379   :type '(choice (string :tag "Use a fixed category")
 380                  (const :tag "Use the current group name"))
 381   :group 'spam-ifile)
 382
 383 (defcustom spam-ifile-all-categories nil
 384   "Whether the ifile check will return all categories, or just spam.
 385 Set this to t if you want to use the `spam-split' invocation of ifile as
 386 your main source of newsgroup names."
 387   :type 'boolean
 388   :group 'spam-ifile)
 389
 390 (defgroup spam-bogofilter nil
 391   "Spam bogofilter configuration."
 392   :group 'spam)
 393
 394 (defcustom spam-bogofilter-path (executable-find "bogofilter")
 395   "File path of the Bogofilter executable program."
 396   :type '(choice (file :tag "Location of bogofilter")
 397                  (const :tag "Bogofilter is not installed"))
 398   :group 'spam-bogofilter)
 399
 400 (defcustom spam-bogofilter-header "X-Bogosity"
 401   "The header that Bogofilter inserts in messages."
 402   :type 'string
 403   :group 'spam-bogofilter)
 404
 405 (defcustom spam-bogofilter-spam-switch "-s"
 406   "The switch that Bogofilter uses to register spam messages."
 407   :type 'string
 408   :group 'spam-bogofilter)
 409
 410 (defcustom spam-bogofilter-ham-switch "-n"
 411   "The switch that Bogofilter uses to register ham messages."
 412   :type 'string
 413   :group 'spam-bogofilter)
 414
 415 (defcustom spam-bogofilter-spam-strong-switch "-S"
 416   "The switch that Bogofilter uses to unregister ham messages."
 417   :type 'string
 418   :group 'spam-bogofilter)
 419
 420 (defcustom spam-bogofilter-ham-strong-switch "-N"
 421   "The switch that Bogofilter uses to unregister spam messages."
 422   :type 'string
 423   :group 'spam-bogofilter)
 424
 425 (defcustom spam-bogofilter-bogosity-positive-spam-header "^\\(Yes\\|Spam\\)"
 426   "The regex on `spam-bogofilter-header' for positive spam identification."
 427   :type 'regexp
 428   :group 'spam-bogofilter)
 429
 430 (defcustom spam-bogofilter-database-directory nil
 431   "Directory path of the Bogofilter databases."
 432   :type '(choice (directory
 433                   :tag "Location of the Bogofilter database directory")
 434                  (const :tag "Use the default"))
 435   :group 'spam-bogofilter)
 436
 437 (defgroup spam-spamoracle nil
 438   "Spam spamoracle configuration."
 439   :group 'spam)
 440
 441 (defcustom spam-spamoracle-database nil
 442   "Location of spamoracle database file.
 443 When nil, use the default spamoracle database."
 444   :type '(choice (directory :tag "Location of spamoracle database file.")
 445                  (const :tag "Use the default"))
 446   :group 'spam-spamoracle)
 447
 448 (defcustom spam-spamoracle-binary (executable-find "spamoracle")
 449   "Location of the spamoracle binary."
 450   :type '(choice (directory :tag "Location of the spamoracle binary")
 451                  (const :tag "Use the default"))
 452   :group 'spam-spamoracle)
 453
 454 (defgroup spam-spamassassin nil
 455   "Spam SpamAssassin configuration."
 456   :group 'spam)
 457
 458 (defcustom spam-spamassassin-path (executable-find "spamassassin")
 459   "File path of the spamassassin executable program.
 460 Hint: set this to \"spamc\" if you have spamd running.  See the spamc and
 461 spamd man pages for more information on these programs."
 462   :type '(choice (file :tag "Location of spamc")
 463                  (const :tag "spamassassin is not installed"))
 464   :group 'spam-spamassassin)
 465
 466 (defcustom spam-spamassassin-arguments ()
 467   "Arguments to pass to the spamassassin executable.
 468 This must be a list.  For example, `(\"-C\" \"configfile\")'."
 469   :type '(restricted-sexp :match-alternatives (listp))
 470   :group 'spam-spamassassin)
 471
 472 (defcustom spam-spamassassin-spam-flag-header "X-Spam-Flag"
 473   "The header inserted by SpamAssassin to flag spam."
 474   :type 'string
 475   :group 'spam-spamassassin)
 476
 477 (defcustom spam-spamassassin-positive-spam-flag-header "YES"
 478   "The regex on `spam-spamassassin-spam-flag-header' for positive spam
 479 identification"
 480   :type 'string
 481   :group 'spam-spamassassin)
 482
 483 (defcustom spam-spamassassin-spam-status-header "X-Spam-Status"
 484   "The header inserted by SpamAssassin, giving extended scoring information"
 485   :type 'string
 486   :group 'spam-spamassassin)
 487
 488 (defcustom spam-sa-learn-path (executable-find "sa-learn")
 489   "File path of the sa-learn executable program."
 490   :type '(choice (file :tag "Location of spamassassin")
 491                  (const :tag "spamassassin is not installed"))
 492   :group 'spam-spamassassin)
 493
 494 (defcustom spam-sa-learn-rebuild t
 495   "Whether sa-learn should rebuild the database every time it is called
 496 Enable this if you want sa-learn to rebuild the database automatically.  Doing
 497 this will slightly increase the running time of the spam registration process.
 498 If you choose not to do this, you will have to run \"sa-learn --rebuild\" in
 499 order for SpamAssassin to recognize the new registered spam."
 500   :type 'boolean
 501   :group 'spam-spamassassin)
 502
 503 (defcustom spam-sa-learn-spam-switch "--spam"
 504   "The switch that sa-learn uses to register spam messages"
 505   :type 'string
 506   :group 'spam-spamassassin)
 507
 508 (defcustom spam-sa-learn-ham-switch "--ham"
 509   "The switch that sa-learn uses to register ham messages"
 510   :type 'string
 511   :group 'spam-spamassassin)
 512
 513 (defcustom spam-sa-learn-unregister-switch "--forget"
 514   "The switch that sa-learn uses to unregister messages messages"
 515   :type 'string
 516   :group 'spam-spamassassin)
 517
 518 ;;; Key bindings for spam control.
 519
 520 (gnus-define-keys gnus-summary-mode-map
 521   "St" spam-generic-score
 522   "Sx" gnus-summary-mark-as-spam
 523   "Mst" spam-generic-score
 524   "Msx" gnus-summary-mark-as-spam
 525   "\M-d" gnus-summary-mark-as-spam)
 526
 527 (defvar spam-cache-lookups t
 528   "Whether spam.el will try to cache lookups using `spam-caches'.")
 529
 530 (defvar spam-caches (make-hash-table
 531                      :size 10
 532                      :test 'equal)
 533   "Cache of spam detection entries.")
 534
 535 (defvar spam-old-ham-articles nil
 536   "List of old ham articles, generated when a group is entered.")
 537
 538 (defvar spam-old-spam-articles nil
 539   "List of old spam articles, generated when a group is entered.")
 540
 541 (defvar spam-split-disabled nil
 542   "If non-nil, `spam-split' is disabled, and always returns nil.")
 543
 544 (defvar spam-split-last-successful-check nil
 545   "Internal variable.
 546 `spam-split' will set this to nil or a spam-use-XYZ check if it
 547 finds ham or spam.")
 548
 549 ;; convenience functions
 550 (defun spam-clear-cache (symbol)
 551   "Clear the spam-caches entry for a check."
 552   (remhash symbol spam-caches))
 553
 554 (defun spam-xor (a b)
 555   "Logical A xor B."
 556   (and (or a b) (not (and a b))))
 557
 558 (defun spam-group-ham-mark-p (group mark &optional spam)
 559   "Checks if MARK is considered a ham mark in GROUP."
 560   (when (stringp group)
 561     (let* ((marks (spam-group-ham-marks group spam))
 562            (marks (if (symbolp mark)
 563                       marks
 564                     (mapcar 'symbol-value marks))))
 565       (memq mark marks))))
 566
 567 (defun spam-group-spam-mark-p (group mark)
 568   "Checks if MARK is considered a spam mark in GROUP."
 569   (spam-group-ham-mark-p group mark t))
 570
 571 (defun spam-group-ham-marks (group &optional spam)
 572   "In GROUP, get all the ham marks."
 573   (when (stringp group)
 574     (let* ((marks (if spam
 575                       (gnus-parameter-spam-marks group)
 576                     (gnus-parameter-ham-marks group)))
 577            (marks (car marks))
 578            (marks (if (listp (car marks)) (car marks) marks)))
 579       marks)))
 580
 581 (defun spam-group-spam-marks (group)
 582   "In GROUP, get all the spam marks."
 583   (spam-group-ham-marks group t))
 584
 585 (defun spam-group-spam-contents-p (group)
 586   "Is GROUP a spam group?"
 587   (if (stringp group)
 588       (or (member group spam-junk-mailgroups)
 589           (memq 'gnus-group-spam-classification-spam
 590                 (gnus-parameter-spam-contents group)))
 591     nil))
 592
 593 (defun spam-group-ham-contents-p (group)
 594   "Is GROUP a ham group?"
 595   (if (stringp group)
 596       (memq 'gnus-group-spam-classification-ham
 597             (gnus-parameter-spam-contents group))
 598     nil))
 599
 600 (defvar spam-list-of-processors
 601   '((gnus-group-spam-exit-processor-report-gmane spam spam-use-gmane)
 602     (gnus-group-spam-exit-processor-bogofilter   spam spam-use-bogofilter)
 603     (gnus-group-spam-exit-processor-blacklist    spam spam-use-blacklist)
 604     (gnus-group-spam-exit-processor-ifile        spam spam-use-ifile)
 605     (gnus-group-spam-exit-processor-stat         spam spam-use-stat)
 606     (gnus-group-spam-exit-processor-spamoracle   spam spam-use-spamoracle)
 607     (gnus-group-spam-exit-processor-spamassassin spam spam-use-spamassassin)
 608     (gnus-group-ham-exit-processor-ifile         ham spam-use-ifile)
 609     (gnus-group-ham-exit-processor-bogofilter    ham spam-use-bogofilter)
 610     (gnus-group-ham-exit-processor-stat          ham spam-use-stat)
 611     (gnus-group-ham-exit-processor-whitelist     ham spam-use-whitelist)
 612     (gnus-group-ham-exit-processor-BBDB          ham spam-use-BBDB)
 613     (gnus-group-ham-exit-processor-copy          ham spam-use-ham-copy)
 614     (gnus-group-ham-exit-processor-spamassassin  ham spam-use-spamassassin)
 615     (gnus-group-ham-exit-processor-spamoracle    ham spam-use-spamoracle))
 616   "The `spam-list-of-processors' list.
 617 This list contains pairs associating a ham/spam exit processor
 618 variable with a classification and a spam-use-* variable.")
 619
 620 (defun spam-group-processor-p (group processor)
 621   (if (and (stringp group)
 622            (symbolp processor))
 623       (or (member processor (nth 0 (gnus-parameter-spam-process group)))
 624           (spam-group-processor-multiple-p
 625            group
 626            (cdr-safe (assoc processor spam-list-of-processors))))
 627     nil))
 628
 629 (defun spam-group-processor-multiple-p (group processor-info)
 630   (let* ((classification (nth 0 processor-info))
 631          (check (nth 1 processor-info))
 632          (parameters (nth 0 (gnus-parameter-spam-process group)))
 633          found)
 634     (dolist (parameter parameters)
 635       (when (and (null found)
 636                  (listp parameter)
 637                  (eq classification (nth 0 parameter))
 638                  (eq check (nth 1 parameter)))
 639         (setq found t)))
 640     found))
 641
 642 (defun spam-group-spam-processor-report-gmane-p (group)
 643   (spam-group-processor-p group 'gnus-group-spam-exit-processor-report-gmane))
 644
 645 (defun spam-group-spam-processor-bogofilter-p (group)
 646   (spam-group-processor-p group 'gnus-group-spam-exit-processor-bogofilter))
 647
 648 (defun spam-group-spam-processor-blacklist-p (group)
 649   (spam-group-processor-p group 'gnus-group-spam-exit-processor-blacklist))
 650
 651 (defun spam-group-spam-processor-ifile-p (group)
 652   (spam-group-processor-p group 'gnus-group-spam-exit-processor-ifile))
 653
 654 (defun spam-group-ham-processor-ifile-p (group)
 655   (spam-group-processor-p group 'gnus-group-ham-exit-processor-ifile))
 656
 657 (defun spam-group-spam-processor-spamoracle-p (group)
 658   (spam-group-processor-p group 'gnus-group-spam-exit-processor-spamoracle))
 659
 660 (defun spam-group-ham-processor-bogofilter-p (group)
 661   (spam-group-processor-p group 'gnus-group-ham-exit-processor-bogofilter))
 662
 663 (defun spam-group-spam-processor-stat-p (group)
 664   (spam-group-processor-p group 'gnus-group-spam-exit-processor-stat))
 665
 666 (defun spam-group-ham-processor-stat-p (group)
 667   (spam-group-processor-p group 'gnus-group-ham-exit-processor-stat))
 668
 669 (defun spam-group-ham-processor-whitelist-p (group)
 670   (spam-group-processor-p group 'gnus-group-ham-exit-processor-whitelist))
 671
 672 (defun spam-group-ham-processor-BBDB-p (group)
 673   (spam-group-processor-p group 'gnus-group-ham-exit-processor-BBDB))
 674
 675 (defun spam-group-ham-processor-copy-p (group)
 676   (spam-group-processor-p group 'gnus-group-ham-exit-processor-copy))
 677
 678 (defun spam-group-ham-processor-spamoracle-p (group)
 679   (spam-group-processor-p group 'gnus-group-ham-exit-processor-spamoracle))
 680
 681 (defun spam-report-articles-gmane (n)
 682   "Report the current message as spam.
 683 Respects the process/prefix convention."
 684   (interactive "P")
 685   (dolist (article (gnus-summary-work-articles n))
 686     (gnus-summary-remove-process-mark article)
 687     (spam-report-gmane article)))
 688
 689 (defun spam-necessary-extra-headers ()
 690   "Return the extra headers spam.el thinks are necessary."
 691   (let (list)
 692     (when (or spam-use-spamassassin
 693               spam-use-spamassassin-headers
 694               spam-use-regex-headers)
 695       (push 'X-Spam-Status list))
 696     list))
 697
 698 (defun spam-user-format-function-S (headers)
 699   (when headers
 700     (spam-summary-score headers)))
 701
 702 (defun spam-article-sort-by-spam-status (h1 h2)
 703   "Sort articles by score."
 704   (let (result)
 705     (dolist (header (spam-necessary-extra-headers))
 706       (let ((s1 (spam-summary-score h1 header))
 707             (s2 (spam-summary-score h2 header)))
 708       (unless (= s1 s2)
 709         (setq result (< s1 s2))
 710         (return))))
 711     result))
 712
 713 (defun spam-extra-header-to-number (header headers)
 714   "Transform an extra header to a number."
 715   (if (gnus-extra-header header headers)
 716       (cond
 717        ((eq header 'X-Spam-Status)
 718         (string-to-number (gnus-replace-in-string
 719                            (gnus-extra-header header headers)
 720                            ".*hits=" "")))
 721        (t nil))
 722     nil))
 723
 724 (defun spam-summary-score (headers &optional specific-header)
 725   "Score an article for the summary buffer, as fast as possible.
 726 With SPECIFIC-HEADER, returns only that header's score.
 727 Will not return a nil score."
 728   (let (score)
 729     (dolist (header
 730              (if specific-header
 731                  (list specific-header)
 732                (spam-necessary-extra-headers)))
 733       (setq score
 734             (spam-extra-header-to-number header headers))
 735       (when score
 736         (return)))
 737     (or score 0)))
 738
 739 (defun spam-generic-score ()
 740   "Invoke whatever scoring method we can."
 741   (interactive)
 742   (if (or
 743        spam-use-spamassassin
 744        spam-use-spamassassin-headers)
 745       (spam-spamassassin-score)
 746     (spam-bogofilter-score)))
 747
 748 ;;; Summary entry and exit processing.
 749
 750 (defun spam-summary-prepare ()
 751   (setq spam-old-ham-articles
 752         (spam-list-articles gnus-newsgroup-articles 'ham))
 753   (setq spam-old-spam-articles
 754         (spam-list-articles gnus-newsgroup-articles 'spam))
 755   (spam-mark-junk-as-spam-routine))
 756
 757 ;; The spam processors are invoked for any group, spam or ham or neither
 758 (defun spam-summary-prepare-exit ()
 759   (unless gnus-group-is-exiting-without-update-p
 760     (gnus-message 6 "Exiting summary buffer and applying spam rules")
 761
 762     ;; first of all, unregister any articles that are no longer ham or spam
 763     ;; we have to iterate over the processors, or else we'll be too slow
 764     (dolist (classification '(spam ham))
 765       (let* ((old-articles (if (eq classification 'spam)
 766                                spam-old-spam-articles
 767                              spam-old-ham-articles))
 768              (new-articles (spam-list-articles
 769                             gnus-newsgroup-articles
 770                             classification))
 771              (changed-articles (spam-set-difference new-articles old-articles)))
 772         ;; now that we have the changed articles, we go through the processors
 773         (dolist (processor-param spam-list-of-processors)
 774           (let ((processor (nth 0 processor-param))
 775                 (processor-classification (nth 1 processor-param))
 776                 (check (nth 2 processor-param))
 777                 unregister-list)
 778             (dolist (article changed-articles)
 779               (let ((id (spam-fetch-field-message-id-fast article)))
 780                 (when (spam-log-unregistration-needed-p
 781                        id 'process classification check)
 782                   (push article unregister-list))))
 783             ;; call spam-register-routine with specific articles to unregister,
 784             ;; when there are articles to unregister and the check is enabled
 785             (when (and unregister-list (symbol-value check))
 786               (spam-register-routine
 787                classification check t unregister-list))))))
 788
 789     ;; find all the spam processors applicable to this group
 790     (dolist (processor-param spam-list-of-processors)
 791       (let ((processor (nth 0 processor-param))
 792             (classification (nth 1 processor-param))
 793             (check (nth 2 processor-param)))
 794         (when (and (eq 'spam classification)
 795                    (spam-group-processor-p gnus-newsgroup-name processor))
 796           (spam-register-routine classification check))))
 797
 798     (unless (and spam-move-spam-nonspam-groups-only
 799                  (spam-group-spam-contents-p gnus-newsgroup-name))
 800       (gnus-message 6 "Marking spam as expired and moving it to %s"
 801                     (gnus-parameter-spam-process-destination
 802                      gnus-newsgroup-name))
 803       (spam-mark-spam-as-expired-and-move-routine
 804        (gnus-parameter-spam-process-destination gnus-newsgroup-name)))
 805
 806     ;; now we redo spam-mark-spam-as-expired-and-move-routine to only
 807     ;; expire spam, in case the above did not expire them
 808     (gnus-message 6 "Marking spam as expired without moving it")
 809     (spam-mark-spam-as-expired-and-move-routine nil)
 810
 811     (when (or (spam-group-ham-contents-p gnus-newsgroup-name)
 812               (and (spam-group-spam-contents-p gnus-newsgroup-name)
 813                    spam-process-ham-in-spam-groups)
 814               spam-process-ham-in-nonham-groups)
 815       ;; find all the ham processors applicable to this group
 816       (dolist (processor-param spam-list-of-processors)
 817         (let ((processor (nth 0 processor-param))
 818               (classification (nth 1 processor-param))
 819               (check (nth 2 processor-param)))
 820           (when (and (eq 'ham classification)
 821                      (spam-group-processor-p gnus-newsgroup-name processor))
 822             (spam-register-routine classification check)))))
 823
 824     (when (spam-group-ham-processor-copy-p gnus-newsgroup-name)
 825       (gnus-message 6 "Copying ham")
 826       (spam-ham-copy-routine
 827        (gnus-parameter-ham-process-destination gnus-newsgroup-name)))
 828
 829     ;; now move all ham articles out of spam groups
 830     (when (spam-group-spam-contents-p gnus-newsgroup-name)
 831       (gnus-message 6 "Moving ham messages from spam group")
 832       (spam-ham-move-routine
 833        (gnus-parameter-ham-process-destination gnus-newsgroup-name))))
 834
 835   (setq spam-old-ham-articles nil)
 836   (setq spam-old-spam-articles nil))
 837
 838 (defun spam-set-difference (list1 list2)
 839   "Return a set difference of LIST1 and LIST2.
 840 When either list is nil, the other is returned."
 841   (if (and list1 list2)
 842       ;; we have two non-nil lists
 843       (progn
 844         (dolist (item (append list1 list2))
 845           (when (and (memq item list1) (memq item list2))
 846             (setq list1 (delq item list1))
 847             (setq list2 (delq item list2))))
 848         (append list1 list2))
 849     ;; if either of the lists was nil, return the other one
 850     (if list1 list1 list2)))
 851
 852 (defun spam-mark-junk-as-spam-routine ()
 853   ;; check the global list of group names spam-junk-mailgroups and the
 854   ;; group parameters
 855   (when (spam-group-spam-contents-p gnus-newsgroup-name)
 856     (gnus-message 6 "Marking %s articles as spam"
 857                   (if spam-mark-only-unseen-as-spam
 858                       "unseen"
 859                     "unread"))
 860     (let ((articles (if spam-mark-only-unseen-as-spam
 861                         gnus-newsgroup-unseen
 862                       gnus-newsgroup-unreads)))
 863       (dolist (article articles)
 864         (gnus-summary-mark-article article gnus-spam-mark)))))
 865
 866 (defun spam-mark-spam-as-expired-and-move-routine (&rest groups)
 867   (if (and (car-safe groups) (listp (car-safe groups)))
 868       (apply 'spam-mark-spam-as-expired-and-move-routine (car groups))
 869     (gnus-summary-kill-process-mark)
 870     (let ((articles gnus-newsgroup-articles)
 871           (backend-supports-deletions
 872            (gnus-check-backend-function
 873             'request-move-article gnus-newsgroup-name))
 874           article tomove deletep)
 875       (dolist (article articles)
 876         (when (eq (gnus-summary-article-mark article) gnus-spam-mark)
 877           (gnus-summary-mark-article article gnus-expirable-mark)
 878           (push article tomove)))
 879
 880       ;; now do the actual copies
 881       (dolist (group groups)
 882         (when (and tomove
 883                    (stringp group))
 884           (dolist (article tomove)
 885             (gnus-summary-set-process-mark article))
 886           (when tomove
 887             (if (or (not backend-supports-deletions)
 888                     (> (length groups) 1))
 889                 (progn
 890                   (gnus-summary-copy-article nil group)
 891                   (setq deletep t))
 892               (gnus-summary-move-article nil group)))))
 893
 894       ;; now delete the articles, if there was a copy done, and the
 895       ;; backend allows it
 896       (when (and deletep backend-supports-deletions)
 897         (dolist (article tomove)
 898           (gnus-summary-set-process-mark article))
 899         (when tomove
 900           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
 901             (gnus-summary-delete-article nil))))
 902
 903       (gnus-summary-yank-process-mark))))
 904
 905 (defun spam-ham-copy-or-move-routine (copy groups)
 906   (gnus-summary-kill-process-mark)
 907   (let ((todo (spam-list-articles gnus-newsgroup-articles 'ham))
 908         (backend-supports-deletions
 909          (gnus-check-backend-function
 910           'request-move-article gnus-newsgroup-name))
 911         (respool-method (gnus-find-method-for-group gnus-newsgroup-name))
 912         article mark todo deletep respool)
 913
 914     (when (member 'respool groups)
 915       (setq respool t)                  ; boolean for later
 916       (setq groups '("fake"))) ; when respooling, groups are dynamic so fake it
 917
 918     ;; now do the actual move
 919     (dolist (group groups)
 920       (when (and todo (stringp group))
 921         (dolist (article todo)
 922           (when spam-mark-ham-unread-before-move-from-spam-group
 923             (gnus-summary-mark-article article gnus-unread-mark))
 924           (gnus-summary-set-process-mark article))
 925
 926         (if respool                        ; respooling is with a "fake" group
 927             (let ((spam-split-disabled
 928                    (or spam-split-disabled
 929                        spam-disable-spam-split-during-ham-respool)))
 930               (gnus-summary-respool-article nil respool-method))
 931           (if (or (not backend-supports-deletions) ; else, we are not respooling
 932                   (> (length groups) 1))
 933               (progn                ; if copying, copy and set deletep
 934                 (gnus-summary-copy-article nil group)
 935                 (setq deletep t))
 936             (gnus-summary-move-article nil group))))) ; else move articles
 937
 938     ;; now delete the articles, unless a) copy is t, and there was a copy done
 939     ;;                                 b) a move was done to a single group
 940     ;;                                 c) backend-supports-deletions is nil
 941     (unless copy
 942       (when (and deletep backend-supports-deletions)
 943         (dolist (article todo)
 944           (gnus-summary-set-process-mark article))
 945         (when todo
 946           (let ((gnus-novice-user nil)) ; don't ask me if I'm sure
 947             (gnus-summary-delete-article nil))))))
 948
 949   (gnus-summary-yank-process-mark))
 950
 951 (defun spam-ham-copy-routine (&rest groups)
 952   (if (and (car-safe groups) (listp (car-safe groups)))
 953       (apply 'spam-ham-copy-routine (car groups))
 954     (spam-ham-copy-or-move-routine t groups)))
 955
 956 (defun spam-ham-move-routine (&rest groups)
 957   (if (and (car-safe groups) (listp (car-safe groups)))
 958       (apply 'spam-ham-move-routine (car groups))
 959     (spam-ham-copy-or-move-routine nil groups)))
 960
 961 (defun spam-get-article-as-string (article)
 962   (when (numberp article)
 963     (with-temp-buffer
 964       (gnus-request-article-this-buffer
 965        article
 966        gnus-newsgroup-name)
 967       (buffer-string))))
 968
 969 ;; disabled for now
 970 ;; (defun spam-get-article-as-filename (article)
 971 ;;   (let ((article-filename))
 972 ;;     (when (numberp article)
 973 ;;       (nnml-possibly-change-directory
 974 ;;        (gnus-group-real-name gnus-newsgroup-name))
 975 ;;       (setq article-filename (expand-file-name
 976 ;;                              (int-to-string article) nnml-current-directory)))
 977 ;;     (if (file-exists-p article-filename)
 978 ;;      article-filename
 979 ;;       nil)))
 980
 981 (defun spam-fetch-field-fast (article field &optional prepared-data-header)
 982   "Fetch a field quickly, using the internal gnus-data-list function"
 983   (when (numberp article)
 984     (let* ((data-header (or prepared-data-header
 985                             (spam-fetch-article-header article))))
 986       (if (arrayp data-header)
 987         (cond
 988          ((equal field 'from)
 989           (mail-header-from data-header))
 990          ((equal field 'message-id)
 991           (mail-header-message-id data-header))
 992          ((equal field 'subject)
 993           (mail-header-subject data-header))
 994          ((equal field 'references)
 995           (mail-header-references data-header))
 996          ((equal field 'date)
 997           (mail-header-date data-header))
 998          ((equal field 'xref)
 999           (mail-header-xref data-header))
1000          ((equal field 'extra)
1001           (mail-header-extra data-header))
1002          (t
1003           nil))
1004         (gnus-message 6 "Article %d has a nil data header" article)))))
1005
1006 (defun spam-fetch-field-from-fast (article &optional prepared-data-header)
1007   (spam-fetch-field-fast article 'from prepared-data-header))
1008
1009 (defun spam-fetch-field-subject-fast (article &optional prepared-data-header)
1010   (spam-fetch-field-fast article 'subject prepared-data-header))
1011
1012 (defun spam-fetch-field-message-id-fast (article &optional prepared-data-header)
1013   (spam-fetch-field-fast article 'message-id prepared-data-header))
1014
1015 (defun spam-generate-fake-headers (article)
1016   (let ((dh (spam-fetch-article-header article)))
1017     (if dh
1018         (concat
1019          (format
1020           (concat "From: %s\nSubject: %s\nMessage-ID: %s\n"
1021                   "Date: %s\nReferences: %s\nXref: %s\n")
1022           (spam-fetch-field-fast article 'from dh)
1023           (spam-fetch-field-fast article 'subject dh)
1024           (spam-fetch-field-fast article 'message-id dh)
1025           (spam-fetch-field-fast article 'date dh)
1026           (spam-fetch-field-fast article 'references dh)
1027           (spam-fetch-field-fast article 'xref dh))
1028          (when (spam-fetch-field-fast article 'extra dh)
1029            (format "%s\n" (spam-fetch-field-fast article 'extra dh))))
1030       (gnus-message
1031        5
1032        "spam-generate-fake-headers: article %d didn't have a valid header"
1033        article))))
1034
1035 (defun spam-fetch-article-header (article)
1036   (save-excursion
1037     (set-buffer gnus-summary-buffer)
1038     (gnus-read-header article)
1039     (nth 3 (assq article gnus-newsgroup-data))))
1040
1041 \f
1042 ;;;; Spam determination.
1043
1044 (defvar spam-list-of-checks
1045   '((spam-use-blacklist                 .       spam-check-blacklist)
1046     (spam-use-regex-headers             .       spam-check-regex-headers)
1047     (spam-use-gmane-xref                .       spam-check-gmane-xref)
1048     (spam-use-regex-body                .       spam-check-regex-body)
1049     (spam-use-whitelist                 .       spam-check-whitelist)
1050     (spam-use-BBDB                      .       spam-check-BBDB)
1051     (spam-use-BBDB-exclusive            .       spam-check-BBDB)
1052     (spam-use-ifile                     .       spam-check-ifile)
1053     (spam-use-spamoracle                .       spam-check-spamoracle)
1054     (spam-use-stat                      .       spam-check-stat)
1055     (spam-use-blackholes                .       spam-check-blackholes)
1056     (spam-use-hashcash                  .       spam-check-hashcash)
1057     (spam-use-spamassassin-headers      .       spam-check-spamassassin-headers)
1058     (spam-use-spamassassin              .       spam-check-spamassassin)
1059     (spam-use-bogofilter-headers        .       spam-check-bogofilter-headers)
1060     (spam-use-bogofilter                .       spam-check-bogofilter))
1061   "The spam-list-of-checks list contains pairs associating a
1062 parameter variable with a spam checking function.  If the
1063 parameter variable is true, then the checking function is called,
1064 and its value decides what happens.  Each individual check may
1065 return nil, t, or a mailgroup name.  The value nil means that the
1066 check does not yield a decision, and so, that further checks are
1067 needed.  The value t means that the message is definitely not
1068 spam, and that further spam checks should be inhibited.
1069 Otherwise, a mailgroup name or the symbol 'spam (depending on
1070 spam-split-symbolic-return) is returned where the mail should go,
1071 and further checks are also inhibited.  The usual mailgroup name
1072 is the value of `spam-split-group', meaning that the message is
1073 definitely a spam.")
1074
1075 (defvar spam-list-of-statistical-checks
1076   '(spam-use-ifile
1077     spam-use-regex-body
1078     spam-use-stat
1079     spam-use-bogofilter
1080     spam-use-blackholes
1081     spam-use-spamassassin
1082     spam-use-spamoracle)
1083   "The spam-list-of-statistical-checks list contains all the mail
1084 splitters that need to have the full message body available.
1085 Note that you should fetch extra headers if you don't like this,
1086 e.g. fetch the 'Received' header for spam-use-blackholes.")
1087
1088 (defun spam-split (&rest specific-checks)
1089   "Split this message into the `spam' group if it is spam.
1090 This function can be used as an entry in the variable `nnmail-split-fancy',
1091 for example like this: (: spam-split).  It can take checks as
1092 parameters.  A string as a parameter will set the
1093 spam-split-group to that string.
1094
1095 See the Info node `(gnus)Fancy Mail Splitting' for more details."
1096   (interactive)
1097   (setq spam-split-last-successful-check nil)
1098   (unless spam-split-disabled
1099     (let ((spam-split-group-choice spam-split-group))
1100       (dolist (check specific-checks)
1101         (when (stringp check)
1102           (setq spam-split-group-choice check)
1103           (setq specific-checks (delq check specific-checks))))
1104
1105       (let ((spam-split-group spam-split-group-choice))
1106         (save-excursion
1107           (save-restriction
1108             (dolist (check spam-list-of-statistical-checks)
1109               (when (and (symbolp check)
1110                          (or (symbol-value check)
1111                              (memq check specific-checks)))
1112                 (widen)
1113                 (gnus-message 8 "spam-split: widening the buffer (%s requires it)"
1114                               (symbol-name check))
1115                 (return)))
1116             ;;   (progn (widen) (debug (buffer-string)))
1117             (let ((list-of-checks spam-list-of-checks)
1118                   decision)
1119               (while (and list-of-checks (not decision))
1120                 (let ((pair (pop list-of-checks)))
1121                   (when (or
1122                          ;; either, given specific checks, this is one of them
1123                          (and specific-checks (memq (car pair) specific-checks))
1124                          ;; or, given no specific checks, spam-use-CHECK is set
1125                          (and (null specific-checks) (symbol-value (car pair))))
1126                     (gnus-message 6 "spam-split: calling the %s function"
1127                                   (symbol-name (cdr pair)))
1128                     (setq decision (funcall (cdr pair)))
1129                     ;; if we got a decision at all, save the current check
1130                     (when decision
1131                       (setq spam-split-last-successful-check (car pair)))
1132
1133                     (when (eq decision 'spam)
1134                       (unless spam-split-symbolic-return
1135                         (gnus-error
1136                          5
1137                          (format "spam-split got %s but %s is nil"
1138                                  (symbol-name decision)
1139                                  (symbol-name spam-split-symbolic-return))))))))
1140               (if (eq decision t)
1141                   (if spam-split-symbolic-return-positive 'ham nil)
1142                 decision))))))))
1143
1144 (defun spam-find-spam ()
1145   "This function will detect spam in the current newsgroup using spam-split."
1146   (interactive)
1147
1148   (let* ((group gnus-newsgroup-name)
1149          (autodetect (gnus-parameter-spam-autodetect group))
1150          (methods (gnus-parameter-spam-autodetect-methods group))
1151          (first-method (nth 0 methods))
1152          (articles (if spam-autodetect-recheck-messages
1153                        gnus-newsgroup-articles
1154                      gnus-newsgroup-unseen))
1155          article-cannot-be-faked)
1156
1157     (dolist (check spam-list-of-statistical-checks)
1158       (when (and (symbolp check)
1159                  (memq check methods))
1160         (setq article-cannot-be-faked t)
1161         (return)))
1162
1163     (when (memq 'default methods)
1164       (setq article-cannot-be-faked t))
1165
1166     (when (and autodetect
1167                (not (equal first-method 'none)))
1168     (mapcar
1169      (lambda (article)
1170        (let ((id (spam-fetch-field-message-id-fast article))
1171              (subject (spam-fetch-field-subject-fast article))
1172              (sender (spam-fetch-field-from-fast article))
1173              registry-lookup)
1174
1175          (unless id
1176            (gnus-message 6 "Article %d has no message ID!" article))
1177
1178          (when (and id spam-log-to-registry)
1179            (setq registry-lookup (spam-log-registration-type id 'incoming))
1180            (when registry-lookup
1181              (gnus-message
1182               9
1183               "spam-find-spam: message %s was already registered incoming"
1184               id)))
1185
1186          (let* ((spam-split-symbolic-return t)
1187                 (spam-split-symbolic-return-positive t)
1188                 (fake-headers (spam-generate-fake-headers article))
1189                 (split-return
1190                  (or registry-lookup
1191                      (with-temp-buffer
1192                        (if article-cannot-be-faked
1193                            (gnus-request-article-this-buffer
1194                             article
1195                             group)
1196                          ;; else, we fake the article
1197                          (when fake-headers (insert fake-headers)))
1198                        (if (or (null first-method)
1199                                (equal first-method 'default))
1200                            (spam-split)
1201                          (apply 'spam-split methods))))))
1202            (if (equal split-return 'spam)
1203                (gnus-summary-mark-article article gnus-spam-mark))
1204
1205            (when (and id split-return spam-log-to-registry)
1206              (when (zerop (gnus-registry-group-count id))
1207                (gnus-registry-add-group
1208                 id group subject sender))
1209
1210              (unless registry-lookup
1211                (spam-log-processing-to-registry
1212                 id
1213                 'incoming
1214                 split-return
1215                 spam-split-last-successful-check
1216                 group))))))
1217     articles))))
1218
1219 (defvar spam-registration-functions
1220   ;; first the ham register, second the spam register function
1221   ;; third the ham unregister, fourth the spam unregister function
1222   '((spam-use-blacklist  nil
1223                          spam-blacklist-register-routine
1224                          nil
1225                          spam-blacklist-unregister-routine)
1226     (spam-use-whitelist  spam-whitelist-register-routine
1227                          nil
1228                          spam-whitelist-unregister-routine
1229                          nil)
1230     (spam-use-ham-copy   nil
1231                          nil
1232                          nil
1233                          nil)
1234     (spam-use-BBDB       spam-BBDB-register-routine
1235                          nil
1236                          spam-BBDB-unregister-routine
1237                          nil)
1238     (spam-use-ifile      spam-ifile-register-ham-routine
1239                          spam-ifile-register-spam-routine
1240                          spam-ifile-unregister-ham-routine
1241                          spam-ifile-unregister-spam-routine)
1242     (spam-use-spamoracle spam-spamoracle-learn-ham
1243                          spam-spamoracle-learn-spam
1244                          spam-spamoracle-unlearn-ham
1245                          spam-spamoracle-unlearn-spam)
1246     (spam-use-stat       spam-stat-register-ham-routine
1247                          spam-stat-register-spam-routine
1248                          spam-stat-unregister-ham-routine
1249                          spam-stat-unregister-spam-routine)
1250     ;; note that spam-use-gmane is not a legitimate check
1251     (spam-use-gmane      nil
1252                          spam-report-gmane-register-routine
1253                          ;; does Gmane support unregistration?
1254                          nil
1255                          nil)
1256     (spam-use-spamassassin spam-spamassassin-register-ham-routine
1257                            spam-spamassassin-register-spam-routine
1258                            spam-spamassassin-unregister-ham-routine
1259                            spam-spamassassin-unregister-spam-routine)
1260     (spam-use-bogofilter spam-bogofilter-register-ham-routine
1261                          spam-bogofilter-register-spam-routine
1262                          spam-bogofilter-unregister-ham-routine
1263                          spam-bogofilter-unregister-spam-routine))
1264   "The spam-registration-functions list contains pairs
1265 associating a parameter variable with the ham and spam
1266 registration functions, and the ham and spam unregistration
1267 functions")
1268
1269 (defun spam-classification-valid-p (classification)
1270   (or  (eq classification 'spam)
1271        (eq classification 'ham)))
1272
1273 (defun spam-process-type-valid-p (process-type)
1274   (or  (eq process-type 'incoming)
1275        (eq process-type 'process)))
1276
1277 (defun spam-registration-check-valid-p (check)
1278   (assoc check spam-registration-functions))
1279
1280 (defun spam-unregistration-check-valid-p (check)
1281   (assoc check spam-registration-functions))
1282
1283 (defun spam-registration-function (classification check)
1284   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1285     (if (eq classification 'spam)
1286         (nth 1 flist)
1287       (nth 0 flist))))
1288
1289 (defun spam-unregistration-function (classification check)
1290   (let ((flist (cdr-safe (assoc check spam-registration-functions))))
1291     (if (eq classification 'spam)
1292         (nth 3 flist)
1293       (nth 2 flist))))
1294
1295 (defun spam-list-articles (articles classification)
1296   (let ((mark-check (if (eq classification 'spam)
1297                         'spam-group-spam-mark-p
1298                       'spam-group-ham-mark-p))
1299         alist mark-cache-yes mark-cache-no)
1300     (dolist (article articles)
1301       (let ((mark (gnus-summary-article-mark article)))
1302         (unless (or (memq mark mark-cache-yes)
1303                     (memq mark mark-cache-no))
1304           (if (funcall mark-check
1305                        gnus-newsgroup-name
1306                        mark)
1307               (push mark mark-cache-yes)
1308             (push mark mark-cache-no)))
1309         (when (memq mark mark-cache-yes)
1310           (push article alist))))
1311     alist))
1312
1313 (defun spam-register-routine (classification
1314                               check
1315                               &optional unregister
1316                               specific-articles)
1317   (when (and (spam-classification-valid-p classification)
1318              (spam-registration-check-valid-p check))
1319     (let* ((register-function
1320             (spam-registration-function classification check))
1321            (unregister-function
1322             (spam-unregistration-function classification check))
1323            (run-function (if unregister
1324                              unregister-function
1325                            register-function))
1326            (log-function (if unregister
1327                              'spam-log-undo-registration
1328                            'spam-log-processing-to-registry))
1329            article articles)
1330
1331       (when run-function
1332         ;; make list of articles, using specific-articles if given
1333         (setq articles (or specific-articles
1334                            (spam-list-articles
1335                             gnus-newsgroup-articles
1336                             classification)))
1337         ;; process them
1338         (gnus-message 5 "%s %d %s articles as %s using backend %s"
1339                       (if unregister "Unregistering" "Registering")
1340                       (length articles)
1341                       (if specific-articles "specific" "")
1342                       (symbol-name classification)
1343                       (symbol-name check))
1344         (funcall run-function articles)
1345         ;; now log all the registrations (or undo them, depending on unregister)
1346         (dolist (article articles)
1347           (funcall log-function
1348                    (spam-fetch-field-message-id-fast article)
1349                    'process
1350                    classification
1351                    check
1352                    gnus-newsgroup-name))))))
1353
1354 ;;; log a ham- or spam-processor invocation to the registry
1355 (defun spam-log-processing-to-registry (id type classification check group)
1356   (when spam-log-to-registry
1357     (if (and (stringp id)
1358              (stringp group)
1359              (spam-process-type-valid-p type)
1360              (spam-classification-valid-p classification)
1361              (spam-registration-check-valid-p check))
1362         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1363               (cell (list classification check group)))
1364           (push cell cell-list)
1365           (gnus-registry-store-extra-entry
1366            id
1367            type
1368            cell-list))
1369
1370       (gnus-message
1371        5
1372        (format "%s call with bad ID, type, classification, spam-check, or group"
1373                "spam-log-processing-to-registry")))))
1374
1375 ;;; check if a ham- or spam-processor registration has been done
1376 (defun spam-log-registered-p (id type)
1377   (when spam-log-to-registry
1378     (if (and (stringp id)
1379              (spam-process-type-valid-p type))
1380         (cdr-safe (gnus-registry-fetch-extra id type))
1381       (progn
1382         (gnus-message
1383          5
1384          (format "%s called with bad ID, type, classification, or spam-check"
1385                  "spam-log-registered-p"))
1386         nil))))
1387
1388 ;;; check what a ham- or spam-processor registration says
1389 ;;; returns nil if conflicting registrations are found
1390 (defun spam-log-registration-type (id type)
1391   (let ((count 0)
1392         decision)
1393     (dolist (reg (spam-log-registered-p id type))
1394       (let ((classification (nth 0 reg)))
1395         (when (spam-classification-valid-p classification)
1396           (when (and decision
1397                      (not (eq classification decision)))
1398             (setq count (+ 1 count)))
1399           (setq decision classification))))
1400     (if (< 0 count)
1401         nil
1402       decision)))
1403
1404
1405 ;;; check if a ham- or spam-processor registration needs to be undone
1406 (defun spam-log-unregistration-needed-p (id type classification check)
1407   (when spam-log-to-registry
1408     (if (and (stringp id)
1409              (spam-process-type-valid-p type)
1410              (spam-classification-valid-p classification)
1411              (spam-registration-check-valid-p check))
1412         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1413               found)
1414           (dolist (cell cell-list)
1415             (unless found
1416               (when (and (eq classification (nth 0 cell))
1417                          (eq check (nth 1 cell)))
1418                 (setq found t))))
1419           found)
1420       (progn
1421         (gnus-message
1422          5
1423          (format "%s called with bad ID, type, classification, or spam-check"
1424                  "spam-log-unregistration-needed-p"))
1425         nil))))
1426
1427
1428 ;;; undo a ham- or spam-processor registration (the group is not used)
1429 (defun spam-log-undo-registration (id type classification check &optional group)
1430   (when (and spam-log-to-registry
1431              (spam-log-unregistration-needed-p id type classification check))
1432     (if (and (stringp id)
1433              (spam-process-type-valid-p type)
1434              (spam-classification-valid-p classification)
1435              (spam-registration-check-valid-p check))
1436         (let ((cell-list (cdr-safe (gnus-registry-fetch-extra id type)))
1437               new-cell-list found)
1438           (dolist (cell cell-list)
1439             (unless (and (eq classification (nth 0 cell))
1440                          (eq check (nth 1 cell)))
1441               (push cell new-cell-list)))
1442           (gnus-registry-store-extra-entry
1443            id
1444            type
1445            new-cell-list))
1446       (progn
1447         (gnus-message 6 (format "%s call with bad ID, type, spam-check, or group"
1448                                 "spam-log-undo-registration"))
1449         nil))))
1450
1451 ;;; set up IMAP widening if it's necessary
1452 (defun spam-setup-widening ()
1453   (dolist (check spam-list-of-statistical-checks)
1454     (when (symbol-value check)
1455       (setq nnimap-split-download-body-default t))))
1456
1457 \f
1458 ;;;; Gmane xrefs
1459 (defun spam-check-gmane-xref ()
1460   (let ((header (or
1461                  (message-fetch-field "Xref")
1462                  (message-fetch-field "Newsgroups")))
1463         (spam-split-group (if spam-split-symbolic-return
1464                               'spam
1465                             spam-split-group)))
1466     (when header                        ; return nil when no header
1467       (when (string-match spam-gmane-xref-spam-group
1468                           header)
1469           spam-split-group))))
1470
1471 \f
1472 ;;;; Regex body
1473
1474 (defun spam-check-regex-body ()
1475   (let ((spam-regex-headers-ham spam-regex-body-ham)
1476         (spam-regex-headers-spam spam-regex-body-spam))
1477     (spam-check-regex-headers t)))
1478
1479 \f
1480 ;;;; Regex headers
1481
1482 (defun spam-check-regex-headers (&optional body)
1483   (let ((type (if body "body" "header"))
1484         (spam-split-group (if spam-split-symbolic-return
1485                               'spam
1486                             spam-split-group))
1487         ret found)
1488     (dolist (h-regex spam-regex-headers-ham)
1489       (unless found
1490         (goto-char (point-min))
1491         (when (re-search-forward h-regex nil t)
1492           (message "Ham regex %s search positive." type)
1493           (setq found t))))
1494     (dolist (s-regex spam-regex-headers-spam)
1495       (unless found
1496         (goto-char (point-min))
1497         (when (re-search-forward s-regex nil t)
1498           (message "Spam regex %s search positive." type)
1499           (setq found t)
1500           (setq ret spam-split-group))))
1501     ret))
1502
1503 \f
1504 ;;;; Blackholes.
1505
1506 (defun spam-reverse-ip-string (ip)
1507   (when (stringp ip)
1508     (mapconcat 'identity
1509                (nreverse (split-string ip "\\."))
1510                ".")))
1511
1512 (defun spam-check-blackholes ()
1513   "Check the Received headers for blackholed relays."
1514   (let ((headers (message-fetch-field "received"))
1515         (spam-split-group (if spam-split-symbolic-return
1516                               'spam
1517                             spam-split-group))
1518         ips matches)
1519     (when headers
1520       (with-temp-buffer
1521         (insert headers)
1522         (goto-char (point-min))
1523         (gnus-message 6 "Checking headers for relay addresses")
1524         (while (re-search-forward
1525                 "\\([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+\\)" nil t)
1526           (gnus-message 9 "Blackhole search found host IP %s." (match-string 1))
1527           (push (spam-reverse-ip-string (match-string 1))
1528                 ips)))
1529       (dolist (server spam-blackhole-servers)
1530         (dolist (ip ips)
1531           (unless (and spam-blackhole-good-server-regex
1532                        ;; match the good-server-regex against the reversed (again) IP string
1533                        (string-match
1534                         spam-blackhole-good-server-regex
1535                         (spam-reverse-ip-string ip)))
1536             (unless matches
1537               (let ((query-string (concat ip "." server)))
1538                 (if spam-use-dig
1539                     (let ((query-result (query-dig query-string)))
1540                       (when query-result
1541                         (gnus-message 6 "(DIG): positive blackhole check '%s'"
1542                                       query-result)
1543                         (push (list ip server query-result)
1544                               matches)))
1545                   ;; else, if not using dig.el
1546                   (when (query-dns query-string)
1547                     (gnus-message 6 "positive blackhole check")
1548                     (push (list ip server (query-dns query-string 'TXT))
1549                           matches)))))))))
1550     (when matches
1551       spam-split-group)))
1552 \f
1553 ;;;; Hashcash.
1554
1555 (condition-case nil
1556     (progn
1557       (require 'hashcash)
1558
1559       (defun spam-check-hashcash ()
1560         "Check the headers for hashcash payments."
1561         (mail-check-payment)))   ;mail-check-payment returns a boolean
1562
1563   (file-error (progn
1564                 (defalias 'mail-check-payment 'ignore)
1565                 (defalias 'spam-check-hashcash 'ignore))))
1566 \f
1567 ;;;; BBDB
1568
1569 ;;; original idea for spam-check-BBDB from Alexander Kotelnikov
1570 ;;; <sacha@giotto.sj.ru>
1571
1572 ;; all this is done inside a condition-case to trap errors
1573
1574 (condition-case nil
1575     (progn
1576       (require 'bbdb)
1577       (require 'bbdb-com)
1578
1579       ;; when the BBDB changes, we want to clear out our cache
1580       (defun spam-clear-cache-BBDB (&rest immaterial)
1581         (spam-clear-cache 'spam-use-BBDB))
1582
1583       (add-hook 'bbdb-change-hook 'spam-clear-cache-BBDB)
1584
1585       (defun spam-enter-ham-BBDB (addresses &optional remove)
1586         "Enter an address into the BBDB; implies ham (non-spam) sender"
1587         (dolist (from addresses)
1588           (when (stringp from)
1589             (let* ((parsed-address (gnus-extract-address-components from))
1590                    (name (or (nth 0 parsed-address) "Ham Sender"))
1591                    (remove-function (if remove
1592                                         'bbdb-delete-record-internal
1593                                       'ignore))
1594                    (net-address (nth 1 parsed-address))
1595                    (record (and net-address
1596                                 (bbdb-search-simple nil net-address))))
1597               (when net-address
1598                 (gnus-message 6 "%s address %s %s BBDB"
1599                               (if remove "Deleting" "Adding")
1600                               from
1601                               (if remove "from" "to"))
1602                 (if record
1603                     (funcall remove-function record)
1604                   (bbdb-create-internal name nil net-address nil nil
1605                                         "ham sender added by spam.el")))))))
1606
1607       (defun spam-BBDB-register-routine (articles &optional unregister)
1608         (let (addresses)
1609           (dolist (article articles)
1610             (when (stringp (spam-fetch-field-from-fast article))
1611               (push (spam-fetch-field-from-fast article) addresses)))
1612           ;; now do the register/unregister action
1613           (spam-enter-ham-BBDB addresses unregister)))
1614
1615       (defun spam-BBDB-unregister-routine (articles)
1616         (spam-BBDB-register-routine articles t))
1617
1618       (defun spam-check-BBDB ()
1619         "Mail from people in the BBDB is classified as ham or non-spam"
1620         (let ((who (message-fetch-field "from"))
1621               (spam-split-group (if spam-split-symbolic-return
1622                                     'spam
1623                                   spam-split-group))
1624               bbdb-cache bbdb-hashtable)
1625           (when spam-cache-lookups
1626             (setq bbdb-cache (gethash 'spam-use-BBDB spam-caches))
1627             (unless bbdb-cache
1628               (setq bbdb-cache
1629                     ;; this is the expanded (bbdb-hashtable) macro
1630                     ;; without the debugging support
1631                     (with-current-buffer (bbdb-buffer)
1632                       (save-excursion
1633                         (save-window-excursion
1634                           (bbdb-records nil t)
1635                           bbdb-hashtable))))
1636               (puthash 'spam-use-BBDB bbdb-cache spam-caches)))
1637           (when who
1638             (setq who (nth 1 (gnus-extract-address-components who)))
1639             (if
1640                 (if spam-cache-lookups
1641                     (symbol-value
1642                      (intern-soft who bbdb-cache))
1643                   (bbdb-search-simple nil who))
1644                 t
1645               (if spam-use-BBDB-exclusive
1646                   spam-split-group
1647                 nil))))))
1648
1649   (file-error (progn
1650                 (defalias 'bbdb-search-simple 'ignore)
1651                 (defalias 'bbdb-records 'ignore)
1652                 (defalias 'bbdb-buffer 'ignore)
1653                 (defalias 'spam-check-BBDB 'ignore)
1654                 (defalias 'spam-BBDB-register-routine 'ignore)
1655                 (defalias 'spam-enter-ham-BBDB 'ignore)
1656                 (defalias 'bbdb-create-internal 'ignore)
1657                 (defalias 'bbdb-delete-record-internal 'ignore)
1658                 (defalias 'bbdb-records 'ignore))))
1659
1660 \f
1661 ;;;; ifile
1662
1663 ;;; check the ifile backend; return nil if the mail was NOT classified
1664 ;;; as spam
1665
1666 (defun spam-get-ifile-database-parameter ()
1667   "Get the command-line parameter for ifile's database from
1668   spam-ifile-database-path."
1669   (if spam-ifile-database-path
1670       (format "--db-file=%s" spam-ifile-database-path)
1671     nil))
1672
1673 (defun spam-check-ifile ()
1674   "Check the ifile backend for the classification of this message."
1675   (let ((article-buffer-name (buffer-name))
1676         (spam-split-group (if spam-split-symbolic-return
1677                               'spam
1678                             spam-split-group))
1679         category return)
1680     (with-temp-buffer
1681       (let ((temp-buffer-name (buffer-name))
1682             (db-param (spam-get-ifile-database-parameter)))
1683         (save-excursion
1684           (set-buffer article-buffer-name)
1685           (apply 'call-process-region
1686                  (point-min) (point-max) spam-ifile-path
1687                  nil temp-buffer-name nil "-c"
1688                  (if db-param `(,db-param "-q") `("-q"))))
1689         ;; check the return now (we're back in the temp buffer)
1690         (goto-char (point-min))
1691         (if (not (eobp))
1692             (setq category (buffer-substring (point) (point-at-eol))))
1693         (when (not (zerop (length category))) ; we need a category here
1694           (if spam-ifile-all-categories
1695               (setq return category)
1696             ;; else, if spam-ifile-all-categories is not set...
1697             (when (string-equal spam-ifile-spam-category category)
1698               (setq return spam-split-group)))))) ; note return is nil otherwise
1699     return))
1700
1701 (defun spam-ifile-register-with-ifile (articles category &optional unregister)
1702   "Register an article, given as a string, with a category.
1703 Uses `gnus-newsgroup-name' if category is nil (for ham registration)."
1704   (let ((category (or category gnus-newsgroup-name))
1705         (add-or-delete-option (if unregister "-d" "-i"))
1706         (db (spam-get-ifile-database-parameter))
1707         parameters)
1708     (with-temp-buffer
1709       (dolist (article articles)
1710         (let ((article-string (spam-get-article-as-string article)))
1711           (when (stringp article-string)
1712             (insert article-string))))
1713       (apply 'call-process-region
1714              (point-min) (point-max) spam-ifile-path
1715              nil nil nil
1716              add-or-delete-option category
1717              (if db `(,db "-h") `("-h"))))))
1718
1719 (defun spam-ifile-register-spam-routine (articles &optional unregister)
1720   (spam-ifile-register-with-ifile articles spam-ifile-spam-category unregister))
1721
1722 (defun spam-ifile-unregister-spam-routine (articles)
1723   (spam-ifile-register-spam-routine articles t))
1724
1725 (defun spam-ifile-register-ham-routine (articles &optional unregister)
1726   (spam-ifile-register-with-ifile articles spam-ifile-ham-category unregister))
1727
1728 (defun spam-ifile-unregister-ham-routine (articles)
1729   (spam-ifile-register-ham-routine articles t))
1730
1731 \f
1732 ;;;; spam-stat
1733
1734 (condition-case nil
1735     (progn
1736       (let ((spam-stat-install-hooks nil))
1737         (require 'spam-stat))
1738
1739       (defun spam-check-stat ()
1740         "Check the spam-stat backend for the classification of this message"
1741         (let ((spam-split-group (if spam-split-symbolic-return
1742                                     'spam
1743                                   spam-split-group))
1744               (spam-stat-split-fancy-spam-group spam-split-group) ; override
1745               (spam-stat-buffer (buffer-name)) ; stat the current buffer
1746               category return)
1747           (spam-stat-split-fancy)))
1748
1749       (defun spam-stat-register-spam-routine (articles &optional unregister)
1750         (dolist (article articles)
1751           (let ((article-string (spam-get-article-as-string article)))
1752             (with-temp-buffer
1753               (insert article-string)
1754               (if unregister
1755                   (spam-stat-buffer-change-to-non-spam)
1756               (spam-stat-buffer-is-spam))))))
1757
1758       (defun spam-stat-unregister-spam-routine (articles)
1759         (spam-stat-register-spam-routine articles t))
1760
1761       (defun spam-stat-register-ham-routine (articles &optional unregister)
1762         (dolist (article articles)
1763           (let ((article-string (spam-get-article-as-string article)))
1764             (with-temp-buffer
1765               (insert article-string)
1766               (if unregister
1767                   (spam-stat-buffer-change-to-spam)
1768               (spam-stat-buffer-is-non-spam))))))
1769
1770       (defun spam-stat-unregister-ham-routine (articles)
1771         (spam-stat-register-ham-routine articles t))
1772
1773       (defun spam-maybe-spam-stat-load ()
1774         (when spam-use-stat (spam-stat-load)))
1775
1776       (defun spam-maybe-spam-stat-save ()
1777         (when spam-use-stat (spam-stat-save))))
1778
1779   (file-error (progn
1780                 (defalias 'spam-stat-load 'ignore)
1781                 (defalias 'spam-stat-save 'ignore)
1782                 (defalias 'spam-maybe-spam-stat-load 'ignore)
1783                 (defalias 'spam-maybe-spam-stat-save 'ignore)
1784                 (defalias 'spam-stat-register-ham-routine 'ignore)
1785                 (defalias 'spam-stat-unregister-ham-routine 'ignore)
1786                 (defalias 'spam-stat-register-spam-routine 'ignore)
1787                 (defalias 'spam-stat-unregister-spam-routine 'ignore)
1788                 (defalias 'spam-stat-buffer-is-spam 'ignore)
1789                 (defalias 'spam-stat-buffer-change-to-spam 'ignore)
1790                 (defalias 'spam-stat-buffer-is-non-spam 'ignore)
1791                 (defalias 'spam-stat-buffer-change-to-non-spam 'ignore)
1792                 (defalias 'spam-stat-split-fancy 'ignore)
1793                 (defalias 'spam-check-stat 'ignore))))
1794
1795 \f
1796
1797 ;;;; Blacklists and whitelists.
1798
1799 (defvar spam-whitelist-cache nil)
1800 (defvar spam-blacklist-cache nil)
1801
1802 (defun spam-kill-whole-line ()
1803   (beginning-of-line)
1804   (let ((kill-whole-line t))
1805     (kill-line)))
1806
1807 ;;; address can be a list, too
1808 (defun spam-enter-whitelist (address &optional remove)
1809   "Enter ADDRESS (list or single) into the whitelist.
1810 With a non-nil REMOVE, remove them."
1811   (interactive "sAddress: ")
1812   (spam-enter-list address spam-whitelist remove)
1813   (setq spam-whitelist-cache nil)
1814   (spam-clear-cache 'spam-use-whitelist))
1815
1816 ;;; address can be a list, too
1817 (defun spam-enter-blacklist (address &optional remove)
1818   "Enter ADDRESS (list or single) into the blacklist.
1819 With a non-nil REMOVE, remove them."
1820   (interactive "sAddress: ")
1821   (spam-enter-list address spam-blacklist remove)
1822   (setq spam-blacklist-cache nil)
1823   (spam-clear-cache 'spam-use-whitelist))
1824
1825 (defun spam-enter-list (addresses file &optional remove)
1826   "Enter ADDRESSES into the given FILE.
1827 Either the whitelist or the blacklist files can be used.  With
1828 REMOVE not nil, remove the ADDRESSES."
1829   (if (stringp addresses)
1830       (spam-enter-list (list addresses) file remove)
1831     ;; else, we have a list of addresses here
1832     (unless (file-exists-p (file-name-directory file))
1833       (make-directory (file-name-directory file) t))
1834     (save-excursion
1835       (set-buffer
1836        (find-file-noselect file))
1837       (dolist (a addresses)
1838         (when (stringp a)
1839           (goto-char (point-min))
1840           (if (re-search-forward (regexp-quote a) nil t)
1841               ;; found the address
1842               (when remove
1843                 (spam-kill-whole-line))
1844             ;; else, the address was not found
1845             (unless remove
1846               (goto-char (point-max))
1847               (unless (bobp)
1848                 (insert "\n"))
1849               (insert a "\n")))))
1850       (save-buffer))))
1851
1852 (defun spam-filelist-build-cache (type)
1853   (let ((cache (if (eq type 'spam-use-blacklist)
1854                    spam-blacklist-cache
1855                  spam-whitelist-cache))
1856         parsed-cache)
1857     (unless (gethash type spam-caches)
1858       (while cache
1859         (let ((address (pop cache)))
1860           (unless (zerop (length address)) ; 0 for a nil address too
1861             (setq address (regexp-quote address))
1862             ;; fix regexp-quote's treatment of user-intended regexes
1863             (while (string-match "\\\\\\*" address)
1864               (setq address (replace-match ".*" t t address))))
1865           (push address parsed-cache)))
1866       (puthash type parsed-cache spam-caches))))
1867
1868 (defun spam-filelist-check-cache (type from)
1869   (when (stringp from)
1870     (spam-filelist-build-cache type)
1871     (let (found)
1872       (dolist (address (gethash type spam-caches))
1873         (when (and address (string-match address from))
1874           (setq found t)
1875           (return)))
1876       found)))
1877
1878 ;;; returns t if the sender is in the whitelist, nil or
1879 ;;; spam-split-group otherwise
1880 (defun spam-check-whitelist ()
1881   ;; FIXME!  Should it detect when file timestamps change?
1882   (let ((spam-split-group (if spam-split-symbolic-return
1883                               'spam
1884                             spam-split-group)))
1885     (unless spam-whitelist-cache
1886       (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
1887     (if (spam-from-listed-p 'spam-use-whitelist)
1888         t
1889       (if spam-use-whitelist-exclusive
1890           spam-split-group
1891         nil))))
1892
1893 (defun spam-check-blacklist ()
1894   ;; FIXME!  Should it detect when file timestamps change?
1895   (let ((spam-split-group (if spam-split-symbolic-return
1896                               'spam
1897                             spam-split-group)))
1898     (unless spam-blacklist-cache
1899       (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
1900     (and (spam-from-listed-p 'spam-use-blacklist) spam-split-group)))
1901
1902 (defun spam-parse-list (file)
1903   (when (file-readable-p file)
1904     (let (contents address)
1905       (with-temp-buffer
1906         (insert-file-contents file)
1907         (while (not (eobp))
1908           (setq address (buffer-substring (point) (point-at-eol)))
1909           (forward-line 1)
1910           ;; insert the e-mail address if detected, otherwise the raw data
1911           (unless (zerop (length address))
1912             (let ((pure-address (nth 1 (gnus-extract-address-components address))))
1913               (push (or pure-address address) contents)))))
1914       (nreverse contents))))
1915
1916 (defun spam-from-listed-p (type)
1917   (let ((from (message-fetch-field "from"))
1918         found)
1919     (spam-filelist-check-cache type from)))
1920
1921 (defun spam-filelist-register-routine (articles blacklist &optional unregister)
1922   (let ((de-symbol (if blacklist 'spam-use-whitelist 'spam-use-blacklist))
1923         (declassification (if blacklist 'ham 'spam))
1924         (enter-function
1925          (if blacklist 'spam-enter-blacklist 'spam-enter-whitelist))
1926         (remove-function
1927          (if blacklist 'spam-enter-whitelist 'spam-enter-blacklist))
1928         from addresses unregister-list article-unregister-list)
1929     (dolist (article articles)
1930       (let ((from (spam-fetch-field-from-fast article))
1931             (id (spam-fetch-field-message-id-fast article))
1932             sender-ignored)
1933         (when (stringp from)
1934           (dolist (ignore-regex spam-blacklist-ignored-regexes)
1935             (when (and (not sender-ignored)
1936                        (stringp ignore-regex)
1937                        (string-match ignore-regex from))
1938               (setq sender-ignored t)))
1939           ;; remember the messages we need to unregister, unless remove is set
1940           (when (and
1941                  (null unregister)
1942                  (spam-log-unregistration-needed-p
1943                   id 'process declassification de-symbol))
1944             (push article article-unregister-list)
1945             (push from unregister-list))
1946           (unless sender-ignored
1947             (push from addresses)))))
1948
1949     (if unregister
1950         (funcall enter-function addresses t) ; unregister all these addresses
1951       ;; else, register normally and unregister what we need to
1952       (funcall remove-function unregister-list t)
1953       (dolist (article article-unregister-list)
1954         (spam-log-undo-registration
1955          (spam-fetch-field-message-id-fast article)
1956          'process
1957          declassification
1958          de-symbol))
1959       (funcall enter-function addresses nil))))
1960
1961 (defun spam-blacklist-unregister-routine (articles)
1962   (spam-blacklist-register-routine articles t))
1963
1964 (defun spam-blacklist-register-routine (articles &optional unregister)
1965   (spam-filelist-register-routine articles t unregister))
1966
1967 (defun spam-whitelist-unregister-routine (articles)
1968   (spam-whitelist-register-routine articles t))
1969
1970 (defun spam-whitelist-register-routine (articles &optional unregister)
1971   (spam-filelist-register-routine articles nil unregister))
1972
1973 \f
1974 ;;;; Spam-report glue
1975 (defun spam-report-gmane-register-routine (articles)
1976   (when articles
1977     (apply 'spam-report-gmane articles)))
1978
1979 \f
1980 ;;;; Bogofilter
1981 (defun spam-check-bogofilter-headers (&optional score)
1982   (let ((header (message-fetch-field spam-bogofilter-header))
1983         (spam-split-group (if spam-split-symbolic-return
1984                               'spam
1985                             spam-split-group)))
1986     (when header                        ; return nil when no header
1987       (if score                         ; scoring mode
1988           (if (string-match "spamicity=\\([0-9.]+\\)" header)
1989               (match-string 1 header)
1990             "0")
1991         ;; spam detection mode
1992         (when (string-match spam-bogofilter-bogosity-positive-spam-header
1993                             header)
1994           spam-split-group)))))
1995
1996 ;; return something sensible if the score can't be determined
1997 (defun spam-bogofilter-score ()
1998   "Get the Bogofilter spamicity score"
1999   (interactive)
2000   (save-window-excursion
2001     (gnus-summary-show-article t)
2002     (set-buffer gnus-article-buffer)
2003     (let ((score (or (spam-check-bogofilter-headers t)
2004                      (spam-check-bogofilter t))))
2005       (gnus-summary-show-article)
2006       (message "Spamicity score %s" score)
2007       (or score "0"))))
2008
2009 (defun spam-check-bogofilter (&optional score)
2010   "Check the Bogofilter backend for the classification of this message"
2011   (let ((article-buffer-name (buffer-name))
2012         (db spam-bogofilter-database-directory)
2013         return)
2014     (with-temp-buffer
2015       (let ((temp-buffer-name (buffer-name)))
2016         (save-excursion
2017           (set-buffer article-buffer-name)
2018           (apply 'call-process-region
2019                  (point-min) (point-max)
2020                  spam-bogofilter-path
2021                  nil temp-buffer-name nil
2022                  (if db `("-d" ,db "-v") `("-v"))))
2023         (setq return (spam-check-bogofilter-headers score))))
2024     return))
2025
2026 (defun spam-bogofilter-register-with-bogofilter (articles
2027                                                  spam
2028                                                  &optional unregister)
2029   "Register an article, given as a string, as spam or non-spam."
2030   (dolist (article articles)
2031     (let ((article-string (spam-get-article-as-string article))
2032           (db spam-bogofilter-database-directory)
2033           (switch (if unregister
2034                       (if spam
2035                           spam-bogofilter-spam-strong-switch
2036                         spam-bogofilter-ham-strong-switch)
2037                     (if spam
2038                         spam-bogofilter-spam-switch
2039                       spam-bogofilter-ham-switch))))
2040       (when (stringp article-string)
2041         (with-temp-buffer
2042           (insert article-string)
2043
2044           (apply 'call-process-region
2045                  (point-min) (point-max)
2046                  spam-bogofilter-path
2047                  nil nil nil switch
2048                  (if db `("-d" ,db "-v") `("-v"))))))))
2049
2050 (defun spam-bogofilter-register-spam-routine (articles &optional unregister)
2051   (spam-bogofilter-register-with-bogofilter articles t unregister))
2052
2053 (defun spam-bogofilter-unregister-spam-routine (articles)
2054   (spam-bogofilter-register-spam-routine articles t))
2055
2056 (defun spam-bogofilter-register-ham-routine (articles &optional unregister)
2057   (spam-bogofilter-register-with-bogofilter articles nil unregister))
2058
2059 (defun spam-bogofilter-unregister-ham-routine (articles)
2060   (spam-bogofilter-register-ham-routine articles t))
2061
2062
2063 \f
2064 ;;;; spamoracle
2065 (defun spam-check-spamoracle ()
2066   "Run spamoracle on an article to determine whether it's spam."
2067   (let ((article-buffer-name (buffer-name))
2068         (spam-split-group (if spam-split-symbolic-return
2069                               'spam
2070                             spam-split-group)))
2071     (with-temp-buffer
2072       (let ((temp-buffer-name (buffer-name)))
2073         (save-excursion
2074           (set-buffer article-buffer-name)
2075           (let ((status
2076                  (apply 'call-process-region
2077                         (point-min) (point-max)
2078                         spam-spamoracle-binary
2079                         nil temp-buffer-name nil
2080                         (if spam-spamoracle-database
2081                             `("-f" ,spam-spamoracle-database "mark")
2082                           '("mark")))))
2083             (if (eq 0 status)
2084                 (progn
2085                   (set-buffer temp-buffer-name)
2086                   (goto-char (point-min))
2087                   (when (re-search-forward "^X-Spam: yes;" nil t)
2088                     spam-split-group))
2089               (error "Error running spamoracle: %s" status))))))))
2090
2091 (defun spam-spamoracle-learn (articles article-is-spam-p &optional unregister)
2092   "Run spamoracle in training mode."
2093   (with-temp-buffer
2094     (let ((temp-buffer-name (buffer-name)))
2095       (save-excursion
2096         (goto-char (point-min))
2097         (dolist (article articles)
2098           (insert (spam-get-article-as-string article)))
2099         (let* ((arg (if (spam-xor unregister article-is-spam-p)
2100                         "-spam"
2101                       "-good"))
2102                (status
2103                 (apply 'call-process-region
2104                        (point-min) (point-max)
2105                        spam-spamoracle-binary
2106                        nil temp-buffer-name nil
2107                        (if spam-spamoracle-database
2108                            `("-f" ,spam-spamoracle-database
2109                              "add" ,arg)
2110                          `("add" ,arg)))))
2111           (unless (eq 0 status)
2112             (error "Error running spamoracle: %s" status)))))))
2113
2114 (defun spam-spamoracle-learn-ham (articles &optional unregister)
2115   (spam-spamoracle-learn articles nil unregister))
2116
2117 (defun spam-spamoracle-unlearn-ham (articles &optional unregister)
2118   (spam-spamoracle-learn-ham articles t))
2119
2120 (defun spam-spamoracle-learn-spam (articles &optional unregister)
2121   (spam-spamoracle-learn articles t unregister))
2122
2123 (defun spam-spamoracle-unlearn-spam (articles &optional unregister)
2124   (spam-spamoracle-learn-spam articles t))
2125
2126 \f
2127 ;;;; SpamAssassin
2128 ;;; based mostly on the bogofilter code
2129 (defun spam-check-spamassassin-headers (&optional score)
2130   "Check the SpamAssassin headers for the classification of this message."
2131   (if score                             ; scoring mode
2132       (let ((header (message-fetch-field spam-spamassassin-spam-status-header)))
2133         (when header
2134           (if (string-match "hits=\\(-?[0-9.]+\\)" header)
2135               (match-string 1 header)
2136             "0")))
2137     ;; spam detection mode
2138     (let ((header (message-fetch-field spam-spamassassin-spam-flag-header))
2139           (spam-split-group (if spam-split-symbolic-return
2140                                  'spam
2141                                spam-split-group)))
2142           (when header                  ; return nil when no header
2143             (when (string-match spam-spamassassin-positive-spam-flag-header
2144                                 header)
2145               spam-split-group)))))
2146
2147 (defun spam-check-spamassassin (&optional score)
2148   "Check the SpamAssassin backend for the classification of this message."
2149   (let ((article-buffer-name (buffer-name)))
2150     (with-temp-buffer
2151       (let ((temp-buffer-name (buffer-name)))
2152         (save-excursion
2153           (set-buffer article-buffer-name)
2154           (apply 'call-process-region
2155                  (point-min) (point-max) spam-spamassassin-path
2156                  nil temp-buffer-name nil spam-spamassassin-arguments))
2157         ;; check the return now (we're back in the temp buffer)
2158         (goto-char (point-min))
2159         (spam-check-spamassassin-headers score)))))
2160
2161 ;; return something sensible if the score can't be determined
2162 (defun spam-spamassassin-score ()
2163   "Get the SpamAssassin score"
2164   (interactive)
2165   (save-window-excursion
2166     (gnus-summary-show-article t)
2167     (set-buffer gnus-article-buffer)
2168     (let ((score (or (spam-check-spamassassin-headers t)
2169                      (spam-check-spamassassin t))))
2170       (gnus-summary-show-article)
2171       (message "SpamAssassin score %s" score)
2172       (or score "0"))))
2173
2174 (defun spam-spamassassin-register-with-sa-learn (articles spam
2175                                                  &optional unregister)
2176   "Register articles with spamassassin's sa-learn as spam or non-spam."
2177   (if articles
2178       (let ((action (if unregister spam-sa-learn-unregister-switch
2179                       (if spam spam-sa-learn-spam-switch
2180                         spam-sa-learn-ham-switch)))
2181             (summary-buffer-name (buffer-name)))
2182         (with-temp-buffer
2183           ;; group the articles into mbox format
2184           (dolist (article articles)
2185             (let (article-string)
2186               (save-excursion
2187                 (set-buffer summary-buffer-name)
2188                 (setq article-string (spam-get-article-as-string article)))
2189               (when (stringp article-string)
2190                 (insert "From \n") ; mbox separator (sa-learn only checks the
2191                                    ; first five chars, so we can get away with
2192                                    ; a bogus line))
2193                 (insert article-string)
2194                 (insert "\n"))))
2195           ;; call sa-learn on all messages at the same time
2196           (apply 'call-process-region
2197                  (point-min) (point-max)
2198                  spam-sa-learn-path
2199                  nil nil nil "--mbox"
2200                  (if spam-sa-learn-rebuild
2201                      (list action)
2202                    `("--no-rebuild" ,action)))))))
2203
2204 (defun spam-spamassassin-register-spam-routine (articles &optional unregister)
2205   (spam-spamassassin-register-with-sa-learn articles t unregister))
2206
2207 (defun spam-spamassassin-register-ham-routine (articles &optional unregister)
2208   (spam-spamassassin-register-with-sa-learn articles nil unregister))
2209
2210 (defun spam-spamassassin-unregister-spam-routine (articles)
2211   (spam-spamassassin-register-with-sa-learn articles t t))
2212
2213 (defun spam-spamassassin-unregister-ham-routine (articles)
2214   (spam-spamassassin-register-with-sa-learn articles nil t))
2215 \f
2216 ;;;; Hooks
2217
2218 ;;;###autoload
2219 (defun spam-initialize (&rest symbols)
2220   "Install the spam.el hooks and do other initialization.
2221 When SYMBOLS is given, set those variables to t.  This is so you
2222 can call spam-initialize before you set spam-use-* variables on
2223 explicitly, and matters only if you need the extra headers
2224 installed through spam-necessary-extra-headers."
2225   (interactive)
2226
2227   (dolist (var symbols)
2228     (set var t))
2229
2230   (dolist (header (spam-necessary-extra-headers))
2231     (add-to-list 'nnmail-extra-headers header)
2232     (add-to-list 'gnus-extra-headers header))
2233
2234   (setq spam-install-hooks t)
2235   ;; TODO: How do we redo this every time spam-face is customized?
2236   (push '((eq mark gnus-spam-mark) . spam-face)
2237         gnus-summary-highlight)
2238   ;; Add hooks for loading and saving the spam stats
2239   (add-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2240   (add-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2241   (add-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2242   (add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2243   (add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2244   (add-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2245   (add-hook 'gnus-summary-prepared-hook 'spam-find-spam))
2246
2247 (defun spam-unload-hook ()
2248   "Uninstall the spam.el hooks"
2249   (interactive)
2250   (remove-hook 'gnus-save-newsrc-hook 'spam-maybe-spam-stat-save)
2251   (remove-hook 'gnus-get-top-new-news-hook 'spam-maybe-spam-stat-load)
2252   (remove-hook 'gnus-startup-hook 'spam-maybe-spam-stat-load)
2253   (remove-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)
2254   (remove-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
2255   (remove-hook 'gnus-get-new-news-hook 'spam-setup-widening)
2256   (remove-hook 'gnus-summary-prepare-hook 'spam-find-spam))
2257
2258 (when spam-install-hooks
2259   (spam-initialize))
2260
2261 (provide 'spam)
2262
2263 ;;; spam.el ends here