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

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