minor documentation change

[gnus] / lisp / spam.el

;;; spam.el --- Identifying spam
;; Copyright (C) 2002 Free Software Foundation, Inc.

;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
;; Keywords: network

;; This file is part of GNU Emacs.

;; GNU Emacs is free software; you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation; either version 2, or (at your option)
;; any later version.

;; GNU Emacs is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
;; GNU General Public License for more details.

;; You should have received a copy of the GNU General Public License
;; along with GNU Emacs; see the file COPYING.  If not, write to the
;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
;; Boston, MA 02111-1307, USA.

;;; Commentary:

;;; This module addresses a few aspects of spam control under Gnus.  Page
;;; breaks are used for grouping declarations and documentation relating to
;;; each particular aspect.

;;; The integration with Gnus is not yet complete.  See various `FIXME'
;;; comments, below, for supplementary explanations or discussions.

;;; Code:

(require 'gnus-sum)

;; FIXME!  We should not require `dns' nor `message' until we actually
;; need them.  Best would be to declare needed functions as auto-loadable.
(require 'dns)
(require 'message)

;;; Main parameters.

(defvar spam-use-blacklist t
  "True if the blacklist should be used.")

(defvar spam-use-whitelist t
  "True if the whitelist should be used.")

(defvar spam-use-blackholes nil
  ;; FIXME!  Turned off for now.  The DNS routines are said to be flaky.
  "True if blackholes should be used.")

(defvar spam-use-bogofilter t
  "True if bogofilter should be used.")

(defvar spam-split-group "spam"
  "Usual group name where spam should be split.")

(defvar spam-junk-mailgroups
  ;; FIXME!  The mailgroup list evidently depends on other choices made by the
  ;; user, so the built-in default below is not likely to be appropriate.
  (cons spam-split-group '("mail.junk" "poste.pourriel"))
  "Mailgroups which are dedicated by splitting to receive various junk.
All unmarked article in such group receive the spam mark on group entry.")

;; FIXME!  For `spam-ham-marks' and `spam-spam-marks', I wonder if it would
;; not be easier for the user to just accept a string of mark letters, instead
;; of a list of Gnus variable names.  In such case, the stunt of deferred
;; evaluation would not be useful anymore.  Lars?? :-)

;; FIXME!  It is rather questionable to see `K', `X' and `Y' as indicating
;; positive ham.  It much depends on how and why people use kill files, score
;; files, and the kill command.  Maybe it would be better, by default, to not
;; process a message neither as ham nor spam, that is, just ignore it for
;; learning purposes, when we are not sure of how the user sees it.
;; But `r' and `R' should undoubtedly be seen as ham.

;; FIXME!  Some might consider overkill to define a list of spam marks.  On
;; the other hand, who knows, some users might for example like that
;; explicitly `E'xpired articles be processed as positive spam.

(defvar spam-ham-marks
  (list gnus-del-mark gnus-read-mark gnus-killed-mark
         gnus-kill-file-mark gnus-low-score-mark)
  "Marks considered as being ham (positively not spam).
Such articles will be transmitted to `bogofilter -n' on group exit.")

(defvar spam-spam-marks
  (list gnus-spam-mark)
  "Marks considered as being spam (positively spam).
Such articles will be transmitted to `bogofilter -s' on group exit.")

;; FIXME!  Ideally, the remainder of this page should be fully integrated
;; within `gnus-sum.el'.

;;; Key bindings for spam control.

;; FIXME!  The justification for `M-d' is that this is what Paul Graham
;; suggests in his original article, and what Eric Raymond's patch for Mutt
;; uses.  But more importantly, that binding was still free in Summary mode!

;; FIXME!  Lars has not blessed the following key bindings yet.  It looks
;; convenient that the score analysis command uses a sequence ending with the
;; letter `t', so it nicely parallels `B t' or `V t'.  `M-d' is a kind of
;; "alternate" `d', it is also the sequence suggested in Paul Graham article,
;; and also in Eric Raymond's patch for Mutt.  `S x' might be the more
;; official key binding for `M-d'.

(gnus-define-keys gnus-summary-mode-map
  "St" spam-bogofilter-score
  "Sx" gnus-summary-mark-as-spam
  "\M-d" gnus-summary-mark-as-spam)

;;; How to highlight a spam summary line.

;; FIXME!  Of course, `gnus-splash-face' has another purpose.  Maybe a
;; special face should be created, named and used instead, for spam lines.

(push '((eq mark gnus-spam-mark) . gnus-splash-face)
      gnus-summary-highlight)

;;; Hooks dispatching.  A bit raw for now.

(defun spam-summary-prepare ()
  (spam-mark-junk-as-spam-routine))

(defun spam-summary-prepare-exit ()
  (spam-bogofilter-register-routine))

(add-hook 'gnus-summary-prepare-hook 'spam-summary-prepare)
(add-hook 'gnus-summary-prepare-exit-hook 'spam-summary-prepare-exit)

(defun spam-mark-junk-as-spam-routine ()
  (when (member gnus-newsgroup-name spam-junk-mailgroups)
    (let ((articles gnus-newsgroup-articles)
          article)
      (while articles
        (setq article (pop articles))
        (when (eq (gnus-summary-article-mark article) gnus-unread-mark)
          (gnus-summary-mark-article article gnus-spam-mark))))))
\f
;;;; Spam determination.

;; The following list contains pairs associating a parameter variable with a
;; spam checking function.  If the parameter variable is true, then the
;; checking function is called, and its value decides what happens.  Each
;; individual check may return `nil', `t', or a mailgroup name.  The value
;; `nil' means that the check does not yield a decision, and so, that further
;; checks are needed.  The value `t' means that the message is definitely not
;; spam, and that further spam checks should be inhibited.  Otherwise, a
;; mailgroup name is returned where the mail should go, and further checks are
;; also inhibited.  The usual mailgroup name is the value of
;; `spam-split-group', meaning that the message is definitely a spam.

(defvar spam-list-of-checks
  '((spam-use-blacklist  . spam-check-blacklist)
    (spam-use-whitelist  . spam-check-whitelist)
    (spam-use-bbdb       . spam-check-bbdb)
    (spam-use-blackholes . spam-check-blackholes)
    (spam-use-bogofilter . spam-check-bogofilter)))

(defun spam-split ()
  "Split this message into the `spam' group if it is spam.
This function can be used as an entry in `nnmail-split-fancy', for
example like this: (: spam-split)

See the Info node `(gnus)Fancy Mail Splitting' for more details."
  (interactive)

  (let ((list-of-checks spam-list-of-checks)
        decision)
    (while (and list-of-checks (not decision))
      (let ((pair (pop list-of-checks)))
        (when (eval (car pair))
          (setq decision (apply (cdr pair))))))
    (if (eq decision t)
        nil
      decision)))
\f
;;;; Blackholes.

(defvar spam-blackhole-servers '("bl.spamcop.net"
                                 "relays.ordb.org"
                                 "dev.null.dk"
                                 "relays.visi.com"
                                 "rbl.maps.vix.com")
  "List of blackhole servers.")

(defun spam-check-blackholes ()
  "Check the Receieved headers for blackholed relays."
  (let ((headers (message-fetch-field "received"))
        ips matches)
    (when headers
      (with-temp-buffer
        (insert headers)
        (goto-char (point-min))
        (while (re-search-forward
                "\\[\\([0-9]+.[0-9]+.[0-9]+.[0-9]+\\)\\]" nil t)
          (message "Blackhole search found host IP %s." (match-string 1))
          (push (mapconcat 'identity
                           (nreverse (split-string (match-string 1) "\\."))
                           ".")
                ips)))
      (dolist (server spam-blackhole-servers)
        (dolist (ip ips)
          (when (query-dns (concat ip "." server))
            (push (list ip server (query-dns (concat ip "." server) 'TXT))
                  matches)))))
    (when matches
      spam-split-group)))
\f
;;;; Blacklists and whitelists.

(defvar spam-directory "~/News/spam/"
  "When spam files are kept.")

(defvar spam-whitelist (expand-file-name "whitelist" spam-directory)
  "The location of the whitelist.
The file format is one regular expression per line.
The regular expression is matched against the address.")

(defvar spam-blacklist (expand-file-name "blacklist" spam-directory)
  "The location of the blacklist.
The file format is one regular expression per line.
The regular expression is matched against the address.")

(defvar spam-whitelist-cache nil)
(defvar spam-blacklist-cache nil)

(defun spam-enter-whitelist (address)
  "Enter ADDRESS into the whitelist."
  (interactive "sAddress: ")
  (spam-enter-list address spam-whitelist)
  (setq spam-whitelist-cache nil))

(defun spam-enter-blacklist (address)
  "Enter ADDRESS into the blacklist."
  (interactive "sAddress: ")
  (spam-enter-list address spam-blacklist)
  (setq spam-blacklist-cache nil))

(defun spam-enter-list (address file)
  "Enter ADDRESS into the given FILE, either the whitelist or the blacklist."
  (unless (file-exists-p (file-name-directory file))
    (make-directory (file-name-directory file) t))
  (save-excursion
    (set-buffer
     (find-file-noselect file))
    (goto-char (point-max))
    (unless (bobp)
      (insert "\n"))
    (insert address "\n")
    (save-buffer)))

(defun spam-check-whitelist ()
  ;; FIXME!  Should it detect when file timestamps change?
  (unless spam-whitelist-cache
    (setq spam-whitelist-cache (spam-parse-list spam-whitelist)))
  (and (spam-from-listed-p spam-whitelist-cache) t))

;;; copied from code by Alexander Kotelnikov <sacha@giotto.sj.ru>
;; FIXME: assumes that bbdb.el is loaded
;; FIXME: not sure about the logic...
(defun spam-check-bbdb ()
  "We want people, who are in bbdb not to be splitted to spam"
  (let ((who (cadr
              (gnus-extract-address-components (message-fetch-field "from"))
              )))
    (bbdb-search (bbdb-records) nil nil (regexp-quote who))))

(defun spam-check-blacklist ()
  ;; FIXME!  Should it detect when file timestamps change?
  (unless spam-blacklist-cache
    (setq spam-blacklist-cache (spam-parse-list spam-blacklist)))
  (and (spam-from-listed-p spam-blacklist-cache) spam-split-group))

(eval-and-compile
  (defalias 'spam-point-at-eol (if (fboundp 'point-at-eol)
                                   'point-at-eol
                                 'line-end-position)))

(defun spam-parse-list (file)
  (when (file-readable-p file)
    (let (contents address)
      (with-temp-buffer
        (insert-file-contents file)
        (while (not (eobp))
          (setq address (buffer-substring (point) (spam-point-at-eol)))
          (forward-line 1)
          (unless (zerop (length address))
            (setq address (regexp-quote address))
            (while (string-match "\\\\\\*" address)
              (setq address (replace-match ".*" t t address)))
            (push address contents))))
      (nreverse contents))))

(defun spam-from-listed-p (cache)
  (let ((from (message-fetch-field "from"))
        found)
    (while cache
      (when (string-match (pop cache) from)
        (setq found t
              cache nil)))
    found))
\f
;;;; Training via Bogofilter.   Last updated 2002-09-02.

;;; See Paul Graham article, at `http://www.paulgraham.com/spam.html'.

;;; This page is for those wanting to control spam with the help of Eric
;;; Raymond's speedy Bogofilter, see http://www.tuxedo.org/~esr/bogofilter.
;;; This has been tested with a locally patched copy of version 0.4.

;;; Make sure Bogofilter is installed.  Bogofilter internally uses Judy fast
;;; associative arrays, so you need to install Judy first, and Bogofilter
;;; next.  Fetch both distributions by visiting the following links and
;;; downloading the latest version of each:
;;;
;;;     http://sourceforge.net/projects/judy/
;;;     http://www.tuxedo.org/~esr/bogofilter/
;;;
;;; Unpack the Judy distribution and enter its main directory.  Then do:
;;;
;;;     ./configure
;;;     make
;;;     make install
;;;
;;; You will likely need to become super-user for the last step.  Then, unpack
;;; the Bogofilter distribution and enter its main directory:
;;;
;;;     make
;;;     make install
;;;
;;; Here as well, you need to become super-user for the last step.  Now,
;;; initialises your word lists by doing, under your own identity:
;;;
;;;     mkdir ~/.bogofilter
;;;     touch ~/.bogofilter/badlist
;;;     touch ~/.bogofilter/goodlist
;;;
;;; These two files are text files you may edit, but you normally don't!

;;; The `M-d' command gets added to Gnus summary mode, marking current article
;;; as spam, showing it with the `H' mark.  Whenever you see a spam article,
;;; make sure to mark its summary line with `M-d' before leaving the group.
;;; Some groups, as per variable `spam-junk-mailgroups' below, receive articles
;;; from Gnus splitting on clues added by spam recognisers, so for these
;;; groups, we tack an `H' mark at group entry for all summary lines which
;;; would otherwise have no other mark.  Make sure to _remove_ `H' marks for
;;; any article which is _not_ genuine spam, before leaving such groups: you
;;; may use `M-u' to "unread" the article, or `d' for declaring it read the
;;; non-spam way.  When you leave a group, all `H' marked articles, saved or
;;; unsaved, are sent to Bogofilter which will study them as spam samples.

;;; Messages may also be deleted in various other ways, and unless
;;; `spam-ham-marks-form' gets overridden below, marks `R' and `r' for default
;;; read or explicit delete, marks `X' and 'K' for automatic or explicit
;;; kills, as well as mark `Y' for low scores, are all considered to be
;;; associated with articles which are not spam.  This assumption might be
;;; false, in particular if you use kill files or score files as means for
;;; detecting genuine spam, you should then adjust `spam-ham-marks-form'.  When
;;; you leave a group, all _unsaved_ articles bearing any the above marks are
;;; sent to Bogofilter which will study these as not-spam samples.  If you
;;; explicit kill a lot, you might sometimes end up with articles marked `K'
;;; which you never saw, and which might accidentally contain spam.  Best is
;;; to make sure that real spam is marked with `H', and nothing else.

;;; All other marks do not contribute to Bogofilter pre-conditioning.  In
;;; particular, ticked, dormant or souped articles are likely to contribute
;;; later, when they will get deleted for real, so there is no need to use
;;; them prematurely.  Explicitly expired articles do not contribute, command
;;; `E' is a way to get rid of an article without Bogofilter ever seeing it.

;;; In a word, with a minimum of care for associating the `H' mark for spam
;;; articles only, Bogofilter training all gets fairly automatic.  You should
;;; do this until you get a few hundreds of articles in each category, spam
;;; or not.  The shell command `head -1 ~/.bogofilter/*' shows both article
;;; counts.  The command `S S' in summary mode, either for debugging or for
;;; curiosity, triggers Bogofilter into displaying in another buffer the
;;; "spamicity" score of the current article (between 0.0 and 1.0), together
;;; with the article words which most significantly contribute to the score.

;;; The real way for using Bogofilter, however, is to have some use tool like
;;; `procmail' for invoking it on message reception, then adding some
;;; recognisable header in case of detected spam.  Gnus splitting rules might
;;; later trip on these added headers and react by sorting such articles into
;;; specific junk folders as per `spam-junk-mailgroups'.  Here is a possible
;;; `.procmailrc' contents (still untested -- please tell me how it goes):
;;;
;;; :0HBf:
;;; * ? bogofilter
;;; | formail -bfI "X-Spam-Status: Yes"

(defvar spam-output-buffer-name "*Bogofilter Output*"
  "Name of buffer when displaying `bogofilter -v' output.")

(defvar spam-spaminfo-header-regexp
  ;; FIXME!  In the following regexp, we should explain which tool produces
  ;; which kind of header.  I do not even remember them all by now.  X-Junk
  ;; (and previously X-NoSpam) are produced by the `NoSpam' tool, which has
  ;; never been published, so it might not be reasonable leaving it in the
  ;; list.
  "^X-\\(jf\\|Junk\\|NoSpam\\|Spam\\|SB\\)[^:]*:"
  "Regexp for spam markups in headers.
Markup from spam recognisers, as well as `Xref', are to be removed from
articles before they get registered by Bogofilter.")

;; FIXME!  I do not know if Gnus has a compatibility function for
;; `executable-find'.  Here is a possible mantra for portability,
;; until Lars decides how we really should do it.
(unless (fboundp 'executable-find)
  (if (fboundp 'locate-file)
      (defun executable-find (command)
        (locate-file command exec-path))
    (autoload 'executable-find "executable")))
;; End of portability mantra for `executable-find'.

(defvar spam-bogofilter-path (executable-find "bogofilter")
  "File path of the Bogofilter executable program.
Force this variable to nil if you want to inhibit the functionality.")

(defun spam-check-bogofilter ()
  ;; Dynamic spam check.  I do not know how to check the exit status,
  ;; so instead, read `bogofilter -v' output.
  (when (and spam-use-bogofilter spam-bogofilter-path)
    (spam-bogofilter-articles nil "-v" (list (gnus-summary-article-number)))
    (when (save-excursion
            (set-buffer spam-output-buffer-name)
            (goto-char (point-min))
            (re-search-forward "Spamicity: \\(0\\.9\\|1\\.0\\)" nil t))
      spam-split-group)))

(defun spam-bogofilter-score ()
  "Use `bogofilter -v' on the current article.
This yields the 15 most discriminant words for this article and the
spamicity coefficient of each, and the overall article spamicity."
  (interactive)
  (when (and spam-use-bogofilter spam-bogofilter-path)
    (spam-bogofilter-articles nil "-v" (list (gnus-summary-article-number)))
    (save-excursion
      (set-buffer spam-output-buffer-name)
      (unless (= (point-min) (point-max))
        (display-message-or-buffer (current-buffer)
                                   spam-output-buffer-name)))))

(defun spam-bogofilter-register-routine ()
  (when (and spam-use-bogofilter spam-bogofilter-path)
    (let ((articles gnus-newsgroup-articles)
          article mark ham-articles spam-articles)
      (while articles
        (setq article (pop articles)
              mark (gnus-summary-article-mark article))
        (cond ((memq mark spam-spam-marks) (push article spam-articles))
              ((memq article gnus-newsgroup-saved))
              ((memq mark spam-ham-marks) (push article ham-articles))))
      (when ham-articles
        (spam-bogofilter-articles "ham" "-n" ham-articles))
      (when spam-articles
        (spam-bogofilter-articles "SPAM" "-s" spam-articles)))))

(defvar spam-bogofilter-initial-timeout 40
  "Timeout in seconds for the initial reply from the `bogofilter' program.")

(defvar spam-bogofilter-subsequent-timeout 15
  "Timeout in seconds for any subsequent reply from the `bogofilter' program.")

(defun spam-bogofilter-articles (type option articles)
  (let ((output-buffer (get-buffer-create spam-output-buffer-name))
        (article-copy (get-buffer-create " *Bogofilter Article Copy*"))
        (remove-regexp (concat spam-spaminfo-header-regexp "\\|Xref:"))
        (counter 0)
        prefix process article)
    (when type
      (setq prefix (format "Studying %d articles as %s..." (length articles)
                           type))
      (message "%s" prefix))
    (save-excursion (set-buffer output-buffer) (erase-buffer))
    (setq process (start-process "bogofilter" output-buffer
                                 spam-bogofilter-path "-F" option))
    (process-kill-without-query process t)
    (unwind-protect
        (save-window-excursion
          (while articles
            (setq counter (1+ counter))
            (when prefix
              (message "%s %d" prefix counter))
            (setq article (pop articles))
            (gnus-summary-goto-subject article)
            (gnus-summary-select-article)
            (gnus-eval-in-buffer-window article-copy
              (insert-buffer-substring gnus-original-article-buffer)
              ;; Remove spam classification redundant headers: they may induce
              ;; unwanted biases in later analysis.
              (goto-char (point-min))
              (while (not (or (eobp) (= (following-char) ?\n)))
                (if (looking-at remove-regexp)
                    (delete-region (point)
                                   (save-excursion (forward-line 1) (point)))
                  (forward-line 1)))
              (goto-char (point-min))
              ;; Bogofilter really wants From envelopes for counting articles.
              ;; Fake one at the beginning, make sure there will be no other.
              (if (looking-at "From ")
                  (forward-line 1)
                (insert "From nobody " (current-time-string) "\n"))
              (let (case-fold-search)
                (while (re-search-forward "^From " nil t)
                  (beginning-of-line)
                  (insert ">")))
              (process-send-region process (point-min) (point-max))
              (erase-buffer))))
      ;; Sending the EOF is unwind-protected.  This is to prevent lost copies
      ;; of `bogofilter', hung on reading their standard input, in case the
      ;; whole registering process gets interrupted by the user.
      (process-send-eof process))
    (kill-buffer article-copy)
    ;; Receive process output.  It sadly seems that we still have to protect
    ;; ourselves against hung `bogofilter' processes.
    (let ((status (process-status process))
          (timeout (* 1000 spam-bogofilter-initial-timeout))
          (quanta 200))                 ; also counted in milliseconds
      (while (and (not (eq status 'exit)) (> timeout 0))
        ;; `accept-process-output' timeout is counted in microseconds.
        (setq timeout (if (accept-process-output process 0 (* 1000 quanta))
                          (* 1000 spam-bogofilter-subsequent-timeout)
                        (- timeout quanta))
              status (process-status process)))
      (if (eq status 'exit)
          (when prefix
            (message "%s done!" prefix))
        ;; Sigh!  The process did time out...  Become brutal!
        (interrupt-process process)
        (message "%s %d INTERRUPTED!  (Article %d, status %s)"
                 (or prefix "Bogofilter process...")
                 counter article status)
        ;; Give some time for user to read.  Sitting redisplays but gives up
        ;; if input is pending.  Sleeping does not give up, but it does not
        ;; redisplay either.  Mix both: let's redisplay and not give up.
        (sit-for 1)
        (sleep-for 3)))))

(provide 'spam)

;;; spam.el ends here.