1;;; spam-stat.el --- detecting spam based on statistics
2
3;; Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
4
5;; Author: Alex Schroeder <alex@gnu.org>
6;; Keywords: network
7;; URL: http://www.emacswiki.org/cgi-bin/wiki.pl?SpamStat
8
9;; This file is part of GNU Emacs.
10
11;; This is free software; you can redistribute it and/or modify it
12;; under the terms of the GNU General Public License as published by
13;; the Free Software Foundation; either version 2, or (at your option)
14;; any later version.
15
16;; This is distributed in the hope that it will be useful, but WITHOUT
17;; ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
18;; or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
19;; License for more details.
20
21;; You should have received a copy of the GNU General Public License
22;; along with GNU Emacs; see the file COPYING.  If not, write to the
23;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
24;; Boston, MA 02110-1301, USA.
25
26;;; Commentary:
27
28;; This implements spam analysis according to Paul Graham in "A Plan
29;; for Spam".  The basis for all this is a statistical distribution of
30;; words for your spam and non-spam mails.  We need this information
31;; in a hash-table so that the analysis can use the information when
32;; looking at your mails.  Therefore, before you begin, you need tons
33;; of mails (Graham uses 4000 non-spam and 4000 spam mails for his
34;; experiments).
35;;
36;; The main interface to using spam-stat, are the following functions:
37;;
38;; `spam-stat-buffer-is-spam' -- called in a buffer, that buffer is
39;; considered to be a new spam mail; use this for new mail that has
40;; not been processed before
41;;
42;; `spam-stat-buffer-is-non-spam' -- called in a buffer, that buffer
43;; is considered to be a new non-spam mail; use this for new mail that
44;; has not been processed before
45;;
46;; `spam-stat-buffer-change-to-spam' -- called in a buffer, that
47;; buffer is no longer considered to be normal mail but spam; use this
48;; to change the status of a mail that has already been processed as
49;; non-spam
50;;
51;; `spam-stat-buffer-change-to-non-spam' -- called in a buffer, that
52;; buffer is no longer considered to be spam but normal mail; use this
53;; to change the status of a mail that has already been processed as
54;; spam
55;;
56;; `spam-stat-save' -- save the hash table to the file; the filename
57;; used is stored in the variable `spam-stat-file'
58;;
59;; `spam-stat-load' -- load the hash table from a file; the filename
60;; used is stored in the variable `spam-stat-file'
61;;
62;; `spam-stat-score-word' -- return the spam score for a word
63;;
64;; `spam-stat-score-buffer' -- return the spam score for a buffer
65;;
66;; `spam-stat-split-fancy' -- for fancy mail splitting; add
67;; the rule (: spam-stat-split-fancy) to `nnmail-split-fancy'
68;;
69;; This requires the following in your ~/.gnus file:
70;;
71;; (require 'spam-stat)
72;; (spam-stat-load)
73
74;;; Testing:
75
76;; Typical test will involve calls to the following functions:
77;;
78;; Reset: (spam-stat-reset)
79;; Learn spam: (spam-stat-process-spam-directory "~/Mail/mail/spam")
80;; Learn non-spam: (spam-stat-process-non-spam-directory "~/Mail/mail/misc")
81;; Save table: (spam-stat-save)
82;; File size: (nth 7 (file-attributes spam-stat-file))
83;; Number of words: (hash-table-count spam-stat)
84;; Test spam: (spam-stat-test-directory "~/Mail/mail/spam")
85;; Test non-spam: (spam-stat-test-directory "~/Mail/mail/misc")
86;; Reduce table size: (spam-stat-reduce-size)
87;; Save table: (spam-stat-save)
88;; File size: (nth 7 (file-attributes spam-stat-file))
89;; Number of words: (hash-table-count spam-stat)
90;; Test spam: (spam-stat-test-directory "~/Mail/mail/spam")
91;; Test non-spam: (spam-stat-test-directory "~/Mail/mail/misc")
92
93;;; Dictionary Creation:
94
95;; Typically, you will filter away mailing lists etc. using specific
96;; rules in `nnmail-split-fancy'.  Somewhere among these rules, you
97;; will filter spam.  Here is how you would create your dictionary:
98
99;; Reset: (spam-stat-reset)
100;; Learn spam: (spam-stat-process-spam-directory "~/Mail/mail/spam")
101;; Learn non-spam: (spam-stat-process-non-spam-directory "~/Mail/mail/misc")
102;; Repeat for any other non-spam group you need...
103;; Reduce table size: (spam-stat-reduce-size)
104;; Save table: (spam-stat-save)
105
106;;; Todo:
107
108;; Speed it up.  Integrate with Gnus such that it uses spam and expiry
109;; marks to call the appropriate functions when leaving the summary
110;; buffer and saves the hash table when leaving Gnus.  More testing:
111;; More mails, disabling SpamAssassin, double checking algorithm, find
112;; improved algorithm.
113
114;;; Thanks:
115
116;; Ted Zlatanov <tzz@lifelogs.com>
117;; Jesper Harder <harder@myrealbox.com>
118;; Dan Schmidt <dfan@dfan.org>
119;; Lasse Rasinen <lrasinen@iki.fi>
120;; Milan Zamazal <pdm@zamazal.org>
121
122
123
124;;; Code:
125
126(defvar gnus-original-article-buffer)
127
128(defgroup spam-stat nil
129  "Statistical spam detection for Emacs.
130Use the functions to build a dictionary of words and their statistical
131distribution in spam and non-spam mails.  Then use a function to determine
132whether a buffer contains spam or not."
133  :version "22.1"
134  :group 'gnus)
135
136(defcustom spam-stat-file "~/.spam-stat.el"
137  "File used to save and load the dictionary.
138See `spam-stat-to-hash-table' for the format of the file."
139  :type 'file
140  :group 'spam-stat)
141
142(defcustom spam-stat-install-hooks t
143  "Whether spam-stat should install its hooks in Gnus.
144This is set to nil if you use spam-stat through spam.el."
145  :type 'boolean
146  :group 'spam-stat)
147
148(defcustom spam-stat-unknown-word-score 0.2
149  "The score to use for unknown words.
150Also used for words that don't appear often enough."
151  :type 'number
152  :group 'spam-stat)
153
154(defcustom spam-stat-max-word-length 15
155  "Only words shorter than this will be considered."
156  :type 'integer
157  :group 'spam-stat)
158
159(defcustom spam-stat-max-buffer-length 10240
160  "Only the beginning of buffers will be analyzed.
161This variable says how many characters this will be."
162  :type 'integer
163  :group 'spam-stat)
164
165(defcustom spam-stat-split-fancy-spam-group "mail.spam"
166  "Name of the group where spam should be stored, if
167`spam-stat-split-fancy' is used in fancy splitting rules.  Has no
168effect when spam-stat is invoked through spam.el."
169  :type 'string
170  :group 'spam-stat)
171
172(defcustom spam-stat-split-fancy-spam-threshhold 0.9
173  "Spam score threshhold in spam-stat-split-fancy."
174  :type 'number
175  :group 'spam-stat)
176
177(defvar spam-stat-syntax-table
178  (let ((table (copy-syntax-table text-mode-syntax-table)))
179    (modify-syntax-entry ?- "w" table)
180    (modify-syntax-entry ?_ "w" table)
181    (modify-syntax-entry ?. "w" table)
182    (modify-syntax-entry ?! "w" table)
183    (modify-syntax-entry ?? "w" table)
184    (modify-syntax-entry ?+ "w" table)
185    table)
186  "Syntax table used when processing mails for statistical analysis.
187The important part is which characters are word constituents.")
188
189(defvar spam-stat-dirty nil
190  "Whether the spam-stat database needs saving.")
191
192(defvar spam-stat-buffer nil
193  "Buffer to use for scoring while splitting.
194This is set by hooking into Gnus.")
195
196(defvar spam-stat-buffer-name " *spam stat buffer*"
197  "Name of the `spam-stat-buffer'.")
198
199;; Functions missing in Emacs 20
200
201(when (memq nil (mapcar 'fboundp
202			'(gethash hash-table-count make-hash-table
203				  mapc puthash)))
204  (require 'cl)
205  (unless (fboundp 'puthash)
206    ;; alias puthash is missing from Emacs 20 cl-extra.el
207    (defalias 'puthash 'cl-puthash)))
208
209(eval-when-compile
210  (unless (fboundp 'with-syntax-table)
211    ;; Imported from Emacs 21.2
212    (defmacro with-syntax-table (table &rest body) "\
213Evaluate BODY with syntax table of current buffer set to a copy of TABLE.
214The syntax table of the current buffer is saved, BODY is evaluated, and the
215saved table is restored, even in case of an abnormal exit.
216Value is what BODY returns."
217      (let ((old-table (make-symbol "table"))
218	    (old-buffer (make-symbol "buffer")))
219	`(let ((,old-table (syntax-table))
220	       (,old-buffer (current-buffer)))
221	   (unwind-protect
222	       (progn
223		 (set-syntax-table (copy-syntax-table ,table))
224		 ,@body)
225	     (save-current-buffer
226	       (set-buffer ,old-buffer)
227	       (set-syntax-table ,old-table))))))))
228
229;; Hooking into Gnus
230
231(defun spam-stat-store-current-buffer ()
232  "Store a copy of the current buffer in `spam-stat-buffer'."
233  (save-excursion
234    (let ((str (buffer-string)))
235      (set-buffer (get-buffer-create spam-stat-buffer-name))
236      (erase-buffer)
237      (insert str)
238      (setq spam-stat-buffer (current-buffer)))))
239
240(defun spam-stat-store-gnus-article-buffer ()
241  "Store a copy of the current article in `spam-stat-buffer'.
242This uses `gnus-article-buffer'."
243  (save-excursion
244    (set-buffer gnus-original-article-buffer)
245    (spam-stat-store-current-buffer)))
246
247;; Data -- not using defstruct in order to save space and time
248
249(defvar spam-stat (make-hash-table :test 'equal)
250  "Hash table used to store the statistics.
251Use `spam-stat-load' to load the file.
252Every word is used as a key in this table.  The value is a vector.
253Use `spam-stat-ngood', `spam-stat-nbad', `spam-stat-good',
254`spam-stat-bad', and `spam-stat-score' to access this vector.")
255
256(defvar spam-stat-ngood 0
257  "The number of good mails in the dictionary.")
258
259(defvar spam-stat-nbad 0
260  "The number of bad mails in the dictionary.")
261
262(defsubst spam-stat-good (entry)
263  "Return the number of times this word belongs to good mails."
264  (aref entry 0))
265
266(defsubst spam-stat-bad (entry)
267  "Return the number of times this word belongs to bad mails."
268  (aref entry 1))
269
270(defsubst spam-stat-score (entry)
271  "Set the score of this word."
272  (if entry
273      (aref entry 2)
274    spam-stat-unknown-word-score))
275
276(defsubst spam-stat-set-good (entry value)
277  "Set the number of times this word belongs to good mails."
278  (aset entry 0 value))
279
280(defsubst spam-stat-set-bad (entry value)
281  "Set the number of times this word belongs to bad mails."
282  (aset entry 1 value))
283
284(defsubst spam-stat-set-score (entry value)
285  "Set the score of this word."
286  (aset entry 2 value))
287
288(defsubst spam-stat-make-entry (good bad)
289  "Return a vector with the given properties."
290  (let ((entry (vector good bad nil)))
291    (spam-stat-set-score entry (spam-stat-compute-score entry))
292    entry))
293
294;; Computing
295
296(defun spam-stat-compute-score (entry)
297  "Compute the score of this word.  1.0 means spam."
298   ;; promote all numbers to floats for the divisions
299   (let* ((g (* 2.0 (spam-stat-good entry)))
300	  (b (float (spam-stat-bad entry))))
301     (cond ((< (+ g b) 5)
302	    .2)
303	   ((= 0 spam-stat-ngood)
304	    .99)
305	   ((= 0 spam-stat-nbad)
306	    .01)
307	   (t
308	    (max .01
309		 (min .99 (/ (/ b spam-stat-nbad)
310			     (+ (/ g spam-stat-ngood)
311				(/ b spam-stat-nbad)))))))))
312
313;; Parsing
314
315(defmacro with-spam-stat-max-buffer-size (&rest body)
316  "Narrows the buffer down to the first 4k characters, then evaluates BODY."
317  `(save-restriction
318     (when (> (- (point-max)
319		 (point-min))
320	      spam-stat-max-buffer-length)
321       (narrow-to-region (point-min)
322			 (+ (point-min) spam-stat-max-buffer-length)))
323     ,@body))
324
325(defun spam-stat-buffer-words ()
326  "Return a hash table of words and number of occurrences in the buffer."
327  (with-spam-stat-max-buffer-size
328   (with-syntax-table spam-stat-syntax-table
329     (goto-char (point-min))
330     (let ((result (make-hash-table :test 'equal))
331	   word count)
332       (while (re-search-forward "\\w+" nil t)
333	 (setq word (match-string-no-properties 0)
334	       count (1+ (gethash word result 0)))
335	 (when (< (length word) spam-stat-max-word-length)
336	   (puthash word count result)))
337       result))))
338
339(defun spam-stat-buffer-is-spam ()
340  "Consider current buffer to be a new spam mail."
341  (setq spam-stat-nbad (1+ spam-stat-nbad))
342  (maphash
343   (lambda (word count)
344     (let ((entry (gethash word spam-stat)))
345       (if entry
346	   (spam-stat-set-bad entry (+ count (spam-stat-bad entry)))
347	 (setq entry (spam-stat-make-entry 0 count)))
348       (spam-stat-set-score entry (spam-stat-compute-score entry))
349       (puthash word entry spam-stat)))
350   (spam-stat-buffer-words))
351  (setq spam-stat-dirty t))
352
353(defun spam-stat-buffer-is-non-spam ()
354  "Consider current buffer to be a new non-spam mail."
355  (setq spam-stat-ngood (1+ spam-stat-ngood))
356  (maphash
357   (lambda (word count)
358     (let ((entry (gethash word spam-stat)))
359       (if entry
360	   (spam-stat-set-good entry (+ count (spam-stat-good entry)))
361	 (setq entry (spam-stat-make-entry count 0)))
362       (spam-stat-set-score entry (spam-stat-compute-score entry))
363       (puthash word entry spam-stat)))
364   (spam-stat-buffer-words))
365  (setq spam-stat-dirty t))
366
367(defun spam-stat-buffer-change-to-spam ()
368  "Consider current buffer no longer normal mail but spam."
369  (setq spam-stat-nbad (1+ spam-stat-nbad)
370	spam-stat-ngood (1- spam-stat-ngood))
371  (maphash
372   (lambda (word count)
373     (let ((entry (gethash word spam-stat)))
374       (if (not entry)
375	   (error "This buffer has unknown words in it")
376	 (spam-stat-set-good entry (- (spam-stat-good entry) count))
377	 (spam-stat-set-bad entry (+ (spam-stat-bad entry) count))
378	 (spam-stat-set-score entry (spam-stat-compute-score entry))
379	 (puthash word entry spam-stat))))
380   (spam-stat-buffer-words))
381  (setq spam-stat-dirty t))
382
383(defun spam-stat-buffer-change-to-non-spam ()
384  "Consider current buffer no longer spam but normal mail."
385  (setq spam-stat-nbad (1- spam-stat-nbad)
386	spam-stat-ngood (1+ spam-stat-ngood))
387  (maphash
388   (lambda (word count)
389     (let ((entry (gethash word spam-stat)))
390       (if (not entry)
391	   (error "This buffer has unknown words in it")
392	 (spam-stat-set-good entry (+ (spam-stat-good entry) count))
393	 (spam-stat-set-bad entry (- (spam-stat-bad entry) count))
394	 (spam-stat-set-score entry (spam-stat-compute-score entry))
395	 (puthash word entry spam-stat))))
396   (spam-stat-buffer-words))
397  (setq spam-stat-dirty t))
398
399;; Saving and Loading
400
401(defun spam-stat-save (&optional force)
402  "Save the `spam-stat' hash table as lisp file.
403With a prefix argument save unconditionally."
404  (interactive "P")
405  (when (or force spam-stat-dirty)
406    (with-temp-buffer
407      (let ((standard-output (current-buffer))
408	    (font-lock-maximum-size 0))
409	(insert "(setq spam-stat-ngood "
410		(number-to-string spam-stat-ngood)
411		" spam-stat-nbad "
412		(number-to-string spam-stat-nbad)
413		" spam-stat (spam-stat-to-hash-table '(")
414	(maphash (lambda (word entry)
415		   (prin1 (list word
416				(spam-stat-good entry)
417				(spam-stat-bad entry))))
418		 spam-stat)
419	(insert ")))")
420	(write-file spam-stat-file)))
421    (setq spam-stat-dirty nil)))
422
423(defun spam-stat-load ()
424  "Read the `spam-stat' hash table from disk."
425  ;; TODO: maybe we should warn the user if spam-stat-dirty is t?
426  (load-file spam-stat-file)
427  (setq spam-stat-dirty nil))
428
429(defun spam-stat-to-hash-table (entries)
430  "Turn list ENTRIES into a hash table and store as `spam-stat'.
431Every element in ENTRIES has the form \(WORD GOOD BAD) where WORD is
432the word string, NGOOD is the number of good mails it has appeared in,
433NBAD is the number of bad mails it has appeared in, GOOD is the number
434of times it appeared in good mails, and BAD is the number of times it
435has appeared in bad mails."
436  (let ((table (make-hash-table :test 'equal)))
437    (mapc (lambda (l)
438	    (puthash (car l)
439		     (spam-stat-make-entry (nth 1 l) (nth 2 l))
440		     table))
441	  entries)
442    table))
443
444(defun spam-stat-reset ()
445  "Reset `spam-stat' to an empty hash-table.
446This deletes all the statistics."
447  (interactive)
448  (setq spam-stat (make-hash-table :test 'equal)
449	spam-stat-ngood 0
450	spam-stat-nbad 0)
451  (setq spam-stat-dirty t))
452
453;; Scoring buffers
454
455(defvar spam-stat-score-data nil
456  "Raw data used in the last run of `spam-stat-score-buffer'.")
457
458(defsubst spam-stat-score-word (word)
459  "Return score for WORD.
460The default score for unknown words is stored in
461`spam-stat-unknown-word-score'."
462  (spam-stat-score (gethash word spam-stat)))
463
464(defun spam-stat-buffer-words-with-scores ()
465  "Process current buffer, return the 15 most conspicuous words.
466These are the words whose spam-stat differs the most from 0.5.
467The list returned contains elements of the form \(WORD SCORE DIFF),
468where DIFF is the difference between SCORE and 0.5."
469  (with-spam-stat-max-buffer-size
470   (with-syntax-table spam-stat-syntax-table
471     (let (result word score)
472       (maphash	(lambda (word ignore)
473		  (setq score (spam-stat-score-word word)
474			result (cons (list word score (abs (- score 0.5)))
475				     result)))
476		(spam-stat-buffer-words))
477       (setq result (sort result (lambda (a b) (< (nth 2 b) (nth 2 a)))))
478       (setcdr (nthcdr 14 result) nil)
479       result))))
480
481(defun spam-stat-score-buffer ()
482  "Return a score describing the spam-probability for this buffer."
483  (setq spam-stat-score-data (spam-stat-buffer-words-with-scores))
484  (let* ((probs (mapcar (lambda (e) (cadr e)) spam-stat-score-data))
485	 (prod (apply #'* probs)))
486    (/ prod (+ prod (apply #'* (mapcar #'(lambda (x) (- 1 x))
487				       probs))))))
488
489(defun spam-stat-split-fancy ()
490  "Return the name of the spam group if the current mail is spam.
491Use this function on `nnmail-split-fancy'.  If you are interested in
492the raw data used for the last run of `spam-stat-score-buffer',
493check the variable `spam-stat-score-data'."
494  (condition-case var
495      (progn
496	(set-buffer spam-stat-buffer)
497	(goto-char (point-min))
498	(when (> (spam-stat-score-buffer) spam-stat-split-fancy-spam-threshhold)
499	  (when (boundp 'nnmail-split-trace)
500	    (mapc (lambda (entry)
501		    (push entry nnmail-split-trace))
502		  spam-stat-score-data))
503	  spam-stat-split-fancy-spam-group))
504    (error (message "Error in spam-stat-split-fancy: %S" var)
505	   nil)))
506
507;; Testing
508
509(defun spam-stat-process-directory (dir func)
510  "Process all the regular files in directory DIR using function FUNC."
511  (let* ((files (directory-files dir t "^[^.]"))
512	 (max (/ (length files) 100.0))
513	 (count 0))
514    (with-temp-buffer
515      (dolist (f files)
516	(when (and (file-readable-p f)
517		   (file-regular-p f)
518                   (> (nth 7 (file-attributes f)) 0))
519	  (setq count (1+ count))
520	  (message "Reading %s: %.2f%%" dir (/ count max))
521	  (insert-file-contents f)
522	  (funcall func)
523	  (erase-buffer))))))
524
525(defun spam-stat-process-spam-directory (dir)
526  "Process all the regular files in directory DIR as spam."
527  (interactive "D")
528  (spam-stat-process-directory dir 'spam-stat-buffer-is-spam))
529
530(defun spam-stat-process-non-spam-directory (dir)
531  "Process all the regular files in directory DIR as non-spam."
532  (interactive "D")
533  (spam-stat-process-directory dir 'spam-stat-buffer-is-non-spam))
534
535(defun spam-stat-count ()
536  "Return size of `spam-stat'."
537  (interactive)
538  (hash-table-count spam-stat))
539
540(defun spam-stat-test-directory (dir)
541  "Test all the regular files in directory DIR for spam.
542If the result is 1.0, then all files are considered spam.
543If the result is 0.0, non of the files is considered spam.
544You can use this to determine error rates."
545  (interactive "D")
546  (let* ((files (directory-files dir t "^[^.]"))
547	 (total (length files))
548	 (score 0.0); float
549	 (max (/ total 100.0)); float
550	 (count 0))
551    (with-temp-buffer
552      (dolist (f files)
553	(when (and (file-readable-p f)
554		   (file-regular-p f)
555                   (> (nth 7 (file-attributes f)) 0))
556	  (setq count (1+ count))
557	  (message "Reading %.2f%%, score %.2f%%"
558		   (/ count max) (/ score count))
559	  (insert-file-contents f)
560	  (when (> (spam-stat-score-buffer) 0.9)
561	    (setq score (1+ score)))
562	  (erase-buffer))))
563    (message "Final score: %d / %d = %f" score total (/ score total))))
564
565;; Shrinking the dictionary
566
567(defun spam-stat-reduce-size (&optional count)
568  "Reduce the size of `spam-stat'.
569This removes all words that occur less than COUNT from the dictionary.
570COUNT defaults to 5"
571  (interactive)
572  (setq count (or count 5))
573  (maphash (lambda (key entry)
574	     (when (< (+ (spam-stat-good entry)
575			 (spam-stat-bad entry))
576		      count)
577	       (remhash key spam-stat)))
578	   spam-stat)
579  (setq spam-stat-dirty t))
580
581(defun spam-stat-install-hooks-function ()
582  "Install the spam-stat function hooks"
583  (interactive)
584  (add-hook 'nnmail-prepare-incoming-message-hook
585	    'spam-stat-store-current-buffer)
586  (add-hook 'gnus-select-article-hook
587	    'spam-stat-store-gnus-article-buffer))
588
589(when spam-stat-install-hooks
590  (spam-stat-install-hooks-function))
591
592(defun spam-stat-unload-hook ()
593  "Uninstall the spam-stat function hooks"
594  (interactive)
595  (remove-hook 'nnmail-prepare-incoming-message-hook
596	       'spam-stat-store-current-buffer)
597  (remove-hook 'gnus-select-article-hook
598	       'spam-stat-store-gnus-article-buffer))
599
600(add-hook 'spam-stat-unload-hook 'spam-stat-unload-hook)
601
602(provide 'spam-stat)
603
604;;; arch-tag: ff1d2200-8ddb-42fb-bb7b-1b5e20448554
605;;; spam-stat.el ends here
606