source: npl/mailserver/dspam/dspam-3.10.2/src/tools/dspam_stats.c @ c5c522c

gcc484ntopperl-5.22
Last change on this file since c5c522c was c5c522c, checked in by Edwin Eefting <edwin@datux.nl>, 8 years ago

initial commit, transferred from cleaned syn3 svn tree
  • Property mode set to 100644
File size: 13.9 KB
Line 
1/* $Id: dspam_stats.c,v 1.36 2011/06/28 00:13:48 sbajic Exp $ */
2
3/*
4 DSPAM
5 COPYRIGHT (C) 2002-2012 DSPAM PROJECT
6
7 This program is free software: you can redistribute it and/or modify
8 it under the terms of the GNU Affero General Public License as
9 published by the Free Software Foundation, either version 3 of the
10 License, or (at your option) any later version.
11
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 GNU Affero General Public License for more details.
16
17 You should have received a copy of the GNU Affero General Public License
18 along with this program.  If not, see <http://www.gnu.org/licenses/>.
19
20*/
21
22#ifdef HAVE_CONFIG_H
23#include <auto-config.h>
24#endif
25
26#include <stdio.h>
27#include <stdlib.h>
28#include <string.h>
29#include <fcntl.h>
30#include <ctype.h>
31#include <sys/types.h>
32#include <sys/stat.h>
33#include <errno.h>
34#include <signal.h>
35#ifndef _WIN32
36#include <unistd.h>
37#include <dirent.h>
38#endif
39#include "config.h"
40
41#include "libdspam.h"
42#include "read_config.h"
43#include "config_api.h"
44#include "language.h"
45#include "util.h"
46
47#define TSYNTAX "syntax: dspam_stats [-h]\|[--profile=PROFILE] [-HrsSt] [user [user...]]"
48
49#ifdef _WIN32
50/* no trusted users under Windows */
51#undef TRUSTED_USER_SECURITY
52#endif
53
54DSPAM_CTX *open_ctx, *open_mtx;
55int opt_humanfriendly;
56int opt_reset;
57int opt_snapshot;
58int opt_stats;
59int opt_total;
60
61int stat_user (const char *username, struct _ds_spam_totals *totals);
62int process_all_users (struct _ds_spam_totals *totals);
63void dieout (int signal);
64void usage (void);
65
66int
67main (int argc, char **argv)
68{
69  int ch, i, users = 0;
70#ifndef HAVE_GETOPT
71  int optind = 1;
72#endif
73  struct _ds_spam_totals totals;
74
75#ifdef TRUSTED_USER_SECURITY
76  struct passwd *p = getpwuid (getuid ());
77  int trusted = 0;                                                                               
78#endif
79
80
81 memset(&totals, 0, sizeof(struct _ds_spam_totals));
82
83 /* Read dspam.conf */
84                                                                               
85  agent_config = read_config(NULL);
86  if (!agent_config) {
87    LOG(LOG_ERR, ERR_AGENT_READ_CONFIG);
88    fprintf (stderr, ERR_AGENT_READ_CONFIG "\n");
89    exit(EXIT_FAILURE);
90  }
91                                                                               
92  if (!_ds_read_attribute(agent_config, "Home")) {
93    LOG(LOG_ERR, ERR_AGENT_DSPAM_HOME);
94    fprintf (stderr, ERR_AGENT_DSPAM_HOME "\n");
95    _ds_destroy_config(agent_config);
96    exit(EXIT_FAILURE);
97  }
98                                                                               
99  if (libdspam_init(_ds_read_attribute(agent_config, "StorageDriver")) != 0) {
100    LOG(LOG_ERR, ERR_DRV_INIT);
101    fprintf (stderr, ERR_DRV_INIT "\n");
102    _ds_destroy_config(agent_config);
103    exit(EXIT_FAILURE);
104  }
105
106#ifdef TRUSTED_USER_SECURITY
107  if (_ds_match_attribute(agent_config, "Trust", p->pw_name) || !p->pw_uid) {
108    trusted = 1;
109  }
110#endif
111
112  for(i=0;i<argc;i++) {
113    if (!strncmp (argv[i], "--profile=", 10))
114    {
115#ifdef TRUSTED_USER_SECURITY
116      if (!trusted) {
117        LOG(LOG_ERR, ERR_TRUSTED_PRIV, "--profile", p->pw_uid, p->pw_name);
118        fprintf (stderr, ERR_TRUSTED_PRIV "\n", "--profile", p->pw_uid, p->pw_name);
119        _ds_destroy_config(agent_config);
120        goto BAIL;
121      }
122#endif
123      if (!_ds_match_attribute(agent_config, "Profile", argv[i]+10)) {
124        LOG(LOG_ERR, ERR_AGENT_NO_SUCH_PROFILE, argv[i]+10);
125        fprintf (stderr, ERR_AGENT_NO_SUCH_PROFILE "\n", argv[i]+10);
126        _ds_destroy_config(agent_config);
127        goto BAIL;
128      } else {
129        _ds_overwrite_attribute(agent_config, "DefaultProfile", argv[i]+10);
130      }
131      break;
132    }
133  }
134
135  open_ctx = open_mtx = NULL;
136
137  signal (SIGINT, dieout);
138#ifndef _WIN32
139  signal (SIGPIPE, dieout);
140#endif
141  signal (SIGTERM, dieout);
142
143  dspam_init_driver (NULL);
144
145  /* Process command line */
146  ch = opt_humanfriendly = 0;
147  opt_reset = opt_snapshot = opt_stats = opt_total =  0;
148
149#ifndef HAVE_GETOPT
150  while ( argv[optind] &&
151            argv[optind][0] == '-' &&
152              (ch = argv[optind][1]) &&
153                argv[optind][2] == '\0' )
154#else
155  while((ch = getopt(argc, argv, "hHrsS")) != -1)
156#endif
157  {
158    switch(ch) {
159      case 'h':
160        /* print help, and then exit. usage exits for us */
161        usage();
162        break;
163      case 'H':
164        opt_humanfriendly = 1;
165        break;
166      case 'r':
167        opt_reset = 1;
168        break;
169      case 's':
170        opt_snapshot = 1;
171        break;
172      case 'S':
173        opt_stats = 1;
174        break;
175      case 't':
176        opt_total = 1;
177        break;
178
179#ifndef HAVE_GETOPT
180      default:
181        fprintf(stderr, "%s: unknown option \"%s\".\n",
182                argv[0], argv[optind] + 1);
183        usage();
184#endif
185    }
186#ifndef HAVE_GETOPT
187    optind++;
188#endif
189  }
190#ifndef HAVE_GETOPT
191  /* reset our option array and index to where we are after getopt */
192  argv += optind;
193  argc -= optind;
194#endif
195
196  /* process arguments */
197  for (i=0; i < argc; i++)
198  {
199      if (argv[i] && strncmp(argv[i], "--", 2)) {
200#ifdef TRUSTED_USER_SECURITY
201        if ( !trusted && strcmp(argv[i], p->pw_name) )
202        {
203          fprintf(stderr, ERR_TRUSTED_MODE "\n");
204          _ds_destroy_config(agent_config);
205          goto BAIL;
206        }
207#endif
208        stat_user(argv[i], &totals);
209        users++;
210      }
211  }
212
213  if (!users)
214  {
215#ifdef TRUSTED_USER_SECURITY
216    if ( !trusted )
217    {
218      fprintf(stderr, ERR_TRUSTED_MODE "\n");
219      _ds_destroy_config(agent_config);
220      goto BAIL;
221    }
222#endif
223
224    process_all_users (&totals);
225  }
226
227  if (opt_total)
228    stat_user(NULL, &totals);
229  dspam_shutdown_driver (NULL);
230  _ds_destroy_config(agent_config);
231  libdspam_shutdown();
232  exit (EXIT_SUCCESS);
233
234BAIL:
235  libdspam_shutdown();
236  exit(EXIT_FAILURE);
237}
238
239int
240process_all_users (struct _ds_spam_totals *totals)
241{
242  DSPAM_CTX *CTX;
243  char *user;
244
245  CTX = dspam_create (NULL, NULL, _ds_read_attribute(agent_config, "Home"), DSM_TOOLS, 0);
246  open_ctx = CTX;
247  if (CTX == NULL)
248  {
249    fprintf (stderr, "Could not initialize context: %s\n", strerror (errno));
250    return EFAILURE;
251  }
252
253  set_libdspam_attributes(CTX);
254  if (dspam_attach(CTX, NULL)) {
255    LOG (LOG_WARNING, "unable to attach dspam context");
256    fprintf (stderr, "Unable to attach DSPAM context\n");
257    dspam_destroy(CTX);
258    return EFAILURE;
259  }
260
261  user = _ds_get_nextuser (CTX);
262  while (user != NULL)
263  {
264    stat_user (user, totals);
265    user = _ds_get_nextuser (CTX);
266  }
267
268  dspam_destroy (CTX);
269  open_ctx = NULL;
270  return 0;
271}
272
273int
274stat_user (const char *username, struct _ds_spam_totals *totals)
275{
276  DSPAM_CTX *MTX = NULL;
277  long total_spam, total_innocent, spam_misclassified, innocent_misclassified, spam_corpusfed, innocent_corpusfed, all_spam, all_innocent;
278  char filename[MAX_FILENAME_LENGTH];
279  FILE *file;
280  struct _ds_spam_totals *tptr;
281
282  if (username) {
283    MTX = dspam_create (username, NULL, _ds_read_attribute(agent_config, "Home"), DSM_CLASSIFY, 0);
284    open_mtx = MTX;
285    if (MTX == NULL)
286    {
287      fprintf (stderr, "Could not init context: %s\n", strerror (errno));
288      return EUNKNOWN;
289    }
290    set_libdspam_attributes(MTX);
291    if (dspam_attach(MTX, NULL)) {
292      LOG (LOG_WARNING, "unable to attach dspam context");
293      fprintf (stderr, "Unable to attach DSPAM context\n");
294      return EUNKNOWN;
295    }
296    tptr = &MTX->totals;
297  } else {
298    tptr = totals;
299  }
300
301  /* Convenience variables. Compiling with optimization will cause this to
302     have 0 slowdown, as it is essentially dead code */
303  total_spam =
304      MAX(0, (tptr->spam_learned + tptr->spam_classified) -
305        (tptr->spam_misclassified + tptr->spam_corpusfed));
306  total_innocent =
307      MAX(0, (tptr->innocent_learned + tptr->innocent_classified) -
308        (tptr->innocent_misclassified + tptr->innocent_corpusfed));
309  spam_misclassified     = tptr->spam_misclassified;
310  innocent_misclassified = tptr->innocent_misclassified;
311  spam_corpusfed         = tptr->spam_corpusfed;
312  innocent_corpusfed     = tptr->innocent_corpusfed;
313
314  if (MTX) {
315    totals->spam_learned += MTX->totals.spam_learned;
316    totals->innocent_learned += MTX->totals.innocent_learned;
317    totals->spam_misclassified += MTX->totals.spam_misclassified;
318    totals->innocent_misclassified += MTX->totals.innocent_misclassified;
319    totals->spam_corpusfed += MTX->totals.spam_corpusfed;
320    totals->innocent_corpusfed += MTX->totals.innocent_corpusfed;
321  }
322
323  /* Subtract the snapshot from the current totals to get stats "since last
324     reset" for the user */
325
326  if (opt_snapshot && username) {
327    long s_total_spam, s_total_innocent, s_spam_misclassified,
328         s_innocent_misclassified, s_spam_corpusfed, s_innocent_corpusfed;
329
330    _ds_userdir_path(filename, _ds_read_attribute(agent_config, "Home"),
331                     username, "rstats");
332    _ds_prepare_path_for (filename);
333
334    file = fopen (filename, "r");
335    if (file != NULL) {
336      if (fscanf(file, "%ld,%ld,%ld,%ld,%ld,%ld",
337                 &s_total_spam,
338                 &s_total_innocent,
339                 &s_spam_misclassified,
340                 &s_innocent_misclassified,
341                 &s_spam_corpusfed,
342                 &s_innocent_corpusfed)==6) {
343        total_spam             -= s_total_spam;
344        total_innocent         -= s_total_innocent;
345        spam_misclassified     -= s_spam_misclassified;
346        innocent_misclassified -= s_innocent_misclassified;
347        spam_corpusfed         -= s_spam_corpusfed;
348        innocent_corpusfed     -= s_innocent_corpusfed;
349      }
350      fclose(file);
351    }
352  }
353
354  all_spam = total_spam + spam_misclassified,
355  all_innocent = total_innocent + innocent_misclassified;
356
357  if (opt_humanfriendly)
358  {
359    printf("%s:\n\
360        \tTP True Positives:                %6ld\n\
361        \tTN True Negatives:                %6ld\n\
362        \tFP False Positives:               %6ld\n\
363        \tFN False Negatives:               %6ld\n\
364        \tSC Spam Corpusfed:                %6ld\n\
365        \tNC Nonspam Corpusfed:             %6ld\n\
366        \tTL Training Left:                 %6ld\n\
367        \tSHR Spam Hit Rate               % 7.2f%%\n\
368        \tHSR Ham Strike Rate:            % 7.2f%%\n\
369        \tPPV Positive predictive value:  % 7.2f%%\n\
370        \tOCA Overall Accuracy:           % 7.2f%%\n\
371        \n",
372            (username) ? username : "TOTAL",
373            total_spam, total_innocent,
374            innocent_misclassified, spam_misclassified,
375            spam_corpusfed, innocent_corpusfed,
376            MAX(0, 2500 - (tptr->innocent_learned +
377                           tptr->innocent_classified)),
378       (all_spam) ?
379          (100.0-((float)spam_misclassified / (float)all_spam )*100.0)
380          : 100.0,
381        (all_innocent) ?
382          100-(100.0-((float)innocent_misclassified / (float)all_innocent )*100.0)
383          : 100.0,
384        (total_spam + innocent_misclassified) ?
385          100-(100.0-((float)total_spam /
386                  (float)(total_spam + innocent_misclassified))*100)
387          : 100.0,
388        (all_spam + all_innocent) ?
389          (100.0-(((float)spam_misclassified +(float)innocent_misclassified) /
390                   (float)(all_spam + all_innocent))*100.0)
391          : 100.0);
392  }
393  else
394  {
395#ifdef LONG_USERNAMES
396    printf ("%s\n    TP:%6ld TN:%6ld FP:%6ld FN:%6ld SC:%6ld NC:%6ld\n",
397#else
398    printf ("%-16s  TP:%6ld TN:%6ld FP:%6ld FN:%6ld SC:%6ld NC:%6ld\n",
399#endif
400            (username) ? username : "TOTAL",
401            total_spam, total_innocent,
402            innocent_misclassified, spam_misclassified,
403            spam_corpusfed, innocent_corpusfed);
404
405    if (opt_stats)
406      printf (
407#ifdef LONG_USERNAMES
408"    "
409#else
410"                  "
411#endif
412              "SHR: % 7.2f%%       HSR: % 7.2f%%       OCA: % 7.2f%%\n",
413        (all_spam) ?
414          (100.0-((float)spam_misclassified / (float)all_spam )*100.0)
415          : 100.0,
416        (all_innocent) ?
417          100.0-
418          (100.0-((float)innocent_misclassified / (float)all_innocent )*100.0)
419          : 0.0,
420        (all_spam + all_innocent) ?
421          (100.0-(((float)spam_misclassified +(float)innocent_misclassified) /
422                   (float)(all_spam + all_innocent))*100.0)
423          : 100.0);
424  }
425
426  if (opt_reset && username) {
427    _ds_userdir_path(filename, _ds_read_attribute(agent_config, "Home"),
428                     username, "rstats");
429    _ds_prepare_path_for (filename);
430    file = fopen (filename, "w");
431    if (file == NULL)
432    {
433      LOG(LOG_ERR, ERR_IO_FILE_WRITE, filename, strerror (errno));
434      if (MTX)
435        dspam_destroy (MTX);
436      open_mtx = NULL;
437      return EFILE;
438    }
439
440    fprintf (file, "%ld,%ld,%ld,%ld,%ld,%ld\n",
441     MAX(0,(tptr->spam_learned + tptr->spam_classified) -
442       (tptr->spam_misclassified + tptr->spam_corpusfed)),
443     MAX(0,(tptr->innocent_learned + tptr->innocent_classified) -
444       (tptr->innocent_misclassified + tptr->innocent_corpusfed)),
445     tptr->spam_misclassified,
446     tptr->innocent_misclassified,
447     tptr->spam_corpusfed,
448     tptr->innocent_corpusfed);
449    fclose(file);
450  }
451
452  if (MTX)
453    dspam_destroy (MTX);
454  open_mtx = NULL;
455  return 0;
456}
457
458void
459dieout (int signal)
460{
461  signal = signal; /* Keep compile happy */
462  fprintf (stderr, "terminated.\n");
463  if (open_ctx != NULL)
464    dspam_destroy (open_ctx);
465  if (open_mtx != NULL)
466    dspam_destroy (open_mtx);
467  _ds_destroy_config(agent_config);
468  exit (EXIT_SUCCESS);
469}
470
471void
472usage (void)
473{
474  (void)fprintf (stderr,
475    "usage: dspam_stats [-h]|[--profile=PROFILE] [-HrsSt] [user [user...]]\n\
476      \tPrint dspam statistics for users.\n\
477      \tIf no users are specified, stats for all users are printed.\n\
478      \t-h: print this message\n\
479      \t-H: print stats in \"human friendly\" format\n\
480      \t-r: Resets the current snapshot\n\
481      \t-s: Displays stats since last snapshot (instead of since epoch)\n\
482      \t-S: Displays accuracy percentages in addition to stats\n\
483      \t-t: Displays a total of all statistics displayed\n");
484  _ds_destroy_config(agent_config);
485  exit(EXIT_FAILURE);
486}
Note: See TracBrowser for help on using the repository browser.