src/src/spam.c

   1 /*************************************************
   2 *     Exim - an Internet mail transport agent    *
   3 *************************************************/
   4
   5 /* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-???? */
   6 /* License: GPL */
   7
   8 /* Code for calling spamassassin's spamd. Called from acl.c. */
   9
  10 #include "exim.h"
  11 #ifdef WITH_CONTENT_SCAN
  12 #include "spam.h"
  13
  14 uschar spam_score_buffer[16];
  15 uschar spam_score_int_buffer[16];
  16 uschar spam_bar_buffer[128];
  17 uschar spam_action_buffer[32];
  18 uschar spam_report_buffer[32600];
  19 uschar prev_user_name[128] = "";
  20 int spam_ok = 0;
  21 int spam_rc = 0;
  22 uschar *prev_spamd_address_work = NULL;
  23
  24 static int timeout_sec;
  25 static const uschar * loglabel = US"spam acl condition:";
  26
  27
  28 static int
  29 spamd_param_init(spamd_address_container *spamd)
  30 {
  31 /* default spamd server weight, time and backup value */
  32 spamd->weight = SPAMD_WEIGHT;
  33 spamd->is_failed = FALSE;
  34 spamd->is_backup = FALSE;
  35 return 0;
  36 }
  37
  38
  39 static int
  40 spamd_param(const uschar *param, spamd_address_container *spamd)
  41 {
  42 static int timesinceday = -1;
  43 const uschar * s;
  44
  45 /* check backup parameter */
  46 if (Ustrcmp(param, "backup") == 0)
  47   {
  48   spamd->is_backup = TRUE;
  49   return 0; /* OK */
  50   }
  51
  52 /*XXX more clever parsing could discard embedded spaces? */
  53
  54 /* check weight parameter */
  55 if (sscanf(param, "weight=%u", &spamd->weight))
  56   {
  57   if (spamd->weight == 0) /* this server disabled: skip it */
  58     return 1;
  59   return 0; /* OK */
  60   }
  61
  62 /* check time parameter */
  63 if (Ustrncmp(param, "time=", 5) == 0)
  64   {
  65   unsigned int start_h = 0, start_m = 0, start_s = 0;
  66   unsigned int end_h = 24, end_m = 0, end_s = 0;
  67   unsigned int time_start, time_end;
  68   const uschar * end_string;
  69
  70   s = param+5;
  71   if ((end_string = Ustrchr(s, '-')))
  72     {
  73     end_string++;
  74     if (  sscanf(CS end_string, "%u.%u.%u", &end_h,   &end_m,   &end_s)   == 0
  75        || sscanf(CS s,          "%u.%u.%u", &start_h, &start_m, &start_s) == 0
  76        )
  77       {
  78       log_write(0, LOG_MAIN,
  79         "%s warning - invalid spamd time value: '%s'", loglabel, s);
  80       return -1; /* syntax error */
  81       }
  82     }
  83   else
  84     {
  85     log_write(0, LOG_MAIN,
  86       "%s warning - invalid spamd time value: '%s'", loglabel, s);
  87     return -1; /* syntax error */
  88     }
  89
  90   if (timesinceday < 0)
  91     {
  92     time_t now = time(NULL);
  93     struct tm *tmp = localtime(&now);
  94     timesinceday = tmp->tm_hour*3600 + tmp->tm_min*60 + tmp->tm_sec;
  95     }
  96
  97   time_start = start_h*3600 + start_m*60 + start_s;
  98   time_end = end_h*3600 + end_m*60 + end_s;
  99
 100   if (timesinceday < time_start || timesinceday >= time_end)
 101     return 1; /* skip spamd server */
 102
 103   return 0; /* OK */
 104   }
 105
 106 if (Ustrcmp(param, "variant=rspamd") == 0)
 107   {
 108   spamd->is_rspamd = TRUE;
 109   return 0;
 110   }
 111
 112 if (Ustrncmp(param, "tmo=", 4) == 0)
 113   {
 114   int sec = readconf_readtime((s = param+4), '\0', FALSE);
 115   if (sec < 0)
 116     {
 117     log_write(0, LOG_MAIN,
 118       "%s warning - invalid spamd timeout value: '%s'", loglabel, s);
 119     return -1; /* syntax error */
 120     }
 121   timeout_sec = sec;
 122   return 0;
 123   }
 124
 125 log_write(0, LOG_MAIN, "%s warning - invalid spamd parameter: '%s'",
 126   loglabel, param);
 127 return -1; /* syntax error */
 128 }
 129
 130
 131 static int
 132 spamd_get_server(spamd_address_container **spamds, int num_servers)
 133 {
 134 unsigned int i;
 135 long rnd, weights = 0;
 136 static BOOL srandomed = 0;
 137 BOOL usebackup = FALSE;
 138
 139 for (;;)
 140   {
 141   /* seedup, if we have only 1 server */
 142   if (num_servers == 1)
 143     return (spamds[0]->is_failed ? -1 : 0);
 144
 145   /* init ranmod */
 146   if (!srandomed)
 147     {
 148     struct timeval tv;
 149     gettimeofday(&tv, NULL);
 150     srandom((unsigned int)(tv.tv_usec/1000));
 151     srandomed = TRUE;
 152     }
 153
 154   /* get sum of all weights */
 155   for (i = 0; i < num_servers; i++)
 156     if (!spamds[i]->is_failed && spamds[i]->is_backup == usebackup)
 157       weights += spamds[i]->weight;
 158
 159   if (weights != 0)
 160     break;
 161   if (usebackup)        /* all servers failed (backups too) */
 162     return -1;
 163   usebackup = TRUE;
 164   }
 165
 166 rnd = random() % weights;
 167
 168 for (i = 0; i < num_servers; i++)
 169   if (!spamds[i]->is_failed && spamds[i]->is_backup == usebackup)
 170     if ((rnd -= spamds[i]->weight) < 0)
 171       return i;
 172
 173 log_write(0, LOG_MAIN|LOG_PANIC,
 174   "%s unknown error (memory/cpu corruption?)", loglabel);
 175 return -1;
 176 }
 177
 178
 179 int
 180 spam(const uschar **listptr)
 181 {
 182 int sep = 0;
 183 const uschar *list = *listptr;
 184 uschar *user_name;
 185 uschar user_name_buffer[128];
 186 unsigned long mbox_size;
 187 FILE *mbox_file;
 188 int spamd_sock = -1;
 189 uschar spamd_buffer[32600];
 190 int i, j, offset, result;
 191 BOOL is_rspamd;
 192 uschar spamd_version[8];
 193 uschar spamd_short_result[8];
 194 uschar spamd_score_char;
 195 double spamd_threshold, spamd_score, spamd_reject_score;
 196 int spamd_report_offset;
 197 uschar *p,*q;
 198 int override = 0;
 199 time_t start;
 200 size_t read, wrote;
 201 struct sockaddr_un server;
 202 #ifndef NO_POLL_H
 203 struct pollfd pollfd;
 204 #else                               /* Patch posted by Erik ? for OS X */
 205 struct timeval select_tv;         /* and applied by PH */
 206 fd_set select_fd;
 207 #endif
 208 uschar *spamd_address_work;
 209
 210 /* stop compiler warning */
 211 result = 0;
 212
 213 /* find the username from the option list */
 214 if ((user_name = string_nextinlist(&list, &sep,
 215                                    user_name_buffer,
 216                                    sizeof(user_name_buffer))) == NULL)
 217   {
 218   /* no username given, this means no scanning should be done */
 219   return FAIL;
 220   }
 221
 222 /* if username is "0" or "false", do not scan */
 223 if ( (Ustrcmp(user_name,"0") == 0) ||
 224      (strcmpic(user_name,US"false") == 0) )
 225   return FAIL;
 226
 227 timeout_sec = SPAMD_TIMEOUT;
 228
 229 /* if there is an additional option, check if it is "true" */
 230 if (strcmpic(list,US"true") == 0)
 231   /* in that case, always return true later */
 232   override = 1;
 233
 234 /* expand spamd_address if needed */
 235 if (*spamd_address == '$')
 236   {
 237   spamd_address_work = expand_string(spamd_address);
 238   if (spamd_address_work == NULL)
 239     {
 240     log_write(0, LOG_MAIN|LOG_PANIC,
 241       "%s spamd_address starts with $, but expansion failed: %s",
 242       loglabel, expand_string_message);
 243     return DEFER;
 244     }
 245   }
 246 else
 247   spamd_address_work = spamd_address;
 248
 249 HDEBUG(D_acl) debug_printf("spamd: addrlist '%s'\n", spamd_address_work);
 250
 251 /* check if previous spamd_address was expanded and has changed. dump cached results if so */
 252 if (  spam_ok
 253    && prev_spamd_address_work != NULL
 254    && Ustrcmp(prev_spamd_address_work, spamd_address_work) != 0
 255    )
 256   spam_ok = 0;
 257
 258 /* if we scanned for this username last time, just return */
 259 if (spam_ok && Ustrcmp(prev_user_name, user_name) == 0)
 260   return override ? OK : spam_rc;
 261
 262 /* make sure the eml mbox file is spooled up */
 263 mbox_file = spool_mbox(&mbox_size, NULL);
 264
 265 if (mbox_file == NULL)
 266   {
 267   /* error while spooling */
 268   log_write(0, LOG_MAIN|LOG_PANIC,
 269          "%s error while creating mbox spool file", loglabel);
 270   return DEFER;
 271   }
 272
 273 start = time(NULL);
 274
 275   {
 276   int num_servers = 0;
 277   int current_server;
 278   uschar *address;
 279   const uschar *spamd_address_list_ptr = spamd_address_work;
 280   spamd_address_container * spamd_address_vector[32];
 281   spamd_address_container * sd;
 282
 283   /* Check how many spamd servers we have
 284      and register their addresses */
 285   while ((address = string_nextinlist(&spamd_address_list_ptr, &sep,
 286                                       NULL, 0)) != NULL)
 287     {
 288     const uschar * sublist;
 289     int sublist_sep = -(int)' ';        /* default space-sep */
 290     unsigned args;
 291     uschar * s;
 292
 293     HDEBUG(D_acl) debug_printf("spamd: addr entry '%s'\n", address);
 294     sd = (spamd_address_container *)store_get(sizeof(spamd_address_container));
 295
 296     for (sublist = address, args = 0, spamd_param_init(sd);
 297          s = string_nextinlist(&sublist, &sublist_sep, NULL, 0);
 298          args++
 299          )
 300       {
 301         HDEBUG(D_acl) debug_printf("spamd: addr parm '%s'\n", s);
 302         switch (args)
 303         {
 304         case 0:   sd->hostspec = s;
 305                   if (*s == '/') args++;        /* local; no port */
 306                   break;
 307         case 1:   sd->hostspec = string_sprintf("%s %s", sd->hostspec, s);
 308                   break;
 309         default:  spamd_param(s, sd);
 310                   break;
 311         }
 312       }
 313     if (args < 2)
 314       {
 315       log_write(0, LOG_MAIN,
 316         "%s warning - invalid spamd address: '%s'", loglabel, address);
 317       continue;
 318       }
 319
 320     spamd_address_vector[num_servers] = sd;
 321     if (++num_servers > 31)
 322       break;
 323     }
 324
 325   /* check if we have at least one server */
 326   if (!num_servers)
 327     {
 328     log_write(0, LOG_MAIN|LOG_PANIC,
 329        "%s no useable spamd server addresses in spamd_address configuration option.",
 330        loglabel);
 331     goto defer;
 332     }
 333
 334   while (1)
 335     {
 336     uschar * errstr;
 337
 338     current_server = spamd_get_server(spamd_address_vector, num_servers);
 339     sd = spamd_address_vector[current_server];
 340
 341     debug_printf("trying server %s\n", sd->hostspec);
 342
 343     /* contact a spamd */
 344     if ((spamd_sock = ip_streamsocket(sd->hostspec, &errstr, 5)) >= 0)
 345       break;
 346
 347     log_write(0, LOG_MAIN, "%s spamd: %s", loglabel, errstr);
 348     sd->is_failed = TRUE;
 349
 350     current_server = spamd_get_server(spamd_address_vector, num_servers);
 351     if (current_server < 0)
 352       {
 353       log_write(0, LOG_MAIN|LOG_PANIC, "%s all spamd servers failed",
 354         loglabel);
 355       goto defer;
 356       }
 357     }
 358     is_rspamd = sd->is_rspamd;
 359   }
 360
 361 if (spamd_sock == -1)
 362   {
 363   log_write(0, LOG_MAIN|LOG_PANIC,
 364       "programming fault, spamd_sock unexpectedly unset");
 365   goto defer;
 366   }
 367
 368 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 369 /* now we are connected to spamd on spamd_sock */
 370 if (is_rspamd)
 371   {                             /* rspamd variant */
 372   uschar *req_str;
 373   const char *helo;
 374   const char *fcrdns;
 375
 376   req_str = string_sprintf("CHECK RSPAMC/1.3\r\nContent-length: %lu\r\n"
 377     "Queue-Id: %s\r\nFrom: <%s>\r\nRecipient-Number: %d\r\n", mbox_size,
 378     message_id, sender_address, recipients_count);
 379   for (i = 0; i < recipients_count; i ++)
 380     req_str = string_sprintf("%sRcpt: <%s>\r\n", req_str, recipients_list[i].address);
 381   if ((helo = expand_string(US"$sender_helo_name")) != NULL && *helo != '\0')
 382     req_str = string_sprintf("%sHelo: %s\r\n", req_str, helo);
 383   if ((fcrdns = expand_string(US"$sender_host_name")) != NULL && *fcrdns != '\0')
 384     req_str = string_sprintf("%sHostname: %s\r\n", req_str, fcrdns);
 385   if (sender_host_address != NULL)
 386     req_str = string_sprintf("%sIP: %s\r\n", req_str, sender_host_address);
 387   req_str = string_sprintf("%s\r\n", req_str);
 388   wrote = send(spamd_sock, req_str, Ustrlen(req_str), 0);
 389   }
 390   else
 391   {                             /* spamassassin variant */
 392   (void)string_format(spamd_buffer,
 393           sizeof(spamd_buffer),
 394           "REPORT SPAMC/1.2\r\nUser: %s\r\nContent-length: %ld\r\n\r\n",
 395           user_name,
 396           mbox_size);
 397   /* send our request */
 398   wrote = send(spamd_sock, spamd_buffer, Ustrlen(spamd_buffer), 0);
 399   }
 400 if (wrote == -1)
 401   {
 402   (void)close(spamd_sock);
 403   log_write(0, LOG_MAIN|LOG_PANIC,
 404        "%s spamd send failed: %s", loglabel, strerror(errno));
 405   goto defer;
 406   }
 407
 408 /* now send the file */
 409 /* spamd sometimes accepts conections but doesn't read data off
 410  * the connection.  We make the file descriptor non-blocking so
 411  * that the write will only write sufficient data without blocking
 412  * and we poll the desciptor to make sure that we can write without
 413  * blocking.  Short writes are gracefully handled and if the whole
 414  * trasaction takes too long it is aborted.
 415  * Note: poll() is not supported in OSX 10.2 and is reported to be
 416  *       broken in more recent versions (up to 10.4).
 417  */
 418 #ifndef NO_POLL_H
 419 pollfd.fd = spamd_sock;
 420 pollfd.events = POLLOUT;
 421 #endif
 422 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 423 do
 424   {
 425   read = fread(spamd_buffer,1,sizeof(spamd_buffer),mbox_file);
 426   if (read > 0)
 427     {
 428     offset = 0;
 429 again:
 430 #ifndef NO_POLL_H
 431     result = poll(&pollfd, 1, 1000);
 432
 433 /* Patch posted by Erik ? for OS X and applied by PH */
 434 #else
 435     select_tv.tv_sec = 1;
 436     select_tv.tv_usec = 0;
 437     FD_ZERO(&select_fd);
 438     FD_SET(spamd_sock, &select_fd);
 439     result = select(spamd_sock+1, NULL, &select_fd, NULL, &select_tv);
 440 #endif
 441 /* End Erik's patch */
 442
 443     if (result == -1 && errno == EINTR)
 444       goto again;
 445     else if (result < 1)
 446       {
 447       if (result == -1)
 448         log_write(0, LOG_MAIN|LOG_PANIC,
 449           "%s %s on spamd socket", loglabel, strerror(errno));
 450       else
 451         {
 452         if (time(NULL) - start < timeout_sec)
 453           goto again;
 454         log_write(0, LOG_MAIN|LOG_PANIC,
 455           "%s timed out writing spamd socket", loglabel);
 456         }
 457       (void)close(spamd_sock);
 458       goto defer;
 459       }
 460
 461     wrote = send(spamd_sock,spamd_buffer + offset,read - offset,0);
 462     if (wrote == -1)
 463       {
 464       log_write(0, LOG_MAIN|LOG_PANIC,
 465           "%s %s on spamd socket", loglabel, strerror(errno));
 466       (void)close(spamd_sock);
 467       goto defer;
 468       }
 469     if (offset + wrote != read)
 470       {
 471       offset += wrote;
 472       goto again;
 473       }
 474     }
 475   }
 476 while (!feof(mbox_file) && !ferror(mbox_file));
 477
 478 if (ferror(mbox_file))
 479   {
 480   log_write(0, LOG_MAIN|LOG_PANIC,
 481     "%s error reading spool file: %s", loglabel, strerror(errno));
 482   (void)close(spamd_sock);
 483   goto defer;
 484   }
 485
 486 (void)fclose(mbox_file);
 487
 488 /* we're done sending, close socket for writing */
 489 shutdown(spamd_sock,SHUT_WR);
 490
 491 /* read spamd response using what's left of the timeout.  */
 492 memset(spamd_buffer, 0, sizeof(spamd_buffer));
 493 offset = 0;
 494 while ((i = ip_recv(spamd_sock,
 495                    spamd_buffer + offset,
 496                    sizeof(spamd_buffer) - offset - 1,
 497                    timeout_sec - time(NULL) + start)) > 0 )
 498   offset += i;
 499
 500 /* error handling */
 501 if (i <= 0 && errno != 0)
 502   {
 503   log_write(0, LOG_MAIN|LOG_PANIC,
 504        "%s error reading from spamd socket: %s", loglabel, strerror(errno));
 505   (void)close(spamd_sock);
 506   return DEFER;
 507   }
 508
 509 /* reading done */
 510 (void)close(spamd_sock);
 511
 512 if (is_rspamd)
 513   {                             /* rspamd variant of reply */
 514   int r;
 515   if ((r = sscanf(CS spamd_buffer,
 516           "RSPAMD/%7s 0 EX_OK\r\nMetric: default; %7s %lf / %lf / %lf\r\n%n",
 517           spamd_version, spamd_short_result, &spamd_score, &spamd_threshold,
 518           &spamd_reject_score, &spamd_report_offset)) != 5)
 519     {
 520       log_write(0, LOG_MAIN|LOG_PANIC,
 521                 "%s cannot parse spamd output: %d", loglabel, r);
 522       return DEFER;
 523     }
 524   /* now parse action */
 525   p = &spamd_buffer[spamd_report_offset];
 526
 527   if (Ustrncmp(p, "Action: ", sizeof("Action: ") - 1) == 0)
 528     {
 529     p += sizeof("Action: ") - 1;
 530     q = &spam_action_buffer[0];
 531     while (*p && *p != '\r' && (q - spam_action_buffer) < sizeof(spam_action_buffer) - 1)
 532       *q++ = *p++;
 533     *q = '\0';
 534     }
 535   }
 536 else
 537   {                             /* spamassassin */
 538   /* dig in the spamd output and put the report in a multiline header,
 539   if requested */
 540   if (sscanf(CS spamd_buffer,
 541        "SPAMD/%7s 0 EX_OK\r\nContent-length: %*u\r\n\r\n%lf/%lf\r\n%n",
 542        spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 543     {
 544       /* try to fall back to pre-2.50 spamd output */
 545       if (sscanf(CS spamd_buffer,
 546            "SPAMD/%7s 0 EX_OK\r\nSpam: %*s ; %lf / %lf\r\n\r\n%n",
 547            spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 548         {
 549         log_write(0, LOG_MAIN|LOG_PANIC,
 550                   "%s cannot parse spamd output", loglabel);
 551         return DEFER;
 552         }
 553     }
 554
 555   Ustrcpy(spam_action_buffer,
 556     spamd_score >= spamd_threshold ? "reject" : "no action");
 557   }
 558
 559 /* Create report. Since this is a multiline string,
 560 we must hack it into shape first */
 561 p = &spamd_buffer[spamd_report_offset];
 562 q = spam_report_buffer;
 563 while (*p != '\0')
 564   {
 565   /* skip \r */
 566   if (*p == '\r')
 567     {
 568     p++;
 569     continue;
 570     }
 571   *q++ = *p;
 572   if (*p++ == '\n')
 573     {
 574     /* add an extra space after the newline to ensure
 575     that it is treated as a header continuation line */
 576     *q++ = ' ';
 577     }
 578   }
 579 /* NULL-terminate */
 580 *q-- = '\0';
 581 /* cut off trailing leftovers */
 582 while (*q <= ' ')
 583   *q-- = '\0';
 584
 585 spam_report = spam_report_buffer;
 586 spam_action = spam_action_buffer;
 587
 588 /* create spam bar */
 589 spamd_score_char = spamd_score > 0 ? '+' : '-';
 590 j = abs((int)(spamd_score));
 591 i = 0;
 592 if (j != 0)
 593   while ((i < j) && (i <= MAX_SPAM_BAR_CHARS))
 594      spam_bar_buffer[i++] = spamd_score_char;
 595 else
 596   {
 597   spam_bar_buffer[0] = '/';
 598   i = 1;
 599   }
 600 spam_bar_buffer[i] = '\0';
 601 spam_bar = spam_bar_buffer;
 602
 603 /* create "float" spam score */
 604 (void)string_format(spam_score_buffer, sizeof(spam_score_buffer),
 605         "%.1f", spamd_score);
 606 spam_score = spam_score_buffer;
 607
 608 /* create "int" spam score */
 609 j = (int)((spamd_score + 0.001)*10);
 610 (void)string_format(spam_score_int_buffer, sizeof(spam_score_int_buffer),
 611         "%d", j);
 612 spam_score_int = spam_score_int_buffer;
 613
 614 /* compare threshold against score */
 615 spam_rc = spamd_score >= spamd_threshold
 616   ? OK  /* spam as determined by user's threshold */
 617   : FAIL;       /* not spam */
 618
 619 /* remember expanded spamd_address if needed */
 620 if (spamd_address_work != spamd_address)
 621   prev_spamd_address_work = string_copy(spamd_address_work);
 622
 623 /* remember user name and "been here" for it */
 624 Ustrcpy(prev_user_name, user_name);
 625 spam_ok = 1;
 626
 627 return override
 628   ? OK          /* always return OK, no matter what the score */
 629   : spam_rc;
 630
 631 defer:
 632   (void)fclose(mbox_file);
 633   return DEFER;
 634 }
 635
 636 #endif
 637 /* vi: aw ai sw=2
 638 */