src/src/spam.c

   1 /*************************************************
   2 *     Exim - an Internet mail transport agent    *
   3 *************************************************/
   4
   5 /* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003 - 2015 */
   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 const uschar * loglabel = US"spam acl condition:";
  25
  26
  27 static int
  28 spamd_param_init(spamd_address_container *spamd)
  29 {
  30 /* default spamd server weight, time and priority value */
  31 spamd->is_rspamd = FALSE;
  32 spamd->is_failed = FALSE;
  33 spamd->weight = SPAMD_WEIGHT;
  34 spamd->timeout = SPAMD_TIMEOUT;
  35 spamd->retry = 0;
  36 spamd->priority = 1;
  37 return 0;
  38 }
  39
  40
  41 static int
  42 spamd_param(const uschar * param, spamd_address_container * spamd)
  43 {
  44 static int timesinceday = -1;
  45 const uschar * s;
  46 const uschar * name;
  47
  48 /*XXX more clever parsing could discard embedded spaces? */
  49
  50 if (sscanf(CCS param, "pri=%u", &spamd->priority))
  51   return 0; /* OK */
  52
  53 if (sscanf(CCS param, "weight=%u", &spamd->weight))
  54   {
  55   if (spamd->weight == 0) /* this server disabled: skip it */
  56     return 1;
  57   return 0; /* OK */
  58   }
  59
  60 if (Ustrncmp(param, "time=", 5) == 0)
  61   {
  62   unsigned int start_h = 0, start_m = 0, start_s = 0;
  63   unsigned int end_h = 24, end_m = 0, end_s = 0;
  64   unsigned int time_start, time_end;
  65   const uschar * end_string;
  66
  67   name = US"time";
  68   s = param+5;
  69   if ((end_string = Ustrchr(s, '-')))
  70     {
  71     end_string++;
  72     if (  sscanf(CS end_string, "%u.%u.%u", &end_h,   &end_m,   &end_s)   == 0
  73        || sscanf(CS s,          "%u.%u.%u", &start_h, &start_m, &start_s) == 0
  74        )
  75       goto badval;
  76     }
  77   else
  78     goto badval;
  79
  80   if (timesinceday < 0)
  81     {
  82     time_t now = time(NULL);
  83     struct tm *tmp = localtime(&now);
  84     timesinceday = tmp->tm_hour*3600 + tmp->tm_min*60 + tmp->tm_sec;
  85     }
  86
  87   time_start = start_h*3600 + start_m*60 + start_s;
  88   time_end = end_h*3600 + end_m*60 + end_s;
  89
  90   if (timesinceday < time_start || timesinceday >= time_end)
  91     return 1; /* skip spamd server */
  92
  93   return 0; /* OK */
  94   }
  95
  96 if (Ustrcmp(param, "variant=rspamd") == 0)
  97   {
  98   spamd->is_rspamd = TRUE;
  99   return 0;
 100   }
 101
 102 if (Ustrncmp(param, "tmo=", 4) == 0)
 103   {
 104   int sec = readconf_readtime((s = param+4), '\0', FALSE);
 105   name = US"timeout";
 106   if (sec < 0)
 107     goto badval;
 108   spamd->timeout = sec;
 109   return 0;
 110   }
 111
 112 if (Ustrncmp(param, "retry=", 6) == 0)
 113   {
 114   int sec = readconf_readtime((s = param+6), '\0', FALSE);
 115   name = US"retry";
 116   if (sec < 0)
 117     goto badval;
 118   spamd->retry = sec;
 119   return 0;
 120   }
 121
 122 log_write(0, LOG_MAIN, "%s warning - invalid spamd parameter: '%s'",
 123   loglabel, param);
 124 return -1; /* syntax error */
 125
 126 badval:
 127   log_write(0, LOG_MAIN,
 128     "%s warning - invalid spamd %s value: '%s'", loglabel, name, s);
 129   return -1; /* syntax error */
 130 }
 131
 132
 133 static int
 134 spamd_get_server(spamd_address_container ** spamds, int num_servers)
 135 {
 136 unsigned int i;
 137 spamd_address_container * sd;
 138 long rnd, weights;
 139 unsigned pri;
 140 static BOOL srandomed = FALSE;
 141
 142 /* speedup, if we have only 1 server */
 143 if (num_servers == 1)
 144   return (spamds[0]->is_failed ? -1 : 0);
 145
 146 /* init ranmod */
 147 if (!srandomed)
 148   {
 149   struct timeval tv;
 150   gettimeofday(&tv, NULL);
 151   srandom((unsigned int)(tv.tv_usec/1000));
 152   srandomed = TRUE;
 153   }
 154
 155 /* scan for highest pri */
 156 for (pri = 0, i = 0; i < num_servers; i++)
 157   {
 158   sd = spamds[i];
 159   if (!sd->is_failed && sd->priority > pri) pri = sd->priority;
 160   }
 161
 162 /* get sum of weights */
 163 for (weights = 0, i = 0; i < num_servers; i++)
 164   {
 165   sd = spamds[i];
 166   if (!sd->is_failed && sd->priority == pri) weights += sd->weight;
 167   }
 168 if (weights == 0)       /* all servers failed */
 169   return -1;
 170
 171 for (rnd = random() % weights, i = 0; i < num_servers; i++)
 172   {
 173   sd = spamds[i];
 174   if (!sd->is_failed && sd->priority == pri)
 175     if ((rnd -= sd->weight) <= 0)
 176       return i;
 177   }
 178
 179 log_write(0, LOG_MAIN|LOG_PANIC,
 180   "%s unknown error (memory/cpu corruption?)", loglabel);
 181 return -1;
 182 }
 183
 184
 185 int
 186 spam(const uschar **listptr)
 187 {
 188 int sep = 0;
 189 const uschar *list = *listptr;
 190 uschar *user_name;
 191 uschar user_name_buffer[128];
 192 unsigned long mbox_size;
 193 FILE *mbox_file;
 194 int spamd_sock = -1;
 195 uschar spamd_buffer[32600];
 196 int i, j, offset, result;
 197 uschar spamd_version[8];
 198 uschar spamd_short_result[8];
 199 uschar spamd_score_char;
 200 double spamd_threshold, spamd_score, spamd_reject_score;
 201 int spamd_report_offset;
 202 uschar *p,*q;
 203 int override = 0;
 204 time_t start;
 205 size_t read, wrote;
 206 #ifndef NO_POLL_H
 207 struct pollfd pollfd;
 208 #else                               /* Patch posted by Erik ? for OS X */
 209 struct timeval select_tv;         /* and applied by PH */
 210 fd_set select_fd;
 211 #endif
 212 uschar *spamd_address_work;
 213 spamd_address_container * sd;
 214
 215 /* stop compiler warning */
 216 result = 0;
 217
 218 /* find the username from the option list */
 219 if ((user_name = string_nextinlist(&list, &sep,
 220                                    user_name_buffer,
 221                                    sizeof(user_name_buffer))) == NULL)
 222   {
 223   /* no username given, this means no scanning should be done */
 224   return FAIL;
 225   }
 226
 227 /* if username is "0" or "false", do not scan */
 228 if ( (Ustrcmp(user_name,"0") == 0) ||
 229      (strcmpic(user_name,US"false") == 0) )
 230   return FAIL;
 231
 232 /* if there is an additional option, check if it is "true" */
 233 if (strcmpic(list,US"true") == 0)
 234   /* in that case, always return true later */
 235   override = 1;
 236
 237 /* expand spamd_address if needed */
 238 if (*spamd_address == '$')
 239   {
 240   spamd_address_work = expand_string(spamd_address);
 241   if (spamd_address_work == NULL)
 242     {
 243     log_write(0, LOG_MAIN|LOG_PANIC,
 244       "%s spamd_address starts with $, but expansion failed: %s",
 245       loglabel, expand_string_message);
 246     return DEFER;
 247     }
 248   }
 249 else
 250   spamd_address_work = spamd_address;
 251
 252 DEBUG(D_acl) debug_printf("spamd: addrlist '%s'\n", spamd_address_work);
 253
 254 /* check if previous spamd_address was expanded and has changed. dump cached results if so */
 255 if (  spam_ok
 256    && prev_spamd_address_work != NULL
 257    && Ustrcmp(prev_spamd_address_work, spamd_address_work) != 0
 258    )
 259   spam_ok = 0;
 260
 261 /* if we scanned for this username last time, just return */
 262 if (spam_ok && Ustrcmp(prev_user_name, user_name) == 0)
 263   return override ? OK : spam_rc;
 264
 265 /* make sure the eml mbox file is spooled up */
 266 mbox_file = spool_mbox(&mbox_size, NULL);
 267
 268 if (mbox_file == NULL)
 269   {
 270   /* error while spooling */
 271   log_write(0, LOG_MAIN|LOG_PANIC,
 272          "%s error while creating mbox spool file", loglabel);
 273   return DEFER;
 274   }
 275
 276 start = time(NULL);
 277
 278   {
 279   int num_servers = 0;
 280   int current_server;
 281   uschar * address;
 282   const uschar * spamd_address_list_ptr = spamd_address_work;
 283   spamd_address_container * spamd_address_vector[32];
 284
 285   /* Check how many spamd servers we have
 286      and register their addresses */
 287   sep = 0;                              /* default colon-sep */
 288   while ((address = string_nextinlist(&spamd_address_list_ptr, &sep,
 289                                       NULL, 0)) != NULL)
 290     {
 291     const uschar * sublist;
 292     int sublist_sep = -(int)' ';        /* default space-sep */
 293     unsigned args;
 294     uschar * s;
 295
 296     DEBUG(D_acl) debug_printf("spamd: addr entry '%s'\n", address);
 297     sd = (spamd_address_container *)store_get(sizeof(spamd_address_container));
 298
 299     for (sublist = address, args = 0, spamd_param_init(sd);
 300          (s = string_nextinlist(&sublist, &sublist_sep, NULL, 0));
 301          args++
 302          )
 303       {
 304         DEBUG(D_acl) debug_printf("spamd:  addr parm '%s'\n", s);
 305         switch (args)
 306         {
 307         case 0:   sd->hostspec = s;
 308                   if (*s == '/') args++;        /* local; no port */
 309                   break;
 310         case 1:   sd->hostspec = string_sprintf("%s %s", sd->hostspec, s);
 311                   break;
 312         default:  spamd_param(s, sd);
 313                   break;
 314         }
 315       }
 316     if (args < 2)
 317       {
 318       log_write(0, LOG_MAIN,
 319         "%s warning - invalid spamd address: '%s'", loglabel, address);
 320       continue;
 321       }
 322
 323     spamd_address_vector[num_servers] = sd;
 324     if (++num_servers > 31)
 325       break;
 326     }
 327
 328   /* check if we have at least one server */
 329   if (!num_servers)
 330     {
 331     log_write(0, LOG_MAIN|LOG_PANIC,
 332        "%s no useable spamd server addresses in spamd_address configuration option.",
 333        loglabel);
 334     goto defer;
 335     }
 336
 337   current_server = spamd_get_server(spamd_address_vector, num_servers);
 338   sd = spamd_address_vector[current_server];
 339   for(;;)
 340     {
 341     uschar * errstr;
 342
 343     DEBUG(D_acl) debug_printf("spamd: trying server %s\n", sd->hostspec);
 344
 345     for (;;)
 346       {
 347       if (  (spamd_sock = ip_streamsocket(sd->hostspec, &errstr, 5)) >= 0
 348          || sd->retry <= 0
 349          )
 350         break;
 351       DEBUG(D_acl) debug_printf("spamd: server %s: retry conn\n", sd->hostspec);
 352       while (sd->retry > 0) sd->retry = sleep(sd->retry);
 353       }
 354     if (spamd_sock >= 0)
 355       break;
 356
 357     log_write(0, LOG_MAIN, "%s spamd: %s", loglabel, errstr);
 358     sd->is_failed = TRUE;
 359
 360     current_server = spamd_get_server(spamd_address_vector, num_servers);
 361     if (current_server < 0)
 362       {
 363       log_write(0, LOG_MAIN|LOG_PANIC, "%s all spamd servers failed", loglabel);
 364       goto defer;
 365       }
 366     sd = spamd_address_vector[current_server];
 367     }
 368   }
 369
 370 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 371 /* now we are connected to spamd on spamd_sock */
 372 if (sd->is_rspamd)
 373   {                             /* rspamd variant */
 374   uschar *req_str;
 375   const uschar * helo;
 376   const uschar * fcrdns;
 377   const uschar * authid;
 378
 379   req_str = string_sprintf("CHECK RSPAMC/1.3\r\nContent-length: %lu\r\n"
 380     "Queue-Id: %s\r\nFrom: <%s>\r\nRecipient-Number: %d\r\n",
 381     mbox_size, message_id, sender_address, recipients_count);
 382   for (i = 0; i < recipients_count; i ++)
 383     req_str = string_sprintf("%sRcpt: <%s>\r\n", req_str, recipients_list[i].address);
 384   if ((helo = expand_string(US"$sender_helo_name")) != NULL && *helo != '\0')
 385     req_str = string_sprintf("%sHelo: %s\r\n", req_str, helo);
 386   if ((fcrdns = expand_string(US"$sender_host_name")) != NULL && *fcrdns != '\0')
 387     req_str = string_sprintf("%sHostname: %s\r\n", req_str, fcrdns);
 388   if (sender_host_address != NULL)
 389     req_str = string_sprintf("%sIP: %s\r\n", req_str, sender_host_address);
 390   if ((authid = expand_string(US"$authenticated_id")) != NULL && *authid != '\0')
 391     req_str = string_sprintf("%sUser: %s\r\n", req_str, authid);
 392   req_str = string_sprintf("%s\r\n", req_str);
 393   wrote = send(spamd_sock, req_str, Ustrlen(req_str), 0);
 394   }
 395 else
 396   {                             /* spamassassin variant */
 397   (void)string_format(spamd_buffer,
 398           sizeof(spamd_buffer),
 399           "REPORT SPAMC/1.2\r\nUser: %s\r\nContent-length: %ld\r\n\r\n",
 400           user_name,
 401           mbox_size);
 402   /* send our request */
 403   wrote = send(spamd_sock, spamd_buffer, Ustrlen(spamd_buffer), 0);
 404   }
 405
 406 if (wrote == -1)
 407   {
 408   (void)close(spamd_sock);
 409   log_write(0, LOG_MAIN|LOG_PANIC,
 410        "%s spamd %s send failed: %s", loglabel, callout_address, strerror(errno));
 411   goto defer;
 412   }
 413
 414 /* now send the file */
 415 /* spamd sometimes accepts conections but doesn't read data off
 416  * the connection.  We make the file descriptor non-blocking so
 417  * that the write will only write sufficient data without blocking
 418  * and we poll the desciptor to make sure that we can write without
 419  * blocking.  Short writes are gracefully handled and if the whole
 420  * trasaction takes too long it is aborted.
 421  * Note: poll() is not supported in OSX 10.2 and is reported to be
 422  *       broken in more recent versions (up to 10.4).
 423  */
 424 #ifndef NO_POLL_H
 425 pollfd.fd = spamd_sock;
 426 pollfd.events = POLLOUT;
 427 #endif
 428 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 429 do
 430   {
 431   read = fread(spamd_buffer,1,sizeof(spamd_buffer),mbox_file);
 432   if (read > 0)
 433     {
 434     offset = 0;
 435 again:
 436 #ifndef NO_POLL_H
 437     result = poll(&pollfd, 1, 1000);
 438
 439 /* Patch posted by Erik ? for OS X and applied by PH */
 440 #else
 441     select_tv.tv_sec = 1;
 442     select_tv.tv_usec = 0;
 443     FD_ZERO(&select_fd);
 444     FD_SET(spamd_sock, &select_fd);
 445     result = select(spamd_sock+1, NULL, &select_fd, NULL, &select_tv);
 446 #endif
 447 /* End Erik's patch */
 448
 449     if (result == -1 && errno == EINTR)
 450       goto again;
 451     else if (result < 1)
 452       {
 453       if (result == -1)
 454         log_write(0, LOG_MAIN|LOG_PANIC,
 455           "%s %s on spamd %s socket", loglabel, callout_address, strerror(errno));
 456       else
 457         {
 458         if (time(NULL) - start < sd->timeout)
 459           goto again;
 460         log_write(0, LOG_MAIN|LOG_PANIC,
 461           "%s timed out writing spamd %s, socket", loglabel, callout_address);
 462         }
 463       (void)close(spamd_sock);
 464       goto defer;
 465       }
 466
 467     wrote = send(spamd_sock,spamd_buffer + offset,read - offset,0);
 468     if (wrote == -1)
 469       {
 470       log_write(0, LOG_MAIN|LOG_PANIC,
 471           "%s %s on spamd %s socket", loglabel, callout_address, strerror(errno));
 472       (void)close(spamd_sock);
 473       goto defer;
 474       }
 475     if (offset + wrote != read)
 476       {
 477       offset += wrote;
 478       goto again;
 479       }
 480     }
 481   }
 482 while (!feof(mbox_file) && !ferror(mbox_file));
 483
 484 if (ferror(mbox_file))
 485   {
 486   log_write(0, LOG_MAIN|LOG_PANIC,
 487     "%s error reading spool file: %s", loglabel, strerror(errno));
 488   (void)close(spamd_sock);
 489   goto defer;
 490   }
 491
 492 (void)fclose(mbox_file);
 493
 494 /* we're done sending, close socket for writing */
 495 shutdown(spamd_sock,SHUT_WR);
 496
 497 /* read spamd response using what's left of the timeout.  */
 498 memset(spamd_buffer, 0, sizeof(spamd_buffer));
 499 offset = 0;
 500 while ((i = ip_recv(spamd_sock,
 501                    spamd_buffer + offset,
 502                    sizeof(spamd_buffer) - offset - 1,
 503                    sd->timeout - time(NULL) + start)) > 0 )
 504   offset += i;
 505
 506 /* error handling */
 507 if (i <= 0 && errno != 0)
 508   {
 509   log_write(0, LOG_MAIN|LOG_PANIC,
 510        "%s error reading from spamd %s, socket: %s", loglabel, callout_address, strerror(errno));
 511   (void)close(spamd_sock);
 512   return DEFER;
 513   }
 514
 515 /* reading done */
 516 (void)close(spamd_sock);
 517
 518 if (sd->is_rspamd)
 519   {                             /* rspamd variant of reply */
 520   int r;
 521   if ((r = sscanf(CS spamd_buffer,
 522           "RSPAMD/%7s 0 EX_OK\r\nMetric: default; %7s %lf / %lf / %lf\r\n%n",
 523           spamd_version, spamd_short_result, &spamd_score, &spamd_threshold,
 524           &spamd_reject_score, &spamd_report_offset)) != 5)
 525     {
 526     log_write(0, LOG_MAIN|LOG_PANIC,
 527               "%s cannot parse spamd %s, output: %d", loglabel, callout_address, r);
 528     return DEFER;
 529     }
 530   /* now parse action */
 531   p = &spamd_buffer[spamd_report_offset];
 532
 533   if (Ustrncmp(p, "Action: ", sizeof("Action: ") - 1) == 0)
 534     {
 535     p += sizeof("Action: ") - 1;
 536     q = &spam_action_buffer[0];
 537     while (*p && *p != '\r' && (q - spam_action_buffer) < sizeof(spam_action_buffer) - 1)
 538       *q++ = *p++;
 539     *q = '\0';
 540     }
 541   }
 542 else
 543   {                             /* spamassassin */
 544   /* dig in the spamd output and put the report in a multiline header,
 545   if requested */
 546   if (sscanf(CS spamd_buffer,
 547        "SPAMD/%7s 0 EX_OK\r\nContent-length: %*u\r\n\r\n%lf/%lf\r\n%n",
 548        spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 549     {
 550       /* try to fall back to pre-2.50 spamd output */
 551       if (sscanf(CS spamd_buffer,
 552            "SPAMD/%7s 0 EX_OK\r\nSpam: %*s ; %lf / %lf\r\n\r\n%n",
 553            spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 554         {
 555         log_write(0, LOG_MAIN|LOG_PANIC,
 556                   "%s cannot parse spamd %s output", loglabel, callout_address);
 557         return DEFER;
 558         }
 559     }
 560
 561   Ustrcpy(spam_action_buffer,
 562     spamd_score >= spamd_threshold ? "reject" : "no action");
 563   }
 564
 565 /* Create report. Since this is a multiline string,
 566 we must hack it into shape first */
 567 p = &spamd_buffer[spamd_report_offset];
 568 q = spam_report_buffer;
 569 while (*p != '\0')
 570   {
 571   /* skip \r */
 572   if (*p == '\r')
 573     {
 574     p++;
 575     continue;
 576     }
 577   *q++ = *p;
 578   if (*p++ == '\n')
 579     {
 580     /* add an extra space after the newline to ensure
 581     that it is treated as a header continuation line */
 582     *q++ = ' ';
 583     }
 584   }
 585 /* NULL-terminate */
 586 *q-- = '\0';
 587 /* cut off trailing leftovers */
 588 while (*q <= ' ')
 589   *q-- = '\0';
 590
 591 spam_report = spam_report_buffer;
 592 spam_action = spam_action_buffer;
 593
 594 /* create spam bar */
 595 spamd_score_char = spamd_score > 0 ? '+' : '-';
 596 j = abs((int)(spamd_score));
 597 i = 0;
 598 if (j != 0)
 599   while ((i < j) && (i <= MAX_SPAM_BAR_CHARS))
 600      spam_bar_buffer[i++] = spamd_score_char;
 601 else
 602   {
 603   spam_bar_buffer[0] = '/';
 604   i = 1;
 605   }
 606 spam_bar_buffer[i] = '\0';
 607 spam_bar = spam_bar_buffer;
 608
 609 /* create "float" spam score */
 610 (void)string_format(spam_score_buffer, sizeof(spam_score_buffer),
 611         "%.1f", spamd_score);
 612 spam_score = spam_score_buffer;
 613
 614 /* create "int" spam score */
 615 j = (int)((spamd_score + 0.001)*10);
 616 (void)string_format(spam_score_int_buffer, sizeof(spam_score_int_buffer),
 617         "%d", j);
 618 spam_score_int = spam_score_int_buffer;
 619
 620 /* compare threshold against score */
 621 spam_rc = spamd_score >= spamd_threshold
 622   ? OK  /* spam as determined by user's threshold */
 623   : FAIL;       /* not spam */
 624
 625 /* remember expanded spamd_address if needed */
 626 if (spamd_address_work != spamd_address)
 627   prev_spamd_address_work = string_copy(spamd_address_work);
 628
 629 /* remember user name and "been here" for it */
 630 Ustrcpy(prev_user_name, user_name);
 631 spam_ok = 1;
 632
 633 return override
 634   ? OK          /* always return OK, no matter what the score */
 635   : spam_rc;
 636
 637 defer:
 638   (void)fclose(mbox_file);
 639   return DEFER;
 640 }
 641
 642 #endif
 643 /* vi: aw ai sw=2
 644 */