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 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_failed = FALSE;
  32 spamd->weight = SPAMD_WEIGHT;
  33 spamd->timeout = SPAMD_TIMEOUT;
  34 spamd->retry = 0;
  35 spamd->priority = 1;
  36 return 0;
  37 }
  38
  39
  40 static int
  41 spamd_param(const uschar * param, spamd_address_container * spamd)
  42 {
  43 static int timesinceday = -1;
  44 const uschar * s;
  45 const uschar * name;
  46
  47 /*XXX more clever parsing could discard embedded spaces? */
  48
  49 if (sscanf(CCS param, "pri=%u", &spamd->priority))
  50   return 0; /* OK */
  51
  52 if (sscanf(CCS param, "weight=%u", &spamd->weight))
  53   {
  54   if (spamd->weight == 0) /* this server disabled: skip it */
  55     return 1;
  56   return 0; /* OK */
  57   }
  58
  59 if (Ustrncmp(param, "time=", 5) == 0)
  60   {
  61   unsigned int start_h = 0, start_m = 0, start_s = 0;
  62   unsigned int end_h = 24, end_m = 0, end_s = 0;
  63   unsigned int time_start, time_end;
  64   const uschar * end_string;
  65
  66   name = US"time";
  67   s = param+5;
  68   if ((end_string = Ustrchr(s, '-')))
  69     {
  70     end_string++;
  71     if (  sscanf(CS end_string, "%u.%u.%u", &end_h,   &end_m,   &end_s)   == 0
  72        || sscanf(CS s,          "%u.%u.%u", &start_h, &start_m, &start_s) == 0
  73        )
  74       goto badval;
  75     }
  76   else
  77     goto badval;
  78
  79   if (timesinceday < 0)
  80     {
  81     time_t now = time(NULL);
  82     struct tm *tmp = localtime(&now);
  83     timesinceday = tmp->tm_hour*3600 + tmp->tm_min*60 + tmp->tm_sec;
  84     }
  85
  86   time_start = start_h*3600 + start_m*60 + start_s;
  87   time_end = end_h*3600 + end_m*60 + end_s;
  88
  89   if (timesinceday < time_start || timesinceday >= time_end)
  90     return 1; /* skip spamd server */
  91
  92   return 0; /* OK */
  93   }
  94
  95 if (Ustrcmp(param, "variant=rspamd") == 0)
  96   {
  97   spamd->is_rspamd = TRUE;
  98   return 0;
  99   }
 100
 101 if (Ustrncmp(param, "tmo=", 4) == 0)
 102   {
 103   int sec = readconf_readtime((s = param+4), '\0', FALSE);
 104   name = US"timeout";
 105   if (sec < 0)
 106     goto badval;
 107   spamd->timeout = sec;
 108   return 0;
 109   }
 110
 111 if (Ustrncmp(param, "retry=", 6) == 0)
 112   {
 113   int sec = readconf_readtime((s = param+6), '\0', FALSE);
 114   name = US"retry";
 115   if (sec < 0)
 116     goto badval;
 117   spamd->retry = sec;
 118   return 0;
 119   }
 120
 121 log_write(0, LOG_MAIN, "%s warning - invalid spamd parameter: '%s'",
 122   loglabel, param);
 123 return -1; /* syntax error */
 124
 125 badval:
 126   log_write(0, LOG_MAIN,
 127     "%s warning - invalid spamd %s value: '%s'", loglabel, name, s);
 128   return -1; /* syntax error */
 129 }
 130
 131
 132 static int
 133 spamd_get_server(spamd_address_container ** spamds, int num_servers)
 134 {
 135 unsigned int i;
 136 spamd_address_container * sd;
 137 long rnd, weights;
 138 unsigned pri;
 139 static BOOL srandomed = FALSE;
 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 /* scan for highest pri */
 155 for (pri = 0, i = 0; i < num_servers; i++)
 156   {
 157   sd = spamds[i];
 158   if (!sd->is_failed && sd->priority > pri) pri = sd->priority;
 159   }
 160
 161 /* get sum of weights */
 162 for (weights = 0, i = 0; i < num_servers; i++)
 163   {
 164   sd = spamds[i];
 165   if (!sd->is_failed && sd->priority == pri) weights += sd->weight;
 166   }
 167 if (weights == 0)       /* all servers failed */
 168   return -1;
 169
 170 for (rnd = random() % weights, i = 0; i < num_servers; i++)
 171   {
 172   sd = spamds[i];
 173   if (!sd->is_failed && sd->priority == pri)
 174     if ((rnd -= sd->weight) <= 0)
 175       return i;
 176   }
 177
 178 log_write(0, LOG_MAIN|LOG_PANIC,
 179   "%s unknown error (memory/cpu corruption?)", loglabel);
 180 return -1;
 181 }
 182
 183
 184 int
 185 spam(const uschar **listptr)
 186 {
 187 int sep = 0;
 188 const uschar *list = *listptr;
 189 uschar *user_name;
 190 uschar user_name_buffer[128];
 191 unsigned long mbox_size;
 192 FILE *mbox_file;
 193 int spamd_sock = -1;
 194 uschar spamd_buffer[32600];
 195 int i, j, offset, result;
 196 uschar spamd_version[8];
 197 uschar spamd_short_result[8];
 198 uschar spamd_score_char;
 199 double spamd_threshold, spamd_score, spamd_reject_score;
 200 int spamd_report_offset;
 201 uschar *p,*q;
 202 int override = 0;
 203 time_t start;
 204 size_t read, wrote;
 205 #ifndef NO_POLL_H
 206 struct pollfd pollfd;
 207 #else                               /* Patch posted by Erik ? for OS X */
 208 struct timeval select_tv;         /* and applied by PH */
 209 fd_set select_fd;
 210 #endif
 211 uschar *spamd_address_work;
 212 spamd_address_container * sd;
 213
 214 /* stop compiler warning */
 215 result = 0;
 216
 217 /* find the username from the option list */
 218 if ((user_name = string_nextinlist(&list, &sep,
 219                                    user_name_buffer,
 220                                    sizeof(user_name_buffer))) == NULL)
 221   {
 222   /* no username given, this means no scanning should be done */
 223   return FAIL;
 224   }
 225
 226 /* if username is "0" or "false", do not scan */
 227 if ( (Ustrcmp(user_name,"0") == 0) ||
 228      (strcmpic(user_name,US"false") == 0) )
 229   return FAIL;
 230
 231 /* if there is an additional option, check if it is "true" */
 232 if (strcmpic(list,US"true") == 0)
 233   /* in that case, always return true later */
 234   override = 1;
 235
 236 /* expand spamd_address if needed */
 237 if (*spamd_address == '$')
 238   {
 239   spamd_address_work = expand_string(spamd_address);
 240   if (spamd_address_work == NULL)
 241     {
 242     log_write(0, LOG_MAIN|LOG_PANIC,
 243       "%s spamd_address starts with $, but expansion failed: %s",
 244       loglabel, expand_string_message);
 245     return DEFER;
 246     }
 247   }
 248 else
 249   spamd_address_work = spamd_address;
 250
 251 DEBUG(D_acl) debug_printf("spamd: addrlist '%s'\n", spamd_address_work);
 252
 253 /* check if previous spamd_address was expanded and has changed. dump cached results if so */
 254 if (  spam_ok
 255    && prev_spamd_address_work != NULL
 256    && Ustrcmp(prev_spamd_address_work, spamd_address_work) != 0
 257    )
 258   spam_ok = 0;
 259
 260 /* if we scanned for this username last time, just return */
 261 if (spam_ok && Ustrcmp(prev_user_name, user_name) == 0)
 262   return override ? OK : spam_rc;
 263
 264 /* make sure the eml mbox file is spooled up */
 265 mbox_file = spool_mbox(&mbox_size, NULL);
 266
 267 if (mbox_file == NULL)
 268   {
 269   /* error while spooling */
 270   log_write(0, LOG_MAIN|LOG_PANIC,
 271          "%s error while creating mbox spool file", loglabel);
 272   return DEFER;
 273   }
 274
 275 start = time(NULL);
 276
 277   {
 278   int num_servers = 0;
 279   int current_server;
 280   uschar * address;
 281   const uschar * spamd_address_list_ptr = spamd_address_work;
 282   spamd_address_container * spamd_address_vector[32];
 283
 284   /* Check how many spamd servers we have
 285      and register their addresses */
 286   sep = 0;                              /* default colon-sep */
 287   while ((address = string_nextinlist(&spamd_address_list_ptr, &sep,
 288                                       NULL, 0)) != NULL)
 289     {
 290     const uschar * sublist;
 291     int sublist_sep = -(int)' ';        /* default space-sep */
 292     unsigned args;
 293     uschar * s;
 294
 295     DEBUG(D_acl) debug_printf("spamd: addr entry '%s'\n", address);
 296     sd = (spamd_address_container *)store_get(sizeof(spamd_address_container));
 297
 298     for (sublist = address, args = 0, spamd_param_init(sd);
 299          s = string_nextinlist(&sublist, &sublist_sep, NULL, 0);
 300          args++
 301          )
 302       {
 303         DEBUG(D_acl) debug_printf("spamd:  addr parm '%s'\n", s);
 304         switch (args)
 305         {
 306         case 0:   sd->hostspec = s;
 307                   if (*s == '/') args++;        /* local; no port */
 308                   break;
 309         case 1:   sd->hostspec = string_sprintf("%s %s", sd->hostspec, s);
 310                   break;
 311         default:  spamd_param(s, sd);
 312                   break;
 313         }
 314       }
 315     if (args < 2)
 316       {
 317       log_write(0, LOG_MAIN,
 318         "%s warning - invalid spamd address: '%s'", loglabel, address);
 319       continue;
 320       }
 321
 322     spamd_address_vector[num_servers] = sd;
 323     if (++num_servers > 31)
 324       break;
 325     }
 326
 327   /* check if we have at least one server */
 328   if (!num_servers)
 329     {
 330     log_write(0, LOG_MAIN|LOG_PANIC,
 331        "%s no useable spamd server addresses in spamd_address configuration option.",
 332        loglabel);
 333     goto defer;
 334     }
 335
 336   current_server = spamd_get_server(spamd_address_vector, num_servers);
 337   sd = spamd_address_vector[current_server];
 338   for(;;)
 339     {
 340     uschar * errstr;
 341
 342     DEBUG(D_acl) debug_printf("spamd: trying server %s\n", sd->hostspec);
 343
 344     for (;;)
 345       {
 346       if (  (spamd_sock = ip_streamsocket(sd->hostspec, &errstr, 5)) >= 0
 347          || sd->retry <= 0
 348          )
 349         break;
 350       DEBUG(D_acl) debug_printf("sspamd: erver %s: retry conn\n", sd->hostspec);
 351       while (sd->retry > 0) sd->retry = sleep(sd->retry);
 352       }
 353     if (spamd_sock >= 0)
 354       break;
 355
 356     log_write(0, LOG_MAIN, "%s spamd: %s", loglabel, errstr);
 357     sd->is_failed = TRUE;
 358
 359     current_server = spamd_get_server(spamd_address_vector, num_servers);
 360     if (current_server < 0)
 361       {
 362       log_write(0, LOG_MAIN|LOG_PANIC, "%s all spamd servers failed", loglabel);
 363       goto defer;
 364       }
 365     sd = spamd_address_vector[current_server];
 366     }
 367   }
 368
 369 if (spamd_sock == -1)
 370   {
 371   log_write(0, LOG_MAIN|LOG_PANIC,
 372       "programming fault, spamd_sock unexpectedly unset");
 373   goto defer;
 374   }
 375
 376 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 377 /* now we are connected to spamd on spamd_sock */
 378 if (sd->is_rspamd)
 379   {                             /* rspamd variant */
 380   uschar *req_str;
 381   const uschar * helo;
 382   const uschar * fcrdns;
 383   const uschar * authid;
 384
 385   req_str = string_sprintf("CHECK RSPAMC/1.3\r\nContent-length: %lu\r\n"
 386     "Queue-Id: %s\r\nFrom: <%s>\r\nRecipient-Number: %d\r\n",
 387     mbox_size, message_id, sender_address, recipients_count);
 388   for (i = 0; i < recipients_count; i ++)
 389     req_str = string_sprintf("%sRcpt: <%s>\r\n", req_str, recipients_list[i].address);
 390   if ((helo = expand_string(US"$sender_helo_name")) != NULL && *helo != '\0')
 391     req_str = string_sprintf("%sHelo: %s\r\n", req_str, helo);
 392   if ((fcrdns = expand_string(US"$sender_host_name")) != NULL && *fcrdns != '\0')
 393     req_str = string_sprintf("%sHostname: %s\r\n", req_str, fcrdns);
 394   if (sender_host_address != NULL)
 395     req_str = string_sprintf("%sIP: %s\r\n", req_str, sender_host_address);
 396   if ((authid = expand_string(US"$authenticated_id")) != NULL && *authid != '\0')
 397     req_str = string_sprintf("%sUser: %s\r\n", req_str, authid);
 398   req_str = string_sprintf("%s\r\n", req_str);
 399   wrote = send(spamd_sock, req_str, Ustrlen(req_str), 0);
 400   }
 401 else
 402   {                             /* spamassassin variant */
 403   (void)string_format(spamd_buffer,
 404           sizeof(spamd_buffer),
 405           "REPORT SPAMC/1.2\r\nUser: %s\r\nContent-length: %ld\r\n\r\n",
 406           user_name,
 407           mbox_size);
 408   /* send our request */
 409   wrote = send(spamd_sock, spamd_buffer, Ustrlen(spamd_buffer), 0);
 410   }
 411
 412 if (wrote == -1)
 413   {
 414   (void)close(spamd_sock);
 415   log_write(0, LOG_MAIN|LOG_PANIC,
 416        "%s spamd send failed: %s", loglabel, strerror(errno));
 417   goto defer;
 418   }
 419
 420 /* now send the file */
 421 /* spamd sometimes accepts conections but doesn't read data off
 422  * the connection.  We make the file descriptor non-blocking so
 423  * that the write will only write sufficient data without blocking
 424  * and we poll the desciptor to make sure that we can write without
 425  * blocking.  Short writes are gracefully handled and if the whole
 426  * trasaction takes too long it is aborted.
 427  * Note: poll() is not supported in OSX 10.2 and is reported to be
 428  *       broken in more recent versions (up to 10.4).
 429  */
 430 #ifndef NO_POLL_H
 431 pollfd.fd = spamd_sock;
 432 pollfd.events = POLLOUT;
 433 #endif
 434 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 435 do
 436   {
 437   read = fread(spamd_buffer,1,sizeof(spamd_buffer),mbox_file);
 438   if (read > 0)
 439     {
 440     offset = 0;
 441 again:
 442 #ifndef NO_POLL_H
 443     result = poll(&pollfd, 1, 1000);
 444
 445 /* Patch posted by Erik ? for OS X and applied by PH */
 446 #else
 447     select_tv.tv_sec = 1;
 448     select_tv.tv_usec = 0;
 449     FD_ZERO(&select_fd);
 450     FD_SET(spamd_sock, &select_fd);
 451     result = select(spamd_sock+1, NULL, &select_fd, NULL, &select_tv);
 452 #endif
 453 /* End Erik's patch */
 454
 455     if (result == -1 && errno == EINTR)
 456       goto again;
 457     else if (result < 1)
 458       {
 459       if (result == -1)
 460         log_write(0, LOG_MAIN|LOG_PANIC,
 461           "%s %s on spamd socket", loglabel, strerror(errno));
 462       else
 463         {
 464         if (time(NULL) - start < sd->timeout)
 465           goto again;
 466         log_write(0, LOG_MAIN|LOG_PANIC,
 467           "%s timed out writing spamd socket", loglabel);
 468         }
 469       (void)close(spamd_sock);
 470       goto defer;
 471       }
 472
 473     wrote = send(spamd_sock,spamd_buffer + offset,read - offset,0);
 474     if (wrote == -1)
 475       {
 476       log_write(0, LOG_MAIN|LOG_PANIC,
 477           "%s %s on spamd socket", loglabel, strerror(errno));
 478       (void)close(spamd_sock);
 479       goto defer;
 480       }
 481     if (offset + wrote != read)
 482       {
 483       offset += wrote;
 484       goto again;
 485       }
 486     }
 487   }
 488 while (!feof(mbox_file) && !ferror(mbox_file));
 489
 490 if (ferror(mbox_file))
 491   {
 492   log_write(0, LOG_MAIN|LOG_PANIC,
 493     "%s error reading spool file: %s", loglabel, strerror(errno));
 494   (void)close(spamd_sock);
 495   goto defer;
 496   }
 497
 498 (void)fclose(mbox_file);
 499
 500 /* we're done sending, close socket for writing */
 501 shutdown(spamd_sock,SHUT_WR);
 502
 503 /* read spamd response using what's left of the timeout.  */
 504 memset(spamd_buffer, 0, sizeof(spamd_buffer));
 505 offset = 0;
 506 while ((i = ip_recv(spamd_sock,
 507                    spamd_buffer + offset,
 508                    sizeof(spamd_buffer) - offset - 1,
 509                    sd->timeout - time(NULL) + start)) > 0 )
 510   offset += i;
 511
 512 /* error handling */
 513 if (i <= 0 && errno != 0)
 514   {
 515   log_write(0, LOG_MAIN|LOG_PANIC,
 516        "%s error reading from spamd socket: %s", loglabel, strerror(errno));
 517   (void)close(spamd_sock);
 518   return DEFER;
 519   }
 520
 521 /* reading done */
 522 (void)close(spamd_sock);
 523
 524 if (sd->is_rspamd)
 525   {                             /* rspamd variant of reply */
 526   int r;
 527   if ((r = sscanf(CS spamd_buffer,
 528           "RSPAMD/%7s 0 EX_OK\r\nMetric: default; %7s %lf / %lf / %lf\r\n%n",
 529           spamd_version, spamd_short_result, &spamd_score, &spamd_threshold,
 530           &spamd_reject_score, &spamd_report_offset)) != 5)
 531     {
 532     log_write(0, LOG_MAIN|LOG_PANIC,
 533               "%s cannot parse spamd output: %d", loglabel, r);
 534     return DEFER;
 535     }
 536   /* now parse action */
 537   p = &spamd_buffer[spamd_report_offset];
 538
 539   if (Ustrncmp(p, "Action: ", sizeof("Action: ") - 1) == 0)
 540     {
 541     p += sizeof("Action: ") - 1;
 542     q = &spam_action_buffer[0];
 543     while (*p && *p != '\r' && (q - spam_action_buffer) < sizeof(spam_action_buffer) - 1)
 544       *q++ = *p++;
 545     *q = '\0';
 546     }
 547   }
 548 else
 549   {                             /* spamassassin */
 550   /* dig in the spamd output and put the report in a multiline header,
 551   if requested */
 552   if (sscanf(CS spamd_buffer,
 553        "SPAMD/%7s 0 EX_OK\r\nContent-length: %*u\r\n\r\n%lf/%lf\r\n%n",
 554        spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 555     {
 556       /* try to fall back to pre-2.50 spamd output */
 557       if (sscanf(CS spamd_buffer,
 558            "SPAMD/%7s 0 EX_OK\r\nSpam: %*s ; %lf / %lf\r\n\r\n%n",
 559            spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 560         {
 561         log_write(0, LOG_MAIN|LOG_PANIC,
 562                   "%s cannot parse spamd output", loglabel);
 563         return DEFER;
 564         }
 565     }
 566
 567   Ustrcpy(spam_action_buffer,
 568     spamd_score >= spamd_threshold ? "reject" : "no action");
 569   }
 570
 571 /* Create report. Since this is a multiline string,
 572 we must hack it into shape first */
 573 p = &spamd_buffer[spamd_report_offset];
 574 q = spam_report_buffer;
 575 while (*p != '\0')
 576   {
 577   /* skip \r */
 578   if (*p == '\r')
 579     {
 580     p++;
 581     continue;
 582     }
 583   *q++ = *p;
 584   if (*p++ == '\n')
 585     {
 586     /* add an extra space after the newline to ensure
 587     that it is treated as a header continuation line */
 588     *q++ = ' ';
 589     }
 590   }
 591 /* NULL-terminate */
 592 *q-- = '\0';
 593 /* cut off trailing leftovers */
 594 while (*q <= ' ')
 595   *q-- = '\0';
 596
 597 spam_report = spam_report_buffer;
 598 spam_action = spam_action_buffer;
 599
 600 /* create spam bar */
 601 spamd_score_char = spamd_score > 0 ? '+' : '-';
 602 j = abs((int)(spamd_score));
 603 i = 0;
 604 if (j != 0)
 605   while ((i < j) && (i <= MAX_SPAM_BAR_CHARS))
 606      spam_bar_buffer[i++] = spamd_score_char;
 607 else
 608   {
 609   spam_bar_buffer[0] = '/';
 610   i = 1;
 611   }
 612 spam_bar_buffer[i] = '\0';
 613 spam_bar = spam_bar_buffer;
 614
 615 /* create "float" spam score */
 616 (void)string_format(spam_score_buffer, sizeof(spam_score_buffer),
 617         "%.1f", spamd_score);
 618 spam_score = spam_score_buffer;
 619
 620 /* create "int" spam score */
 621 j = (int)((spamd_score + 0.001)*10);
 622 (void)string_format(spam_score_int_buffer, sizeof(spam_score_int_buffer),
 623         "%d", j);
 624 spam_score_int = spam_score_int_buffer;
 625
 626 /* compare threshold against score */
 627 spam_rc = spamd_score >= spamd_threshold
 628   ? OK  /* spam as determined by user's threshold */
 629   : FAIL;       /* not spam */
 630
 631 /* remember expanded spamd_address if needed */
 632 if (spamd_address_work != spamd_address)
 633   prev_spamd_address_work = string_copy(spamd_address_work);
 634
 635 /* remember user name and "been here" for it */
 636 Ustrcpy(prev_user_name, user_name);
 637 spam_ok = 1;
 638
 639 return override
 640   ? OK          /* always return OK, no matter what the score */
 641   : spam_rc;
 642
 643 defer:
 644   (void)fclose(mbox_file);
 645   return DEFER;
 646 }
 647
 648 #endif
 649 /* vi: aw ai sw=2
 650 */