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 int
  25 spam(uschar **listptr)
  26 {
  27   int sep = 0;
  28   uschar *list = *listptr;
  29   uschar *user_name;
  30   uschar user_name_buffer[128];
  31   unsigned long mbox_size;
  32   FILE *mbox_file;
  33   int spamd_sock = -1;
  34   uschar spamd_buffer[32600];
  35   int i, j, offset, result;
  36   BOOL is_rspamd;
  37   uschar spamd_version[8];
  38   uschar spamd_short_result[8];
  39   uschar spamd_score_char;
  40   double spamd_threshold, spamd_score, spamd_reject_score;
  41   int spamd_report_offset;
  42   uschar *p,*q;
  43   int override = 0;
  44   time_t start;
  45   size_t read, wrote;
  46   struct sockaddr_un server;
  47 #ifndef NO_POLL_H
  48   struct pollfd pollfd;
  49 #else                               /* Patch posted by Erik ? for OS X */
  50   struct timeval select_tv;         /* and applied by PH */
  51   fd_set select_fd;
  52 #endif
  53   uschar *spamd_address_work;
  54   static const uschar * loglabel = US"spam acl condition:";
  55
  56   /* stop compiler warning */
  57   result = 0;
  58
  59   /* find the username from the option list */
  60   if ((user_name = string_nextinlist(&list, &sep,
  61                                      user_name_buffer,
  62                                      sizeof(user_name_buffer))) == NULL)
  63     {
  64     /* no username given, this means no scanning should be done */
  65     return FAIL;
  66     }
  67
  68   /* if username is "0" or "false", do not scan */
  69   if ( (Ustrcmp(user_name,"0") == 0) ||
  70        (strcmpic(user_name,US"false") == 0) )
  71     return FAIL;
  72
  73   /* if there is an additional option, check if it is "true" */
  74   if (strcmpic(list,US"true") == 0)
  75     /* in that case, always return true later */
  76     override = 1;
  77
  78   /* expand spamd_address if needed */
  79   if (*spamd_address == '$')
  80     {
  81     spamd_address_work = expand_string(spamd_address);
  82     if (spamd_address_work == NULL)
  83       {
  84       log_write(0, LOG_MAIN|LOG_PANIC,
  85         "%s spamd_address starts with $, but expansion failed: %s",
  86         loglabel, expand_string_message);
  87       return DEFER;
  88       }
  89     }
  90   else
  91     spamd_address_work = spamd_address;
  92
  93   /* check if previous spamd_address was expanded and has changed. dump cached results if so */
  94   if (  spam_ok
  95      && prev_spamd_address_work != NULL
  96      && Ustrcmp(prev_spamd_address_work, spamd_address_work) != 0
  97      )
  98     spam_ok = 0;
  99
 100   /* if we scanned for this username last time, just return */
 101   if (spam_ok && Ustrcmp(prev_user_name, user_name) == 0)
 102     return override ? OK : spam_rc;
 103
 104   /* make sure the eml mbox file is spooled up */
 105   mbox_file = spool_mbox(&mbox_size, NULL);
 106
 107   if (mbox_file == NULL)
 108     {
 109     /* error while spooling */
 110     log_write(0, LOG_MAIN|LOG_PANIC,
 111            "%s error while creating mbox spool file", loglabel);
 112     return DEFER;
 113     }
 114
 115   start = time(NULL);
 116
 117   /* socket does not start with '/' -> network socket */
 118   if (*spamd_address_work != '/')
 119     {
 120     int num_servers = 0;
 121     int current_server;
 122     uschar *address = NULL;
 123     uschar *spamd_address_list_ptr = spamd_address_work;
 124     uschar address_buffer[256];
 125     spamd_address_container * spamd_address_vector[32];
 126
 127     /* Check how many spamd servers we have
 128        and register their addresses */
 129     while ((address = string_nextinlist(&spamd_address_list_ptr, &sep,
 130                                         address_buffer,
 131                                         sizeof(address_buffer))) != NULL)
 132       {
 133
 134       /* Potential memory leak as we never free the store. */
 135       spamd_address_container *this_spamd =
 136         (spamd_address_container *)store_get(sizeof(spamd_address_container));
 137
 138       /* Check for spamd variant */
 139       this_spamd->is_rspamd = Ustrstr(address, "variant=rspamd") != NULL;
 140
 141       /* grok spamd address and port */
 142       if (sscanf(CS address, "%23s %hu", this_spamd->tcp_addr, &this_spamd->tcp_port) != 2)
 143         {
 144         log_write(0, LOG_MAIN,
 145           "%s warning - invalid spamd address: '%s'", loglabel, address);
 146         continue;
 147         }
 148
 149       spamd_address_vector[num_servers] = this_spamd;
 150       if (  ++num_servers
 151          >= sizeof(spamd_address_vector)/sizeof(spamd_address_vector[0]))
 152         break;
 153       }
 154
 155     /* check if we have at least one server */
 156     if (!num_servers)
 157       {
 158       log_write(0, LOG_MAIN|LOG_PANIC,
 159          "%s no useable spamd server addresses in spamd_address configuration option.",
 160          loglabel);
 161       (void)fclose(mbox_file);
 162       return DEFER;
 163       }
 164
 165     while (num_servers > 0)
 166       {
 167       int i;
 168
 169       /* Randomly pick a server to try */
 170       current_server = random_number(num_servers);
 171
 172       debug_printf("trying server %s, port %u\n",
 173                    spamd_address_vector[current_server]->tcp_addr,
 174                    spamd_address_vector[current_server]->tcp_port);
 175
 176       /* contact a spamd */
 177       if ((spamd_sock = ip_socket(SOCK_STREAM, AF_INET)) < 0)
 178         {
 179         log_write(0, LOG_MAIN|LOG_PANIC,
 180            "%s error creating IP socket for spamd", loglabel);
 181         (void)fclose(mbox_file);
 182         return DEFER;
 183         }
 184
 185       if (ip_connect(spamd_sock,
 186                      AF_INET,
 187                      spamd_address_vector[current_server]->tcp_addr,
 188                      spamd_address_vector[current_server]->tcp_port,
 189                      5 ) > -1)
 190         {
 191         /* connection OK */
 192         is_rspamd = spamd_address_vector[current_server]->is_rspamd;
 193         break;
 194         }
 195
 196       log_write(0, LOG_MAIN|LOG_PANIC,
 197          "%s warning - spamd connection to %s, port %u failed: %s",
 198          loglabel,
 199          spamd_address_vector[current_server]->tcp_addr,
 200          spamd_address_vector[current_server]->tcp_port,
 201          strerror(errno));
 202
 203       (void)close(spamd_sock);
 204
 205       /* Remove the server from the list. XXX We should free the memory */
 206       num_servers--;
 207       for (i = current_server; i < num_servers; i++)
 208         spamd_address_vector[i] = spamd_address_vector[i+1];
 209       }
 210
 211     if (num_servers == 0)
 212       {
 213       log_write(0, LOG_MAIN|LOG_PANIC, "%s all spamd servers failed", loglabel);
 214       (void)fclose(mbox_file);
 215       return DEFER;
 216       }
 217     }
 218   else
 219     {
 220     /* open the local socket */
 221
 222     if ((spamd_sock = socket(AF_UNIX, SOCK_STREAM, 0)) < 0)
 223       {
 224       log_write(0, LOG_MAIN|LOG_PANIC,
 225                 "%s spamd: unable to acquire socket (%s)",
 226                 loglabel,
 227                 strerror(errno));
 228       (void)fclose(mbox_file);
 229       return DEFER;
 230       }
 231
 232     server.sun_family = AF_UNIX;
 233
 234     is_rspamd = (p = Ustrstr(spamd_address_work, "variant=rspamd")) != NULL;
 235     if (is_rspamd)
 236       {
 237       /* strip spaces */
 238       p--;
 239       while (p > spamd_address_work && isspace (*p))
 240         p--;
 241       Ustrncpy(server.sun_path, spamd_address_work, p - spamd_address_work + 1);
 242       /* zero terminate */
 243       server.sun_path[p - spamd_address_work + 1] = 0;
 244       }
 245     else
 246       Ustrcpy(server.sun_path, spamd_address_work);
 247
 248     if (connect(spamd_sock, (struct sockaddr *) &server, sizeof(struct sockaddr_un)) < 0)
 249       {
 250       log_write(0, LOG_MAIN|LOG_PANIC,
 251                 "%s spamd: unable to connect to UNIX socket %s (%s)",
 252                 loglabel,
 253                 server.sun_path, strerror(errno) );
 254       (void)fclose(mbox_file);
 255       (void)close(spamd_sock);
 256       return DEFER;
 257       }
 258     }
 259
 260   if (spamd_sock == -1)
 261     {
 262     log_write(0, LOG_MAIN|LOG_PANIC,
 263         "programming fault, spamd_sock unexpectedly unset");
 264     (void)fclose(mbox_file);
 265     (void)close(spamd_sock);
 266     return DEFER;
 267     }
 268
 269   (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 270   /* now we are connected to spamd on spamd_sock */
 271   if (is_rspamd)
 272     {                           /* rspamd variant */
 273     uschar *req_str;
 274     const char *helo;
 275     const char *fcrdns;
 276
 277     req_str = string_sprintf("CHECK RSPAMC/1.3\r\nContent-length: %lu\r\n"
 278       "Queue-Id: %s\r\nFrom: <%s>\r\nRecipient-Number: %d\r\n", mbox_size,
 279       message_id, sender_address, recipients_count);
 280     for (i = 0; i < recipients_count; i ++)
 281       req_str = string_sprintf("%sRcpt: <%s>\r\n", req_str, recipients_list[i].address);
 282     if ((helo = expand_string(US"$sender_helo_name")) != NULL && *helo != '\0')
 283       req_str = string_sprintf("%sHelo: %s\r\n", req_str, helo);
 284     if ((fcrdns = expand_string(US"$sender_host_name")) != NULL && *fcrdns != '\0')
 285       req_str = string_sprintf("%sHostname: %s\r\n", req_str, fcrdns);
 286     if (sender_host_address != NULL)
 287       req_str = string_sprintf("%sIP: %s\r\n", req_str, sender_host_address);
 288     req_str = string_sprintf("%s\r\n", req_str);
 289     wrote = send(spamd_sock, req_str, Ustrlen(req_str), 0);
 290     }
 291     else
 292     {                           /* spamassassin variant */
 293     (void)string_format(spamd_buffer,
 294             sizeof(spamd_buffer),
 295             "REPORT SPAMC/1.2\r\nUser: %s\r\nContent-length: %ld\r\n\r\n",
 296             user_name,
 297             mbox_size);
 298     /* send our request */
 299     wrote = send(spamd_sock, spamd_buffer, Ustrlen(spamd_buffer), 0);
 300     }
 301   if (wrote == -1)
 302     {
 303     (void)close(spamd_sock);
 304     log_write(0, LOG_MAIN|LOG_PANIC,
 305          "%s spamd send failed: %s", loglabel, strerror(errno));
 306     (void)fclose(mbox_file);
 307     (void)close(spamd_sock);
 308     return DEFER;
 309     }
 310
 311   /* now send the file */
 312   /* spamd sometimes accepts conections but doesn't read data off
 313    * the connection.  We make the file descriptor non-blocking so
 314    * that the write will only write sufficient data without blocking
 315    * and we poll the desciptor to make sure that we can write without
 316    * blocking.  Short writes are gracefully handled and if the whole
 317    * trasaction takes too long it is aborted.
 318    * Note: poll() is not supported in OSX 10.2 and is reported to be
 319    *       broken in more recent versions (up to 10.4).
 320    */
 321 #ifndef NO_POLL_H
 322   pollfd.fd = spamd_sock;
 323   pollfd.events = POLLOUT;
 324 #endif
 325   (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 326   do
 327     {
 328     read = fread(spamd_buffer,1,sizeof(spamd_buffer),mbox_file);
 329     if (read > 0)
 330       {
 331       offset = 0;
 332 again:
 333 #ifndef NO_POLL_H
 334       result = poll(&pollfd, 1, 1000);
 335
 336 /* Patch posted by Erik ? for OS X and applied by PH */
 337 #else
 338       select_tv.tv_sec = 1;
 339       select_tv.tv_usec = 0;
 340       FD_ZERO(&select_fd);
 341       FD_SET(spamd_sock, &select_fd);
 342       result = select(spamd_sock+1, NULL, &select_fd, NULL, &select_tv);
 343 #endif
 344 /* End Erik's patch */
 345
 346       if (result == -1 && errno == EINTR)
 347         goto again;
 348       else if (result < 1)
 349         {
 350         if (result == -1)
 351           log_write(0, LOG_MAIN|LOG_PANIC,
 352             "%s %s on spamd socket", loglabel, strerror(errno));
 353         else
 354           {
 355           if (time(NULL) - start < SPAMD_TIMEOUT)
 356             goto again;
 357           log_write(0, LOG_MAIN|LOG_PANIC,
 358             "%s timed out writing spamd socket", loglabel);
 359           }
 360         (void)close(spamd_sock);
 361         (void)fclose(mbox_file);
 362         return DEFER;
 363         }
 364
 365       wrote = send(spamd_sock,spamd_buffer + offset,read - offset,0);
 366       if (wrote == -1)
 367         {
 368         log_write(0, LOG_MAIN|LOG_PANIC,
 369             "%s %s on spamd socket", loglabel, strerror(errno));
 370         (void)close(spamd_sock);
 371         (void)fclose(mbox_file);
 372         return DEFER;
 373         }
 374       if (offset + wrote != read)
 375         {
 376         offset += wrote;
 377         goto again;
 378         }
 379       }
 380     }
 381   while (!feof(mbox_file) && !ferror(mbox_file));
 382
 383   if (ferror(mbox_file))
 384     {
 385     log_write(0, LOG_MAIN|LOG_PANIC,
 386       "%s error reading spool file: %s", loglabel, strerror(errno));
 387     (void)close(spamd_sock);
 388     (void)fclose(mbox_file);
 389     return DEFER;
 390     }
 391
 392   (void)fclose(mbox_file);
 393
 394   /* we're done sending, close socket for writing */
 395   shutdown(spamd_sock,SHUT_WR);
 396
 397   /* read spamd response using what's left of the timeout.
 398    */
 399   memset(spamd_buffer, 0, sizeof(spamd_buffer));
 400   offset = 0;
 401   while ((i = ip_recv(spamd_sock,
 402                      spamd_buffer + offset,
 403                      sizeof(spamd_buffer) - offset - 1,
 404                      SPAMD_TIMEOUT - time(NULL) + start)) > 0 )
 405     offset += i;
 406
 407   /* error handling */
 408   if (i <= 0 && errno != 0)
 409     {
 410     log_write(0, LOG_MAIN|LOG_PANIC,
 411          "%s error reading from spamd socket: %s", loglabel, strerror(errno));
 412     (void)close(spamd_sock);
 413     return DEFER;
 414     }
 415
 416   /* reading done */
 417   (void)close(spamd_sock);
 418
 419   if (is_rspamd)
 420     {                           /* rspamd variant of reply */
 421     int r;
 422     if ((r = sscanf(CS spamd_buffer,
 423             "RSPAMD/%7s 0 EX_OK\r\nMetric: default; %7s %lf / %lf / %lf\r\n%n",
 424             spamd_version, spamd_short_result, &spamd_score, &spamd_threshold,
 425             &spamd_reject_score, &spamd_report_offset)) != 5)
 426       {
 427         log_write(0, LOG_MAIN|LOG_PANIC,
 428                   "%s cannot parse spamd output: %d", loglabel, r);
 429         return DEFER;
 430       }
 431     /* now parse action */
 432     p = &spamd_buffer[spamd_report_offset];
 433
 434     if (Ustrncmp(p, "Action: ", sizeof("Action: ") - 1) == 0)
 435       {
 436       p += sizeof("Action: ") - 1;
 437       q = &spam_action_buffer[0];
 438       while (*p && *p != '\r' && (q - spam_action_buffer) < sizeof(spam_action_buffer) - 1)
 439         *q++ = *p++;
 440       *q = '\0';
 441       }
 442     }
 443   else
 444     {                           /* spamassassin */
 445     /* dig in the spamd output and put the report in a multiline header,
 446     if requested */
 447     if (sscanf(CS spamd_buffer,
 448          "SPAMD/%7s 0 EX_OK\r\nContent-length: %*u\r\n\r\n%lf/%lf\r\n%n",
 449          spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 450       {
 451         /* try to fall back to pre-2.50 spamd output */
 452         if (sscanf(CS spamd_buffer,
 453              "SPAMD/%7s 0 EX_OK\r\nSpam: %*s ; %lf / %lf\r\n\r\n%n",
 454              spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 455           {
 456           log_write(0, LOG_MAIN|LOG_PANIC,
 457                     "%s cannot parse spamd output", loglabel);
 458           return DEFER;
 459           }
 460       }
 461
 462     Ustrcpy(spam_action_buffer,
 463       spamd_score >= spamd_threshold ? "reject" : "no action");
 464     }
 465
 466   /* Create report. Since this is a multiline string,
 467   we must hack it into shape first */
 468   p = &spamd_buffer[spamd_report_offset];
 469   q = spam_report_buffer;
 470   while (*p != '\0')
 471     {
 472     /* skip \r */
 473     if (*p == '\r')
 474       {
 475       p++;
 476       continue;
 477       }
 478     *q++ = *p;
 479     if (*p++ == '\n')
 480       {
 481       /* add an extra space after the newline to ensure
 482       that it is treated as a header continuation line */
 483       *q++ = ' ';
 484       }
 485     }
 486   /* NULL-terminate */
 487   *q-- = '\0';
 488   /* cut off trailing leftovers */
 489   while (*q <= ' ')
 490     *q-- = '\0';
 491
 492   spam_report = spam_report_buffer;
 493   spam_action = spam_action_buffer;
 494
 495   /* create spam bar */
 496   spamd_score_char = spamd_score > 0 ? '+' : '-';
 497   j = abs((int)(spamd_score));
 498   i = 0;
 499   if (j != 0)
 500     while ((i < j) && (i <= MAX_SPAM_BAR_CHARS))
 501        spam_bar_buffer[i++] = spamd_score_char;
 502   else
 503     {
 504     spam_bar_buffer[0] = '/';
 505     i = 1;
 506     }
 507   spam_bar_buffer[i] = '\0';
 508   spam_bar = spam_bar_buffer;
 509
 510   /* create "float" spam score */
 511   (void)string_format(spam_score_buffer, sizeof(spam_score_buffer),
 512           "%.1f", spamd_score);
 513   spam_score = spam_score_buffer;
 514
 515   /* create "int" spam score */
 516   j = (int)((spamd_score + 0.001)*10);
 517   (void)string_format(spam_score_int_buffer, sizeof(spam_score_int_buffer),
 518           "%d", j);
 519   spam_score_int = spam_score_int_buffer;
 520
 521   /* compare threshold against score */
 522   spam_rc = spamd_score >= spamd_threshold
 523     ? OK        /* spam as determined by user's threshold */
 524     : FAIL;     /* not spam */
 525
 526   /* remember expanded spamd_address if needed */
 527   if (spamd_address_work != spamd_address)
 528     prev_spamd_address_work = string_copy(spamd_address_work);
 529
 530   /* remember user name and "been here" for it */
 531   Ustrcpy(prev_user_name, user_name);
 532   spam_ok = 1;
 533
 534   return override
 535     ? OK                /* always return OK, no matter what the score */
 536     : spam_rc;
 537 }
 538
 539 #endif