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 backup value */
  31 spamd->is_failed = FALSE;
  32 spamd->is_backup = FALSE;
  33 spamd->weight = SPAMD_WEIGHT;
  34 spamd->timeout = SPAMD_TIMEOUT;
  35 spamd->retry = 0;
  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 /* check backup parameter */
  48 if (Ustrcmp(param, "backup") == 0)
  49   {
  50   spamd->is_backup = TRUE;
  51   return 0; /* OK */
  52   }
  53
  54 /*XXX more clever parsing could discard embedded spaces? */
  55
  56 /* check weight parameter */
  57 if (sscanf(param, "weight=%u", &spamd->weight))
  58   {
  59   if (spamd->weight == 0) /* this server disabled: skip it */
  60     return 1;
  61   return 0; /* OK */
  62   }
  63
  64 /* check time parameter */
  65 if (Ustrncmp(param, "time=", 5) == 0)
  66   {
  67   unsigned int start_h = 0, start_m = 0, start_s = 0;
  68   unsigned int end_h = 24, end_m = 0, end_s = 0;
  69   unsigned int time_start, time_end;
  70   const uschar * end_string;
  71
  72   name = US"time";
  73   s = param+5;
  74   if ((end_string = Ustrchr(s, '-')))
  75     {
  76     end_string++;
  77     if (  sscanf(CS end_string, "%u.%u.%u", &end_h,   &end_m,   &end_s)   == 0
  78        || sscanf(CS s,          "%u.%u.%u", &start_h, &start_m, &start_s) == 0
  79        )
  80       goto badval;
  81     }
  82   else
  83     goto badval;
  84
  85   if (timesinceday < 0)
  86     {
  87     time_t now = time(NULL);
  88     struct tm *tmp = localtime(&now);
  89     timesinceday = tmp->tm_hour*3600 + tmp->tm_min*60 + tmp->tm_sec;
  90     }
  91
  92   time_start = start_h*3600 + start_m*60 + start_s;
  93   time_end = end_h*3600 + end_m*60 + end_s;
  94
  95   if (timesinceday < time_start || timesinceday >= time_end)
  96     return 1; /* skip spamd server */
  97
  98   return 0; /* OK */
  99   }
 100
 101 if (Ustrcmp(param, "variant=rspamd") == 0)
 102   {
 103   spamd->is_rspamd = TRUE;
 104   return 0;
 105   }
 106
 107 if (Ustrncmp(param, "tmo=", 4) == 0)
 108   {
 109   int sec = readconf_readtime((s = param+4), '\0', FALSE);
 110   name = US"timeout";
 111   if (sec < 0)
 112     goto badval;
 113   spamd->timeout = sec;
 114   return 0;
 115   }
 116
 117 if (Ustrncmp(param, "retry=", 6) == 0)
 118   {
 119   int sec = readconf_readtime((s = param+6), '\0', FALSE);
 120   name = US"retry";
 121   if (sec < 0)
 122     goto badval;
 123   spamd->retry = sec;
 124   return 0;
 125   }
 126
 127 log_write(0, LOG_MAIN, "%s warning - invalid spamd parameter: '%s'",
 128   loglabel, param);
 129 return -1; /* syntax error */
 130
 131 badval:
 132   log_write(0, LOG_MAIN,
 133     "%s warning - invalid spamd %s value: '%s'", loglabel, name, s);
 134   return -1; /* syntax error */
 135 }
 136
 137
 138 static int
 139 spamd_get_server(spamd_address_container **spamds, int num_servers)
 140 {
 141 unsigned int i;
 142 long rnd, weights = 0;
 143 static BOOL srandomed = 0;
 144 BOOL usebackup = FALSE;
 145
 146 for (;;)
 147   {
 148   /* seedup, if we have only 1 server */
 149   if (num_servers == 1)
 150     return (spamds[0]->is_failed ? -1 : 0);
 151
 152   /* init ranmod */
 153   if (!srandomed)
 154     {
 155     struct timeval tv;
 156     gettimeofday(&tv, NULL);
 157     srandom((unsigned int)(tv.tv_usec/1000));
 158     srandomed = TRUE;
 159     }
 160
 161   /* get sum of all weights */
 162   for (i = 0; i < num_servers; i++)
 163     if (!spamds[i]->is_failed && spamds[i]->is_backup == usebackup)
 164       weights += spamds[i]->weight;
 165
 166   if (weights != 0)
 167     break;
 168   if (usebackup)        /* all servers failed (backups too) */
 169     return -1;
 170   usebackup = TRUE;
 171   }
 172
 173 rnd = random() % weights;
 174
 175 for (i = 0; i < num_servers; i++)
 176   if (!spamds[i]->is_failed && spamds[i]->is_backup == usebackup)
 177     if ((rnd -= spamds[i]->weight) < 0)
 178       return i;
 179
 180 log_write(0, LOG_MAIN|LOG_PANIC,
 181   "%s unknown error (memory/cpu corruption?)", loglabel);
 182 return -1;
 183 }
 184
 185
 186 int
 187 spam(const uschar **listptr)
 188 {
 189 int sep = 0;
 190 const uschar *list = *listptr;
 191 uschar *user_name;
 192 uschar user_name_buffer[128];
 193 unsigned long mbox_size;
 194 FILE *mbox_file;
 195 int spamd_sock = -1;
 196 uschar spamd_buffer[32600];
 197 int i, j, offset, result;
 198 uschar spamd_version[8];
 199 uschar spamd_short_result[8];
 200 uschar spamd_score_char;
 201 double spamd_threshold, spamd_score, spamd_reject_score;
 202 int spamd_report_offset;
 203 uschar *p,*q;
 204 int override = 0;
 205 time_t start;
 206 size_t read, wrote;
 207 struct sockaddr_un server;
 208 #ifndef NO_POLL_H
 209 struct pollfd pollfd;
 210 #else                               /* Patch posted by Erik ? for OS X */
 211 struct timeval select_tv;         /* and applied by PH */
 212 fd_set select_fd;
 213 #endif
 214 uschar *spamd_address_work;
 215 spamd_address_container * sd;
 216
 217 /* stop compiler warning */
 218 result = 0;
 219
 220 /* find the username from the option list */
 221 if ((user_name = string_nextinlist(&list, &sep,
 222                                    user_name_buffer,
 223                                    sizeof(user_name_buffer))) == NULL)
 224   {
 225   /* no username given, this means no scanning should be done */
 226   return FAIL;
 227   }
 228
 229 /* if username is "0" or "false", do not scan */
 230 if ( (Ustrcmp(user_name,"0") == 0) ||
 231      (strcmpic(user_name,US"false") == 0) )
 232   return FAIL;
 233
 234 /* if there is an additional option, check if it is "true" */
 235 if (strcmpic(list,US"true") == 0)
 236   /* in that case, always return true later */
 237   override = 1;
 238
 239 /* expand spamd_address if needed */
 240 if (*spamd_address == '$')
 241   {
 242   spamd_address_work = expand_string(spamd_address);
 243   if (spamd_address_work == NULL)
 244     {
 245     log_write(0, LOG_MAIN|LOG_PANIC,
 246       "%s spamd_address starts with $, but expansion failed: %s",
 247       loglabel, expand_string_message);
 248     return DEFER;
 249     }
 250   }
 251 else
 252   spamd_address_work = spamd_address;
 253
 254 HDEBUG(D_acl) debug_printf("spamd: addrlist '%s'\n", spamd_address_work);
 255
 256 /* check if previous spamd_address was expanded and has changed. dump cached results if so */
 257 if (  spam_ok
 258    && prev_spamd_address_work != NULL
 259    && Ustrcmp(prev_spamd_address_work, spamd_address_work) != 0
 260    )
 261   spam_ok = 0;
 262
 263 /* if we scanned for this username last time, just return */
 264 if (spam_ok && Ustrcmp(prev_user_name, user_name) == 0)
 265   return override ? OK : spam_rc;
 266
 267 /* make sure the eml mbox file is spooled up */
 268 mbox_file = spool_mbox(&mbox_size, NULL);
 269
 270 if (mbox_file == NULL)
 271   {
 272   /* error while spooling */
 273   log_write(0, LOG_MAIN|LOG_PANIC,
 274          "%s error while creating mbox spool file", loglabel);
 275   return DEFER;
 276   }
 277
 278 start = time(NULL);
 279
 280   {
 281   int num_servers = 0;
 282   int current_server;
 283   uschar *address;
 284   const uschar *spamd_address_list_ptr = spamd_address_work;
 285   spamd_address_container * spamd_address_vector[32];
 286
 287   /* Check how many spamd servers we have
 288      and register their addresses */
 289   while ((address = string_nextinlist(&spamd_address_list_ptr, &sep,
 290                                       NULL, 0)) != NULL)
 291     {
 292     const uschar * sublist;
 293     int sublist_sep = -(int)' ';        /* default space-sep */
 294     unsigned args;
 295     uschar * s;
 296
 297     HDEBUG(D_acl) debug_printf("spamd: addr entry '%s'\n", address);
 298     sd = (spamd_address_container *)store_get(sizeof(spamd_address_container));
 299
 300     for (sublist = address, args = 0, spamd_param_init(sd);
 301          s = string_nextinlist(&sublist, &sublist_sep, NULL, 0);
 302          args++
 303          )
 304       {
 305         HDEBUG(D_acl) debug_printf("spamd:  addr parm '%s'\n", s);
 306         switch (args)
 307         {
 308         case 0:   sd->hostspec = s;
 309                   if (*s == '/') args++;        /* local; no port */
 310                   break;
 311         case 1:   sd->hostspec = string_sprintf("%s %s", sd->hostspec, s);
 312                   break;
 313         default:  spamd_param(s, sd);
 314                   break;
 315         }
 316       }
 317     if (args < 2)
 318       {
 319       log_write(0, LOG_MAIN,
 320         "%s warning - invalid spamd address: '%s'", loglabel, address);
 321       continue;
 322       }
 323
 324     spamd_address_vector[num_servers] = sd;
 325     if (++num_servers > 31)
 326       break;
 327     }
 328
 329   /* check if we have at least one server */
 330   if (!num_servers)
 331     {
 332     log_write(0, LOG_MAIN|LOG_PANIC,
 333        "%s no useable spamd server addresses in spamd_address configuration option.",
 334        loglabel);
 335     goto defer;
 336     }
 337
 338   current_server = spamd_get_server(spamd_address_vector, num_servers);
 339   sd = spamd_address_vector[current_server];
 340   for(;;)
 341     {
 342     uschar * errstr;
 343
 344     debug_printf("trying server %s\n", sd->hostspec);
 345
 346     for (;;)
 347       {
 348       if (  (spamd_sock = ip_streamsocket(sd->hostspec, &errstr, 5)) >= 0
 349          || sd->retry <= 0
 350          )
 351         break;
 352       debug_printf("server %s: retry conn\n", sd->hostspec);
 353       while (sd->retry > 0) sd->retry = sleep(sd->retry);
 354       }
 355     if (spamd_sock >= 0)
 356       break;
 357
 358     log_write(0, LOG_MAIN, "%s spamd: %s", loglabel, errstr);
 359     sd->is_failed = TRUE;
 360
 361     current_server = spamd_get_server(spamd_address_vector, num_servers);
 362     if (current_server < 0)
 363       {
 364       log_write(0, LOG_MAIN|LOG_PANIC, "%s all spamd servers failed", loglabel);
 365       goto defer;
 366       }
 367     sd = spamd_address_vector[current_server];
 368     }
 369   }
 370
 371 if (spamd_sock == -1)
 372   {
 373   log_write(0, LOG_MAIN|LOG_PANIC,
 374       "programming fault, spamd_sock unexpectedly unset");
 375   goto defer;
 376   }
 377
 378 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 379 /* now we are connected to spamd on spamd_sock */
 380 if (sd->is_rspamd)
 381   {                             /* rspamd variant */
 382   uschar *req_str;
 383   const char *helo;
 384   const char *fcrdns;
 385
 386   req_str = string_sprintf("CHECK RSPAMC/1.3\r\nContent-length: %lu\r\n"
 387     "Queue-Id: %s\r\nFrom: <%s>\r\nRecipient-Number: %d\r\n", mbox_size,
 388     message_id, sender_address, recipients_count);
 389   for (i = 0; i < recipients_count; i ++)
 390     req_str = string_sprintf("%sRcpt: <%s>\r\n", req_str, recipients_list[i].address);
 391   if ((helo = expand_string(US"$sender_helo_name")) != NULL && *helo != '\0')
 392     req_str = string_sprintf("%sHelo: %s\r\n", req_str, helo);
 393   if ((fcrdns = expand_string(US"$sender_host_name")) != NULL && *fcrdns != '\0')
 394     req_str = string_sprintf("%sHostname: %s\r\n", req_str, fcrdns);
 395   if (sender_host_address != NULL)
 396     req_str = string_sprintf("%sIP: %s\r\n", req_str, sender_host_address);
 397   req_str = string_sprintf("%s\r\n", req_str);
 398   wrote = send(spamd_sock, req_str, Ustrlen(req_str), 0);
 399   }
 400   else
 401   {                             /* spamassassin variant */
 402   (void)string_format(spamd_buffer,
 403           sizeof(spamd_buffer),
 404           "REPORT SPAMC/1.2\r\nUser: %s\r\nContent-length: %ld\r\n\r\n",
 405           user_name,
 406           mbox_size);
 407   /* send our request */
 408   wrote = send(spamd_sock, spamd_buffer, Ustrlen(spamd_buffer), 0);
 409   }
 410 if (wrote == -1)
 411   {
 412   (void)close(spamd_sock);
 413   log_write(0, LOG_MAIN|LOG_PANIC,
 414        "%s spamd send failed: %s", loglabel, strerror(errno));
 415   goto defer;
 416   }
 417
 418 /* now send the file */
 419 /* spamd sometimes accepts conections but doesn't read data off
 420  * the connection.  We make the file descriptor non-blocking so
 421  * that the write will only write sufficient data without blocking
 422  * and we poll the desciptor to make sure that we can write without
 423  * blocking.  Short writes are gracefully handled and if the whole
 424  * trasaction takes too long it is aborted.
 425  * Note: poll() is not supported in OSX 10.2 and is reported to be
 426  *       broken in more recent versions (up to 10.4).
 427  */
 428 #ifndef NO_POLL_H
 429 pollfd.fd = spamd_sock;
 430 pollfd.events = POLLOUT;
 431 #endif
 432 (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 433 do
 434   {
 435   read = fread(spamd_buffer,1,sizeof(spamd_buffer),mbox_file);
 436   if (read > 0)
 437     {
 438     offset = 0;
 439 again:
 440 #ifndef NO_POLL_H
 441     result = poll(&pollfd, 1, 1000);
 442
 443 /* Patch posted by Erik ? for OS X and applied by PH */
 444 #else
 445     select_tv.tv_sec = 1;
 446     select_tv.tv_usec = 0;
 447     FD_ZERO(&select_fd);
 448     FD_SET(spamd_sock, &select_fd);
 449     result = select(spamd_sock+1, NULL, &select_fd, NULL, &select_tv);
 450 #endif
 451 /* End Erik's patch */
 452
 453     if (result == -1 && errno == EINTR)
 454       goto again;
 455     else if (result < 1)
 456       {
 457       if (result == -1)
 458         log_write(0, LOG_MAIN|LOG_PANIC,
 459           "%s %s on spamd socket", loglabel, strerror(errno));
 460       else
 461         {
 462         if (time(NULL) - start < sd->timeout)
 463           goto again;
 464         log_write(0, LOG_MAIN|LOG_PANIC,
 465           "%s timed out writing spamd socket", loglabel);
 466         }
 467       (void)close(spamd_sock);
 468       goto defer;
 469       }
 470
 471     wrote = send(spamd_sock,spamd_buffer + offset,read - offset,0);
 472     if (wrote == -1)
 473       {
 474       log_write(0, LOG_MAIN|LOG_PANIC,
 475           "%s %s on spamd socket", loglabel, strerror(errno));
 476       (void)close(spamd_sock);
 477       goto defer;
 478       }
 479     if (offset + wrote != read)
 480       {
 481       offset += wrote;
 482       goto again;
 483       }
 484     }
 485   }
 486 while (!feof(mbox_file) && !ferror(mbox_file));
 487
 488 if (ferror(mbox_file))
 489   {
 490   log_write(0, LOG_MAIN|LOG_PANIC,
 491     "%s error reading spool file: %s", loglabel, strerror(errno));
 492   (void)close(spamd_sock);
 493   goto defer;
 494   }
 495
 496 (void)fclose(mbox_file);
 497
 498 /* we're done sending, close socket for writing */
 499 shutdown(spamd_sock,SHUT_WR);
 500
 501 /* read spamd response using what's left of the timeout.  */
 502 memset(spamd_buffer, 0, sizeof(spamd_buffer));
 503 offset = 0;
 504 while ((i = ip_recv(spamd_sock,
 505                    spamd_buffer + offset,
 506                    sizeof(spamd_buffer) - offset - 1,
 507                    sd->timeout - time(NULL) + start)) > 0 )
 508   offset += i;
 509
 510 /* error handling */
 511 if (i <= 0 && errno != 0)
 512   {
 513   log_write(0, LOG_MAIN|LOG_PANIC,
 514        "%s error reading from spamd socket: %s", loglabel, strerror(errno));
 515   (void)close(spamd_sock);
 516   return DEFER;
 517   }
 518
 519 /* reading done */
 520 (void)close(spamd_sock);
 521
 522 if (sd->is_rspamd)
 523   {                             /* rspamd variant of reply */
 524   int r;
 525   if ((r = sscanf(CS spamd_buffer,
 526           "RSPAMD/%7s 0 EX_OK\r\nMetric: default; %7s %lf / %lf / %lf\r\n%n",
 527           spamd_version, spamd_short_result, &spamd_score, &spamd_threshold,
 528           &spamd_reject_score, &spamd_report_offset)) != 5)
 529     {
 530       log_write(0, LOG_MAIN|LOG_PANIC,
 531                 "%s cannot parse spamd output: %d", loglabel, r);
 532       return DEFER;
 533     }
 534   /* now parse action */
 535   p = &spamd_buffer[spamd_report_offset];
 536
 537   if (Ustrncmp(p, "Action: ", sizeof("Action: ") - 1) == 0)
 538     {
 539     p += sizeof("Action: ") - 1;
 540     q = &spam_action_buffer[0];
 541     while (*p && *p != '\r' && (q - spam_action_buffer) < sizeof(spam_action_buffer) - 1)
 542       *q++ = *p++;
 543     *q = '\0';
 544     }
 545   }
 546 else
 547   {                             /* spamassassin */
 548   /* dig in the spamd output and put the report in a multiline header,
 549   if requested */
 550   if (sscanf(CS spamd_buffer,
 551        "SPAMD/%7s 0 EX_OK\r\nContent-length: %*u\r\n\r\n%lf/%lf\r\n%n",
 552        spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 553     {
 554       /* try to fall back to pre-2.50 spamd output */
 555       if (sscanf(CS spamd_buffer,
 556            "SPAMD/%7s 0 EX_OK\r\nSpam: %*s ; %lf / %lf\r\n\r\n%n",
 557            spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 558         {
 559         log_write(0, LOG_MAIN|LOG_PANIC,
 560                   "%s cannot parse spamd output", loglabel);
 561         return DEFER;
 562         }
 563     }
 564
 565   Ustrcpy(spam_action_buffer,
 566     spamd_score >= spamd_threshold ? "reject" : "no action");
 567   }
 568
 569 /* Create report. Since this is a multiline string,
 570 we must hack it into shape first */
 571 p = &spamd_buffer[spamd_report_offset];
 572 q = spam_report_buffer;
 573 while (*p != '\0')
 574   {
 575   /* skip \r */
 576   if (*p == '\r')
 577     {
 578     p++;
 579     continue;
 580     }
 581   *q++ = *p;
 582   if (*p++ == '\n')
 583     {
 584     /* add an extra space after the newline to ensure
 585     that it is treated as a header continuation line */
 586     *q++ = ' ';
 587     }
 588   }
 589 /* NULL-terminate */
 590 *q-- = '\0';
 591 /* cut off trailing leftovers */
 592 while (*q <= ' ')
 593   *q-- = '\0';
 594
 595 spam_report = spam_report_buffer;
 596 spam_action = spam_action_buffer;
 597
 598 /* create spam bar */
 599 spamd_score_char = spamd_score > 0 ? '+' : '-';
 600 j = abs((int)(spamd_score));
 601 i = 0;
 602 if (j != 0)
 603   while ((i < j) && (i <= MAX_SPAM_BAR_CHARS))
 604      spam_bar_buffer[i++] = spamd_score_char;
 605 else
 606   {
 607   spam_bar_buffer[0] = '/';
 608   i = 1;
 609   }
 610 spam_bar_buffer[i] = '\0';
 611 spam_bar = spam_bar_buffer;
 612
 613 /* create "float" spam score */
 614 (void)string_format(spam_score_buffer, sizeof(spam_score_buffer),
 615         "%.1f", spamd_score);
 616 spam_score = spam_score_buffer;
 617
 618 /* create "int" spam score */
 619 j = (int)((spamd_score + 0.001)*10);
 620 (void)string_format(spam_score_int_buffer, sizeof(spam_score_int_buffer),
 621         "%d", j);
 622 spam_score_int = spam_score_int_buffer;
 623
 624 /* compare threshold against score */
 625 spam_rc = spamd_score >= spamd_threshold
 626   ? OK  /* spam as determined by user's threshold */
 627   : FAIL;       /* not spam */
 628
 629 /* remember expanded spamd_address if needed */
 630 if (spamd_address_work != spamd_address)
 631   prev_spamd_address_work = string_copy(spamd_address_work);
 632
 633 /* remember user name and "been here" for it */
 634 Ustrcpy(prev_user_name, user_name);
 635 spam_ok = 1;
 636
 637 return override
 638   ? OK          /* always return OK, no matter what the score */
 639   : spam_rc;
 640
 641 defer:
 642   (void)fclose(mbox_file);
 643   return DEFER;
 644 }
 645
 646 #endif
 647 /* vi: aw ai sw=2
 648 */