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  * Copyright (c) The Exim Maintainers 2016 - 2018
   8  */
   9
  10 /* Code for calling spamassassin's spamd. Called from acl.c. */
  11
  12 #include "exim.h"
  13 #ifdef WITH_CONTENT_SCAN
  14 #include "spam.h"
  15
  16 uschar spam_score_buffer[16];
  17 uschar spam_score_int_buffer[16];
  18 uschar spam_bar_buffer[128];
  19 uschar spam_action_buffer[32];
  20 uschar spam_report_buffer[32600];
  21 uschar prev_user_name[128] = "";
  22 int spam_ok = 0;
  23 int spam_rc = 0;
  24 uschar *prev_spamd_address_work = NULL;
  25
  26 static const uschar * loglabel = US"spam acl condition:";
  27
  28
  29 static int
  30 spamd_param_init(spamd_address_container *spamd)
  31 {
  32 /* default spamd server weight, time and priority value */
  33 spamd->is_failed = FALSE;
  34 spamd->weight = SPAMD_WEIGHT;
  35 spamd->timeout = SPAMD_TIMEOUT;
  36 spamd->retry = 0;
  37 spamd->priority = 1;
  38 return 0;
  39 }
  40
  41
  42 static int
  43 spamd_param(const uschar * param, spamd_address_container * spamd)
  44 {
  45 static int timesinceday = -1;
  46 const uschar * s;
  47 const uschar * name;
  48
  49 /*XXX more clever parsing could discard embedded spaces? */
  50
  51 if (sscanf(CCS param, "pri=%u", &spamd->priority))
  52   return 0; /* OK */
  53
  54 if (sscanf(CCS param, "weight=%u", &spamd->weight))
  55   {
  56   if (spamd->weight == 0) /* this server disabled: skip it */
  57     return 1;
  58   return 0; /* OK */
  59   }
  60
  61 if (Ustrncmp(param, "time=", 5) == 0)
  62   {
  63   unsigned int start_h = 0, start_m = 0, start_s = 0;
  64   unsigned int end_h = 24, end_m = 0, end_s = 0;
  65   unsigned int time_start, time_end;
  66   const uschar * end_string;
  67
  68   name = US"time";
  69   s = param+5;
  70   if ((end_string = Ustrchr(s, '-')))
  71     {
  72     end_string++;
  73     if (  sscanf(CS end_string, "%u.%u.%u", &end_h,   &end_m,   &end_s)   == 0
  74        || sscanf(CS s,          "%u.%u.%u", &start_h, &start_m, &start_s) == 0
  75        )
  76       goto badval;
  77     }
  78   else
  79     goto badval;
  80
  81   if (timesinceday < 0)
  82     {
  83     time_t now = time(NULL);
  84     struct tm *tmp = localtime(&now);
  85     timesinceday = tmp->tm_hour*3600 + tmp->tm_min*60 + tmp->tm_sec;
  86     }
  87
  88   time_start = start_h*3600 + start_m*60 + start_s;
  89   time_end = end_h*3600 + end_m*60 + end_s;
  90
  91   if (timesinceday < time_start || timesinceday >= time_end)
  92     return 1; /* skip spamd server */
  93
  94   return 0; /* OK */
  95   }
  96
  97 if (Ustrcmp(param, "variant=rspamd") == 0)
  98   {
  99   spamd->is_rspamd = TRUE;
 100   return 0;
 101   }
 102
 103 if (Ustrncmp(param, "tmo=", 4) == 0)
 104   {
 105   int sec = readconf_readtime((s = param+4), '\0', FALSE);
 106   name = US"timeout";
 107   if (sec < 0)
 108     goto badval;
 109   spamd->timeout = sec;
 110   return 0;
 111   }
 112
 113 if (Ustrncmp(param, "retry=", 6) == 0)
 114   {
 115   int sec = readconf_readtime((s = param+6), '\0', FALSE);
 116   name = US"retry";
 117   if (sec < 0)
 118     goto badval;
 119   spamd->retry = sec;
 120   return 0;
 121   }
 122
 123 log_write(0, LOG_MAIN, "%s warning - invalid spamd parameter: '%s'",
 124   loglabel, param);
 125 return -1; /* syntax error */
 126
 127 badval:
 128   log_write(0, LOG_MAIN,
 129     "%s warning - invalid spamd %s value: '%s'", loglabel, name, s);
 130   return -1; /* syntax error */
 131 }
 132
 133
 134 static int
 135 spamd_get_server(spamd_address_container ** spamds, int num_servers)
 136 {
 137 unsigned int i;
 138 spamd_address_container * sd;
 139 long rnd, weights;
 140 unsigned pri;
 141 static BOOL srandomed = FALSE;
 142
 143 /* speedup, if we have only 1 server */
 144 if (num_servers == 1)
 145   return (spamds[0]->is_failed ? -1 : 0);
 146
 147 /* init ranmod */
 148 if (!srandomed)
 149   {
 150   struct timeval tv;
 151   gettimeofday(&tv, NULL);
 152   srandom((unsigned int)(tv.tv_usec/1000));
 153   srandomed = TRUE;
 154   }
 155
 156 /* scan for highest pri */
 157 for (pri = 0, i = 0; i < num_servers; i++)
 158   {
 159   sd = spamds[i];
 160   if (!sd->is_failed && sd->priority > pri) pri = sd->priority;
 161   }
 162
 163 /* get sum of weights */
 164 for (weights = 0, i = 0; i < num_servers; i++)
 165   {
 166   sd = spamds[i];
 167   if (!sd->is_failed && sd->priority == pri) weights += sd->weight;
 168   }
 169 if (weights == 0)       /* all servers failed */
 170   return -1;
 171
 172 for (rnd = random() % weights, i = 0; i < num_servers; i++)
 173   {
 174   sd = spamds[i];
 175   if (!sd->is_failed && sd->priority == pri)
 176     if ((rnd -= sd->weight) <= 0)
 177       return i;
 178   }
 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 client_conn_ctx spamd_cctx = {.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 #ifndef NO_POLL_H
 208 struct pollfd pollfd;
 209 #else                               /* Patch posted by Erik ? for OS X */
 210 struct timeval select_tv;         /* and applied by PH */
 211 fd_set select_fd;
 212 #endif
 213 uschar *spamd_address_work;
 214 spamd_address_container * sd;
 215
 216 /* stop compiler warning */
 217 result = 0;
 218
 219 /* find the username from the option list */
 220 if ((user_name = string_nextinlist(&list, &sep,
 221                                    user_name_buffer,
 222                                    sizeof(user_name_buffer))) == NULL)
 223   {
 224   /* no username given, this means no scanning should be done */
 225   return FAIL;
 226   }
 227
 228 /* if username is "0" or "false", do not scan */
 229 if ( (Ustrcmp(user_name,"0") == 0) ||
 230      (strcmpic(user_name,US"false") == 0) )
 231   return FAIL;
 232
 233 /* if there is an additional option, check if it is "true" */
 234 if (strcmpic(list,US"true") == 0)
 235   /* in that case, always return true later */
 236   override = 1;
 237
 238 /* expand spamd_address if needed */
 239 if (*spamd_address == '$')
 240   {
 241   spamd_address_work = expand_string(spamd_address);
 242   if (spamd_address_work == NULL)
 243     {
 244     log_write(0, LOG_MAIN|LOG_PANIC,
 245       "%s spamd_address starts with $, but expansion failed: %s",
 246       loglabel, expand_string_message);
 247     return DEFER;
 248     }
 249   }
 250 else
 251   spamd_address_work = spamd_address;
 252
 253 DEBUG(D_acl) debug_printf_indent("spamd: addrlist '%s'\n", spamd_address_work);
 254
 255 /* check if previous spamd_address was expanded and has changed. dump cached results if so */
 256 if (  spam_ok
 257    && prev_spamd_address_work != NULL
 258    && Ustrcmp(prev_spamd_address_work, spamd_address_work) != 0
 259    )
 260   spam_ok = 0;
 261
 262 /* if we scanned for this username last time, just return */
 263 if (spam_ok && Ustrcmp(prev_user_name, user_name) == 0)
 264   return override ? OK : spam_rc;
 265
 266 /* make sure the eml mbox file is spooled up */
 267
 268 if (!(mbox_file = spool_mbox(&mbox_size, NULL, NULL)))
 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, NULL, 0)))
 288     {
 289     const uschar * sublist;
 290     int sublist_sep = -(int)' ';        /* default space-sep */
 291     unsigned args;
 292     uschar * s;
 293
 294     DEBUG(D_acl) debug_printf_indent("spamd: addr entry '%s'\n", address);
 295     sd = (spamd_address_container *)store_get(sizeof(spamd_address_container));
 296
 297     for (sublist = address, args = 0, spamd_param_init(sd);
 298          (s = string_nextinlist(&sublist, &sublist_sep, NULL, 0));
 299          args++
 300          )
 301       {
 302         DEBUG(D_acl) debug_printf_indent("spamd:  addr parm '%s'\n", s);
 303         switch (args)
 304         {
 305         case 0:   sd->hostspec = s;
 306                   if (*s == '/') args++;        /* local; no port */
 307                   break;
 308         case 1:   sd->hostspec = string_sprintf("%s %s", sd->hostspec, s);
 309                   break;
 310         default:  spamd_param(s, sd);
 311                   break;
 312         }
 313       }
 314     if (args < 2)
 315       {
 316       log_write(0, LOG_MAIN,
 317         "%s warning - invalid spamd address: '%s'", loglabel, address);
 318       continue;
 319       }
 320
 321     spamd_address_vector[num_servers] = sd;
 322     if (++num_servers > 31)
 323       break;
 324     }
 325
 326   /* check if we have at least one server */
 327   if (!num_servers)
 328     {
 329     log_write(0, LOG_MAIN|LOG_PANIC,
 330        "%s no useable spamd server addresses in spamd_address configuration option.",
 331        loglabel);
 332     goto defer;
 333     }
 334
 335   current_server = spamd_get_server(spamd_address_vector, num_servers);
 336   sd = spamd_address_vector[current_server];
 337   for(;;)
 338     {
 339     uschar * errstr;
 340
 341     DEBUG(D_acl) debug_printf_indent("spamd: trying server %s\n", sd->hostspec);
 342
 343     for (;;)
 344       {
 345       /*XXX could potentially use TFO early-data here */
 346       if (  (spamd_cctx.sock = ip_streamsocket(sd->hostspec, &errstr, 5)) >= 0
 347          || sd->retry <= 0
 348          )
 349         break;
 350       DEBUG(D_acl) debug_printf_indent("spamd: server %s: retry conn\n", sd->hostspec);
 351       while (sd->retry > 0) sd->retry = sleep(sd->retry);
 352       }
 353     if (spamd_cctx.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 (void)fcntl(spamd_cctx.sock, F_SETFL, O_NONBLOCK);
 370 /* now we are connected to spamd on spamd_cctx.sock */
 371 if (sd->is_rspamd)
 372   {                             /* rspamd variant */
 373   uschar *req_str;
 374   const char *helo;
 375   const char *fcrdns;
 376
 377   req_str = string_sprintf("CHECK RSPAMC/1.3\r\nContent-length: %lu\r\n"
 378     "Queue-Id: %s\r\nFrom: <%s>\r\nRecipient-Number: %d\r\n",
 379     mbox_size, message_id, sender_address, recipients_count);
 380   for (i = 0; i < recipients_count; i ++)
 381     req_str = string_sprintf("%sRcpt: <%s>\r\n", req_str, recipients_list[i].address);
 382   if ((helo = expand_string(US"$sender_helo_name")) != NULL && *helo != '\0')
 383     req_str = string_sprintf("%sHelo: %s\r\n", req_str, helo);
 384   if ((fcrdns = expand_string(US"$sender_host_name")) != NULL && *fcrdns != '\0')
 385     req_str = string_sprintf("%sHostname: %s\r\n", req_str, fcrdns);
 386   if (sender_host_address != NULL)
 387     req_str = string_sprintf("%sIP: %s\r\n", req_str, sender_host_address);
 388   req_str = string_sprintf("%s\r\n", req_str);
 389   wrote = send(spamd_cctx.sock, req_str->s, req_str->ptr, 0);
 390   }
 391   else
 392   {                             /* spamassassin variant */
 393   (void)string_format(spamd_buffer,
 394           sizeof(spamd_buffer),
 395           "REPORT SPAMC/1.2\r\nUser: %s\r\nContent-length: %ld\r\n\r\n",
 396           user_name,
 397           mbox_size);
 398   /* send our request */
 399   wrote = send(spamd_cctx.sock, spamd_buffer, Ustrlen(spamd_buffer), 0);
 400   }
 401 if (wrote == -1)
 402   {
 403   (void)close(spamd_cctx.sock);
 404   log_write(0, LOG_MAIN|LOG_PANIC,
 405        "%s spamd %s send failed: %s", loglabel, callout_address, strerror(errno));
 406   goto defer;
 407   }
 408
 409 /* now send the file */
 410 /* spamd sometimes accepts connections but doesn't read data off
 411  * the connection.  We make the file descriptor non-blocking so
 412  * that the write will only write sufficient data without blocking
 413  * and we poll the descriptor to make sure that we can write without
 414  * blocking.  Short writes are gracefully handled and if the whole
 415  * transaction takes too long it is aborted.
 416  * Note: poll() is not supported in OSX 10.2 and is reported to be
 417  *       broken in more recent versions (up to 10.4).
 418  */
 419 #ifndef NO_POLL_H
 420 pollfd.fd = spamd_cctx.sock;
 421 pollfd.events = POLLOUT;
 422 #endif
 423 (void)fcntl(spamd_cctx.sock, F_SETFL, O_NONBLOCK);
 424 do
 425   {
 426   read = fread(spamd_buffer,1,sizeof(spamd_buffer),mbox_file);
 427   if (read > 0)
 428     {
 429     offset = 0;
 430 again:
 431 #ifndef NO_POLL_H
 432     result = poll(&pollfd, 1, 1000);
 433
 434 /* Patch posted by Erik ? for OS X and applied by PH */
 435 #else
 436     select_tv.tv_sec = 1;
 437     select_tv.tv_usec = 0;
 438     FD_ZERO(&select_fd);
 439     FD_SET(spamd_cctx.sock, &select_fd);
 440     result = select(spamd_cctx.sock+1, NULL, &select_fd, NULL, &select_tv);
 441 #endif
 442 /* End Erik's patch */
 443
 444     if (result == -1 && errno == EINTR)
 445       goto again;
 446     else if (result < 1)
 447       {
 448       if (result == -1)
 449         log_write(0, LOG_MAIN|LOG_PANIC,
 450           "%s %s on spamd %s socket", loglabel, callout_address, strerror(errno));
 451       else
 452         {
 453         if (time(NULL) - start < sd->timeout)
 454           goto again;
 455         log_write(0, LOG_MAIN|LOG_PANIC,
 456           "%s timed out writing spamd %s, socket", loglabel, callout_address);
 457         }
 458       (void)close(spamd_cctx.sock);
 459       goto defer;
 460       }
 461
 462     wrote = send(spamd_cctx.sock,spamd_buffer + offset,read - offset,0);
 463     if (wrote == -1)
 464       {
 465       log_write(0, LOG_MAIN|LOG_PANIC,
 466           "%s %s on spamd %s socket", loglabel, callout_address, strerror(errno));
 467       (void)close(spamd_cctx.sock);
 468       goto defer;
 469       }
 470     if (offset + wrote != read)
 471       {
 472       offset += wrote;
 473       goto again;
 474       }
 475     }
 476   }
 477 while (!feof(mbox_file) && !ferror(mbox_file));
 478
 479 if (ferror(mbox_file))
 480   {
 481   log_write(0, LOG_MAIN|LOG_PANIC,
 482     "%s error reading spool file: %s", loglabel, strerror(errno));
 483   (void)close(spamd_cctx.sock);
 484   goto defer;
 485   }
 486
 487 (void)fclose(mbox_file);
 488
 489 /* we're done sending, close socket for writing */
 490 if (!sd->is_rspamd)
 491   shutdown(spamd_cctx.sock, SHUT_WR);
 492
 493 /* read spamd response using what's left of the timeout.  */
 494 memset(spamd_buffer, 0, sizeof(spamd_buffer));
 495 offset = 0;
 496 while ((i = ip_recv(&spamd_cctx,
 497                    spamd_buffer + offset,
 498                    sizeof(spamd_buffer) - offset - 1,
 499                    sd->timeout - time(NULL) + start)) > 0)
 500   offset += i;
 501 spamd_buffer[offset] = '\0';    /* guard byte */
 502
 503 /* error handling */
 504 if (i <= 0 && errno != 0)
 505   {
 506   log_write(0, LOG_MAIN|LOG_PANIC,
 507        "%s error reading from spamd %s, socket: %s", loglabel, callout_address, strerror(errno));
 508   (void)close(spamd_cctx.sock);
 509   return DEFER;
 510   }
 511
 512 /* reading done */
 513 (void)close(spamd_cctx.sock);
 514
 515 if (sd->is_rspamd)
 516   {                             /* rspamd variant of reply */
 517   int r;
 518   if (  (r = sscanf(CS spamd_buffer,
 519           "RSPAMD/%7s 0 EX_OK\r\nMetric: default; %7s %lf / %lf / %lf\r\n%n",
 520           spamd_version, spamd_short_result, &spamd_score, &spamd_threshold,
 521           &spamd_reject_score, &spamd_report_offset)) != 5
 522      || spamd_report_offset >= offset           /* verify within buffer */
 523      )
 524     {
 525     log_write(0, LOG_MAIN|LOG_PANIC,
 526               "%s cannot parse spamd %s, output: %d", loglabel, callout_address, r);
 527     return DEFER;
 528     }
 529   /* now parse action */
 530   p = &spamd_buffer[spamd_report_offset];
 531
 532   if (Ustrncmp(p, "Action: ", sizeof("Action: ") - 1) == 0)
 533     {
 534     p += sizeof("Action: ") - 1;
 535     q = &spam_action_buffer[0];
 536     while (*p && *p != '\r' && (q - spam_action_buffer) < sizeof(spam_action_buffer) - 1)
 537       *q++ = *p++;
 538     *q = '\0';
 539     }
 540   }
 541 else
 542   {                             /* spamassassin */
 543   /* dig in the spamd output and put the report in a multiline header,
 544   if requested */
 545   if (sscanf(CS spamd_buffer,
 546        "SPAMD/%7s 0 EX_OK\r\nContent-length: %*u\r\n\r\n%lf/%lf\r\n%n",
 547        spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 548     {
 549       /* try to fall back to pre-2.50 spamd output */
 550       if (sscanf(CS spamd_buffer,
 551            "SPAMD/%7s 0 EX_OK\r\nSpam: %*s ; %lf / %lf\r\n\r\n%n",
 552            spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3)
 553         {
 554         log_write(0, LOG_MAIN|LOG_PANIC,
 555                   "%s cannot parse spamd %s output", loglabel, callout_address);
 556         return DEFER;
 557         }
 558     }
 559
 560   Ustrcpy(spam_action_buffer,
 561     spamd_score >= spamd_threshold ? "reject" : "no action");
 562   }
 563
 564 /* Create report. Since this is a multiline string,
 565 we must hack it into shape first */
 566 p = &spamd_buffer[spamd_report_offset];
 567 q = spam_report_buffer;
 568 while (*p != '\0')
 569   {
 570   /* skip \r */
 571   if (*p == '\r')
 572     {
 573     p++;
 574     continue;
 575     }
 576   *q++ = *p;
 577   if (*p++ == '\n')
 578     {
 579     /* add an extra space after the newline to ensure
 580     that it is treated as a header continuation line */
 581     *q++ = ' ';
 582     }
 583   }
 584 /* NULL-terminate */
 585 *q-- = '\0';
 586 /* cut off trailing leftovers */
 587 while (*q <= ' ')
 588   *q-- = '\0';
 589
 590 spam_report = spam_report_buffer;
 591 spam_action = spam_action_buffer;
 592
 593 /* create spam bar */
 594 spamd_score_char = spamd_score > 0 ? '+' : '-';
 595 j = abs((int)(spamd_score));
 596 i = 0;
 597 if (j != 0)
 598   while ((i < j) && (i <= MAX_SPAM_BAR_CHARS))
 599      spam_bar_buffer[i++] = spamd_score_char;
 600 else
 601   {
 602   spam_bar_buffer[0] = '/';
 603   i = 1;
 604   }
 605 spam_bar_buffer[i] = '\0';
 606 spam_bar = spam_bar_buffer;
 607
 608 /* create "float" spam score */
 609 (void)string_format(spam_score_buffer, sizeof(spam_score_buffer),
 610         "%.1f", spamd_score);
 611 spam_score = spam_score_buffer;
 612
 613 /* create "int" spam score */
 614 j = (int)((spamd_score + 0.001)*10);
 615 (void)string_format(spam_score_int_buffer, sizeof(spam_score_int_buffer),
 616         "%d", j);
 617 spam_score_int = spam_score_int_buffer;
 618
 619 /* compare threshold against score */
 620 spam_rc = spamd_score >= spamd_threshold
 621   ? OK  /* spam as determined by user's threshold */
 622   : FAIL;       /* not spam */
 623
 624 /* remember expanded spamd_address if needed */
 625 if (spamd_address_work != spamd_address)
 626   prev_spamd_address_work = string_copy(spamd_address_work);
 627
 628 /* remember user name and "been here" for it */
 629 Ustrcpy(prev_user_name, user_name);
 630 spam_ok = 1;
 631
 632 return override
 633   ? OK          /* always return OK, no matter what the score */
 634   : spam_rc;
 635
 636 defer:
 637   (void)fclose(mbox_file);
 638   return DEFER;
 639 }
 640
 641 #endif
 642 /* vi: aw ai sw=2
 643 */