src/src/spam.c

   1 /* $Cambridge: exim/src/src/spam.c,v 1.18 2010/06/05 11:13:30 pdp Exp $ */
   2
   3 /*************************************************
   4 *     Exim - an Internet mail transport agent    *
   5 *************************************************/
   6
   7 /* Copyright (c) Tom Kistner <tom@duncanthrax.net> 2003-???? */
   8 /* License: GPL */
   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_report_buffer[32600];
  20 uschar prev_user_name[128] = "";
  21 int spam_ok = 0;
  22 int spam_rc = 0;
  23 uschar *prev_spamd_address_work = NULL;
  24
  25 int spam(uschar **listptr) {
  26   int sep = 0;
  27   uschar *list = *listptr;
  28   uschar *user_name;
  29   uschar user_name_buffer[128];
  30   unsigned long mbox_size;
  31   FILE *mbox_file;
  32   int spamd_sock;
  33   uschar spamd_buffer[32600];
  34   int i, j, offset, result;
  35   uschar spamd_version[8];
  36   uschar spamd_score_char;
  37   double spamd_threshold, spamd_score;
  38   int spamd_report_offset;
  39   uschar *p,*q;
  40   int override = 0;
  41   time_t start;
  42   size_t read, wrote;
  43   struct sockaddr_un server;
  44 #ifndef NO_POLL_H
  45   struct pollfd pollfd;
  46 #else                               /* Patch posted by Erik ? for OS X */
  47   struct timeval select_tv;         /* and applied by PH */
  48   fd_set select_fd;
  49 #endif
  50   uschar *spamd_address_work;
  51
  52   /* stop compiler warning */
  53   result = 0;
  54
  55   /* find the username from the option list */
  56   if ((user_name = string_nextinlist(&list, &sep,
  57                                      user_name_buffer,
  58                                      sizeof(user_name_buffer))) == NULL) {
  59     /* no username given, this means no scanning should be done */
  60     return FAIL;
  61   };
  62
  63   /* if username is "0" or "false", do not scan */
  64   if ( (Ustrcmp(user_name,"0") == 0) ||
  65        (strcmpic(user_name,US"false") == 0) ) {
  66     return FAIL;
  67   };
  68
  69   /* if there is an additional option, check if it is "true" */
  70   if (strcmpic(list,US"true") == 0) {
  71     /* in that case, always return true later */
  72     override = 1;
  73   };
  74
  75   /* expand spamd_address if needed */
  76   if (*spamd_address == '$') {
  77     spamd_address_work = expand_string(spamd_address);
  78     if (spamd_address_work == NULL) {
  79       log_write(0, LOG_MAIN|LOG_PANIC,
  80         "spamassassin acl condition: spamd_address starts with $, but expansion failed: %s", expand_string_message);
  81       return DEFER;
  82     }
  83   }
  84   else
  85     spamd_address_work = spamd_address;
  86
  87   /* check if previous spamd_address was expanded and has changed. dump cached results if so */
  88   if ( spam_ok && ( prev_spamd_address_work != NULL) && (Ustrcmp(prev_spamd_address_work, spamd_address_work) != 0)) {
  89     spam_ok = 0;
  90   }
  91
  92   /* if we scanned for this username last time, just return */
  93   if ( spam_ok && ( Ustrcmp(prev_user_name, user_name) == 0 ) ) {
  94     if (override)
  95       return OK;
  96     else
  97       return spam_rc;
  98   };
  99
 100   /* make sure the eml mbox file is spooled up */
 101   mbox_file = spool_mbox(&mbox_size, NULL);
 102
 103   if (mbox_file == NULL) {
 104     /* error while spooling */
 105     log_write(0, LOG_MAIN|LOG_PANIC,
 106            "spam acl condition: error while creating mbox spool file");
 107     return DEFER;
 108   };
 109
 110   start = time(NULL);
 111
 112   /* socket does not start with '/' -> network socket */
 113   if (*spamd_address_work != '/') {
 114     int num_servers = 0;
 115     int current_server;
 116     uschar *address = NULL;
 117     uschar *spamd_address_list_ptr = spamd_address_work;
 118     uschar address_buffer[256];
 119     spamd_address_container * spamd_address_vector[32];
 120
 121     /* Check how many spamd servers we have
 122        and register their addresses */
 123     while ((address = string_nextinlist(&spamd_address_list_ptr, &sep,
 124                                         address_buffer,
 125                                         sizeof(address_buffer))) != NULL) {
 126
 127       /* Potential memory leak as we never free the store. */
 128       spamd_address_container *this_spamd =
 129         (spamd_address_container *)store_get(sizeof(spamd_address_container));
 130
 131       /* grok spamd address and port */
 132       if( sscanf(CS address, "%s %u", this_spamd->tcp_addr, &(this_spamd->tcp_port)) != 2 ) {
 133         log_write(0, LOG_MAIN,
 134           "spam acl condition: warning - invalid spamd address: '%s'", address);
 135         continue;
 136       };
 137
 138       spamd_address_vector[num_servers] = this_spamd;
 139       num_servers++;
 140       if (num_servers > 31)
 141         break;
 142     };
 143
 144     /* check if we have at least one server */
 145     if (!num_servers) {
 146       log_write(0, LOG_MAIN|LOG_PANIC,
 147          "spam acl condition: no useable spamd server addresses in spamd_address configuration option.");
 148       (void)fclose(mbox_file);
 149       return DEFER;
 150     };
 151
 152     while ( num_servers > 0 ) {
 153       int i;
 154
 155       /* Randomly pick a server to try */
 156       current_server = random_number( num_servers );
 157
 158       debug_printf("trying server %s, port %u\n",
 159                    spamd_address_vector[current_server]->tcp_addr,
 160                    spamd_address_vector[current_server]->tcp_port);
 161
 162       /* contact a spamd */
 163       if ( (spamd_sock = ip_socket(SOCK_STREAM, AF_INET)) < 0) {
 164         log_write(0, LOG_MAIN|LOG_PANIC,
 165            "spam acl condition: error creating IP socket for spamd");
 166         (void)fclose(mbox_file);
 167         return DEFER;
 168       };
 169
 170       if (ip_connect( spamd_sock,
 171                       AF_INET,
 172                       spamd_address_vector[current_server]->tcp_addr,
 173                       spamd_address_vector[current_server]->tcp_port,
 174                       5 ) > -1) {
 175         /* connection OK */
 176         break;
 177       };
 178
 179       log_write(0, LOG_MAIN|LOG_PANIC,
 180          "spam acl condition: warning - spamd connection to %s, port %u failed: %s",
 181          spamd_address_vector[current_server]->tcp_addr,
 182          spamd_address_vector[current_server]->tcp_port,
 183          strerror(errno));
 184
 185       (void)close(spamd_sock);
 186
 187       /* Remove the server from the list. XXX We should free the memory */
 188       num_servers--;
 189       for( i = current_server; i < num_servers; i++ )
 190         spamd_address_vector[i] = spamd_address_vector[i+1];
 191     }
 192
 193     if ( num_servers == 0 ) {
 194       log_write(0, LOG_MAIN|LOG_PANIC, "spam acl condition: all spamd servers failed");
 195       (void)fclose(mbox_file);
 196       return DEFER;
 197     }
 198
 199   }
 200   else {
 201     /* open the local socket */
 202
 203     if ((spamd_sock = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) {
 204       log_write(0, LOG_MAIN|LOG_PANIC,
 205                 "malware acl condition: spamd: unable to acquire socket (%s)",
 206                 strerror(errno));
 207       (void)fclose(mbox_file);
 208       return DEFER;
 209     }
 210
 211     server.sun_family = AF_UNIX;
 212     Ustrcpy(server.sun_path, spamd_address_work);
 213
 214     if (connect(spamd_sock, (struct sockaddr *) &server, sizeof(struct sockaddr_un)) < 0) {
 215       log_write(0, LOG_MAIN|LOG_PANIC,
 216                 "malware acl condition: spamd: unable to connect to UNIX socket %s (%s)",
 217                 spamd_address_work, strerror(errno) );
 218       (void)fclose(mbox_file);
 219       (void)close(spamd_sock);
 220       return DEFER;
 221     }
 222
 223   }
 224
 225   /* now we are connected to spamd on spamd_sock */
 226   (void)string_format(spamd_buffer,
 227            sizeof(spamd_buffer),
 228            "REPORT SPAMC/1.2\r\nUser: %s\r\nContent-length: %ld\r\n\r\n",
 229            user_name,
 230            mbox_size);
 231
 232   /* send our request */
 233   if (send(spamd_sock, spamd_buffer, Ustrlen(spamd_buffer), 0) < 0) {
 234     (void)close(spamd_sock);
 235     log_write(0, LOG_MAIN|LOG_PANIC,
 236          "spam acl condition: spamd send failed: %s", strerror(errno));
 237     (void)fclose(mbox_file);
 238     (void)close(spamd_sock);
 239     return DEFER;
 240   };
 241
 242   /* now send the file */
 243   /* spamd sometimes accepts conections but doesn't read data off
 244    * the connection.  We make the file descriptor non-blocking so
 245    * that the write will only write sufficient data without blocking
 246    * and we poll the desciptor to make sure that we can write without
 247    * blocking.  Short writes are gracefully handled and if the whole
 248    * trasaction takes too long it is aborted.
 249    * Note: poll() is not supported in OSX 10.2 and is reported to be
 250    *       broken in more recent versions (up to 10.4).
 251    */
 252 #ifndef NO_POLL_H
 253   pollfd.fd = spamd_sock;
 254   pollfd.events = POLLOUT;
 255 #endif
 256   (void)fcntl(spamd_sock, F_SETFL, O_NONBLOCK);
 257   do {
 258     read = fread(spamd_buffer,1,sizeof(spamd_buffer),mbox_file);
 259     if (read > 0) {
 260       offset = 0;
 261 again:
 262 #ifndef NO_POLL_H
 263       result = poll(&pollfd, 1, 1000);
 264
 265 /* Patch posted by Erik ? for OS X and applied by PH */
 266 #else
 267       select_tv.tv_sec = 1;
 268       select_tv.tv_usec = 0;
 269       FD_ZERO(&select_fd);
 270       FD_SET(spamd_sock, &select_fd);
 271       result = select(spamd_sock+1, NULL, &select_fd, NULL, &select_tv);
 272 #endif
 273 /* End Erik's patch */
 274
 275       if (result == -1 && errno == EINTR)
 276         goto again;
 277       else if (result < 1) {
 278         if (result == -1)
 279           log_write(0, LOG_MAIN|LOG_PANIC,
 280             "spam acl condition: %s on spamd socket", strerror(errno));
 281         else {
 282           if (time(NULL) - start < SPAMD_TIMEOUT)
 283           goto again;
 284           log_write(0, LOG_MAIN|LOG_PANIC,
 285             "spam acl condition: timed out writing spamd socket");
 286         }
 287         (void)close(spamd_sock);
 288         (void)fclose(mbox_file);
 289         return DEFER;
 290       }
 291
 292       wrote = send(spamd_sock,spamd_buffer + offset,read - offset,0);
 293       if (wrote == -1)
 294       {
 295           log_write(0, LOG_MAIN|LOG_PANIC,
 296             "spam acl condition: %s on spamd socket", strerror(errno));
 297         (void)close(spamd_sock);
 298         (void)fclose(mbox_file);
 299         return DEFER;
 300       }
 301       if (offset + wrote != read) {
 302         offset += wrote;
 303         goto again;
 304       }
 305     }
 306   }
 307   while (!feof(mbox_file) && !ferror(mbox_file));
 308   if (ferror(mbox_file)) {
 309     log_write(0, LOG_MAIN|LOG_PANIC,
 310       "spam acl condition: error reading spool file: %s", strerror(errno));
 311     (void)close(spamd_sock);
 312     (void)fclose(mbox_file);
 313     return DEFER;
 314   }
 315
 316   (void)fclose(mbox_file);
 317
 318   /* we're done sending, close socket for writing */
 319   shutdown(spamd_sock,SHUT_WR);
 320
 321   /* read spamd response using what's left of the timeout.
 322    */
 323   memset(spamd_buffer, 0, sizeof(spamd_buffer));
 324   offset = 0;
 325   while((i = ip_recv(spamd_sock,
 326                      spamd_buffer + offset,
 327                      sizeof(spamd_buffer) - offset - 1,
 328                      SPAMD_TIMEOUT - time(NULL) + start)) > 0 ) {
 329     offset += i;
 330   }
 331
 332   /* error handling */
 333   if((i <= 0) && (errno != 0)) {
 334     log_write(0, LOG_MAIN|LOG_PANIC,
 335          "spam acl condition: error reading from spamd socket: %s", strerror(errno));
 336     (void)close(spamd_sock);
 337     return DEFER;
 338   }
 339
 340   /* reading done */
 341   (void)close(spamd_sock);
 342
 343   /* dig in the spamd output and put the report in a multiline header, if requested */
 344   if( sscanf(CS spamd_buffer,"SPAMD/%7s 0 EX_OK\r\nContent-length: %*u\r\n\r\n%lf/%lf\r\n%n",
 345              spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3 ) {
 346
 347     /* try to fall back to pre-2.50 spamd output */
 348     if( sscanf(CS spamd_buffer,"SPAMD/%7s 0 EX_OK\r\nSpam: %*s ; %lf / %lf\r\n\r\n%n",
 349                spamd_version,&spamd_score,&spamd_threshold,&spamd_report_offset) != 3 ) {
 350       log_write(0, LOG_MAIN|LOG_PANIC,
 351          "spam acl condition: cannot parse spamd output");
 352       return DEFER;
 353     };
 354   };
 355
 356   /* Create report. Since this is a multiline string,
 357   we must hack it into shape first */
 358   p = &spamd_buffer[spamd_report_offset];
 359   q = spam_report_buffer;
 360   while (*p != '\0') {
 361     /* skip \r */
 362     if (*p == '\r') {
 363       p++;
 364       continue;
 365     };
 366     *q = *p;
 367     q++;
 368     if (*p == '\n') {
 369       /* add an extra space after the newline to ensure
 370       that it is treated as a header continuation line */
 371       *q = ' ';
 372       q++;
 373     };
 374     p++;
 375   };
 376   /* NULL-terminate */
 377   *q = '\0';
 378   q--;
 379   /* cut off trailing leftovers */
 380   while (*q <= ' ') {
 381     *q = '\0';
 382     q--;
 383   };
 384   spam_report = spam_report_buffer;
 385
 386   /* create spam bar */
 387   spamd_score_char = spamd_score > 0 ? '+' : '-';
 388   j = abs((int)(spamd_score));
 389   i = 0;
 390   if( j != 0 ) {
 391     while((i < j) && (i <= MAX_SPAM_BAR_CHARS))
 392        spam_bar_buffer[i++] = spamd_score_char;
 393   }
 394   else{
 395     spam_bar_buffer[0] = '/';
 396     i = 1;
 397   }
 398   spam_bar_buffer[i] = '\0';
 399   spam_bar = spam_bar_buffer;
 400
 401   /* create "float" spam score */
 402   (void)string_format(spam_score_buffer, sizeof(spam_score_buffer),"%.1f", spamd_score);
 403   spam_score = spam_score_buffer;
 404
 405   /* create "int" spam score */
 406   j = (int)((spamd_score + 0.001)*10);
 407   (void)string_format(spam_score_int_buffer, sizeof(spam_score_int_buffer), "%d", j);
 408   spam_score_int = spam_score_int_buffer;
 409
 410   /* compare threshold against score */
 411   if (spamd_score >= spamd_threshold) {
 412     /* spam as determined by user's threshold */
 413     spam_rc = OK;
 414   }
 415   else {
 416     /* not spam */
 417     spam_rc = FAIL;
 418   };
 419
 420   /* remember expanded spamd_address if needed */
 421   if (spamd_address_work != spamd_address) {
 422     prev_spamd_address_work = string_copy(spamd_address_work);
 423   }
 424   /* remember user name and "been here" for it */
 425   Ustrcpy(prev_user_name, user_name);
 426   spam_ok = 1;
 427
 428   if (override) {
 429     /* always return OK, no matter what the score */
 430     return OK;
 431   }
 432   else {
 433     return spam_rc;
 434   };
 435 }
 436
 437 #endif