Mercurial > dnsbl
annotate src/spamass.cpp.in @ 470:cef0da2a3b32
Added tag stable-6-0-77 for changeset 0370916c7626
| author | Carl Byington <carl@five-ten-sg.com> |
|---|---|
| date | Wed, 16 Dec 2020 11:25:19 -0800 |
| parents | f5b394bec28c |
| children | 5209e92b4885 |
| rev | line source |
|---|---|
| 163 | 1 /* |
| 2 | |
| 3 Copyright (c) 2007 Carl Byington - 510 Software Group, released under | |
| 4 the GPL version 3 or any later version at your choice available at | |
| 5 http://www.gnu.org/licenses/gpl-3.0.txt | |
| 6 | |
| 7 Based on spamass-milter by Georg C. F. Greve <greve@gnu.org> | |
| 8 | |
| 9 */ | |
| 10 | |
| 177 | 11 #include "includes.h" |
| 163 | 12 #include <errno.h> |
| 13 #include <fcntl.h> | |
| 14 #include <poll.h> | |
| 15 #include <signal.h> | |
| 16 #include <string> | |
| 17 #include <sys/types.h> | |
| 18 #include <sys/wait.h> | |
| 19 #include <unistd.h> | |
| 20 | |
| 21 | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
22 const char *spamc = "@SPAMC@"; |
|
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
23 const char *spamc_empty = ""; |
| 177 | 24 static bool warnedmacro = false; // have we logged that we couldn't fetch a macro? |
| 25 const int maxlen = 1000; // used for snprintf buffers | |
| 163 | 26 |
| 27 | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
28 SpamAssassin::SpamAssassin(mlfiPriv *priv_, int ip, const char *helo_, const char *from, const char *qid) |
| 163 | 29 { |
| 30 error = false; | |
| 31 running = false; | |
| 32 first_recipient = true; | |
|
167
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
33 priv = priv_; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
34 ip4 = ip; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
35 helo = helo_; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
36 envfrom = from; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
37 queueid = qid; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
38 pid = 0; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
39 pipe_io[0][0] = -1; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
40 pipe_io[0][1] = -1; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
41 pipe_io[1][0] = -1; |
|
9b129ed78d7d
actually use spamassassin result, allow build without spam assassin, only call it if some recipient needs it.
carl
parents:
164
diff
changeset
|
42 pipe_io[1][1] = -1; |
| 163 | 43 } |
| 44 | |
| 45 | |
| 46 SpamAssassin::~SpamAssassin() | |
| 47 { | |
| 48 // close all pipes that are still open | |
| 49 if (pipe_io[0][0] > -1) close(pipe_io[0][0]); | |
| 50 if (pipe_io[0][1] > -1) close(pipe_io[0][1]); | |
| 51 if (pipe_io[1][0] > -1) close(pipe_io[1][0]); | |
| 52 if (pipe_io[1][1] > -1) close(pipe_io[1][1]); | |
| 53 // child still running? | |
| 54 if (running) { | |
| 55 // make sure the pid is valid | |
| 56 if (pid > 0) { | |
| 57 // slaughter child | |
| 58 kill(pid, SIGKILL); | |
| 59 // wait for child to terminate | |
| 60 int status; | |
| 61 waitpid(pid, &status, 0); | |
| 62 } | |
| 63 } | |
| 64 } | |
| 65 | |
| 66 | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
67 void SpamAssassin::mlfi_envrcpt(SMFICTX *ctx, const char *envrcpt) |
| 163 | 68 { |
| 69 if (first_recipient) { | |
| 177 | 70 first_recipient = false; |
| 163 | 71 /* Send the envelope headers as X-Envelope-From: and |
| 72 X-Envelope-To: so that SpamAssassin can use them in its | |
| 73 whitelist checks. Also forge as complete a dummy | |
| 74 Received: header as possible because SA gets a lot of | |
| 75 info from it. | |
| 76 | |
| 77 HReceived: $?sfrom $s $.$?_($?s$|from $.$_) | |
| 78 $.$?{auth_type}(authenticated$?{auth_ssf} bits=${auth_ssf}$.) | |
| 79 $.by $j ($v/$Z)$?r with $r$. id $i$?{tls_version} | |
| 80 (version=${tls_version} cipher=${cipher} bits=${cipher_bits} verify=${verify})$.$?u | |
| 81 for $u; $|; | |
| 82 $.$b$?g | |
| 83 (envelope-from $g)$. | |
| 84 | |
| 85 */ | |
| 86 const char *macro_b, *macro_i, *macro_j, *macro_r, | |
| 87 *macro_s, *macro_v, *macro_Z, *macro__; | |
| 88 char date[32]; | |
| 89 time_t tval; | |
| 90 time(&tval); | |
| 91 strftime(date, sizeof(date), "%a, %d %b %Y %H:%M:%S %z", localtime(&tval)); | |
| 92 macro_b = date; | |
| 93 | |
| 168 | 94 // queue ID |
| 163 | 95 macro_i = queueid; |
| 96 | |
| 168 | 97 // FQDN of this site |
| 98 macro_j = getorwarnmacro(ctx, "j", "localhost", "ENVRCPT"); | |
| 163 | 99 |
| 170 | 100 // Protocol used to receive the message, not really needed by spam assassin |
| 101 macro_r = "SMTP"; | |
| 163 | 102 |
| 170 | 103 // helo value we already have |
| 163 | 104 macro_s = helo; |
| 105 | |
| 170 | 106 // Sendmail binary version, not really needed by spam assassin |
| 107 macro_v = "8.13.0"; | |
| 163 | 108 |
| 170 | 109 // Sendmail .cf version, not really needed by spam assassin |
| 110 macro_Z = "8.13.0"; | |
| 163 | 111 |
| 168 | 112 // Validated sending site's address |
| 170 | 113 macro__ = getorwarnmacro(ctx, "_", "unknown", "ENVRCPT"); |
| 163 | 114 |
| 115 output(string("Received: from ") + macro_s + " (" + macro__+ ")\r\n\t" + | |
| 116 "by " + macro_j + " (" + macro_v + "/" + macro_Z + ") with " + macro_r + " id " + macro_i + "\r\n\t" + | |
| 117 "for " + envfrom + ";\r\n\t" + | |
| 118 macro_b + "\r\n"); | |
| 119 | |
| 120 output(string("X-Envelope-From: ") + envfrom + "\r\n"); | |
| 121 } | |
| 122 output(string("X-Envelope-To: ") + envrcpt + "\r\n"); | |
| 123 } | |
| 124 | |
| 125 | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
126 void SpamAssassin::mlfi_header(const char* headerf, const char* headerv) |
| 163 | 127 { |
| 128 if (!running) Connect(); | |
| 177 | 129 if (running) { |
| 130 output(spamc_input); | |
|
227
3fee608becbc
Fixes to compile on old systems without memrchr or string::clear().
Carl Byington <carl@five-ten-sg.com>
parents:
214
diff
changeset
|
131 spamc_input = ""; |
| 177 | 132 } |
| 133 | |
| 163 | 134 output(headerf); |
| 135 output(": "); | |
| 136 output(headerv); | |
| 137 output("\r\n"); | |
| 138 } | |
| 139 | |
| 140 | |
| 141 void SpamAssassin::mlfi_eoh() | |
| 142 { | |
| 143 output("\r\n"); | |
| 144 } | |
| 145 | |
| 146 | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
147 void SpamAssassin::mlfi_body(const u_char *bodyp, size_t bodylen) |
| 163 | 148 { |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
149 output((const char *)bodyp, bodylen); |
| 163 | 150 } |
| 151 | |
| 152 | |
| 153 int SpamAssassin::mlfi_eom() | |
| 154 { | |
| 155 close_output(); // signal EOF to SpamAssassin | |
| 156 input(); // read what the Assassin is telling us | |
| 157 my_syslog(priv, "spamc returned " + spamc_output); | |
|
445
78eedbbce636
round spamassassin scores; check >= rather than >
Carl Byington <carl@five-ten-sg.com>
parents:
240
diff
changeset
|
158 return int(round(atof(spamc_output.c_str()))); |
| 163 | 159 } |
| 160 | |
| 161 | |
| 162 void SpamAssassin::Connect() | |
| 163 { | |
| 164 if (error) return; | |
| 165 // set up pipes for in- and output | |
| 166 error |= (pipe(pipe_io[0])); | |
| 167 error |= (pipe(pipe_io[1])); | |
| 168 if (error) return; | |
| 169 | |
| 170 // now execute SpamAssassin client for contact with SpamAssassin spamd | |
| 171 // start child process | |
| 172 pid = fork(); | |
| 173 switch (pid) { | |
| 174 case -1: | |
| 175 // forking trouble. | |
| 176 my_syslog(priv, "unable to fork for spamc"); | |
| 177 error = true; | |
| 178 close(pipe_io[0][0]); | |
| 179 close(pipe_io[0][1]); | |
| 180 close(pipe_io[1][0]); | |
| 181 close(pipe_io[1][1]); | |
| 182 pipe_io[0][0] = -1; | |
| 183 pipe_io[0][1] = -1; | |
| 184 pipe_io[1][0] = -1; | |
| 185 pipe_io[1][1] = -1; | |
| 186 return; | |
| 187 case 0: | |
| 188 // +++ CHILD +++ | |
| 189 | |
| 190 // close unused pipes | |
| 191 close(pipe_io[1][0]); | |
| 192 close(pipe_io[0][1]); | |
| 193 | |
| 194 // redirect stdin(0), stdout(1) and stderr(2) | |
| 195 dup2(pipe_io[0][0],0); | |
| 196 dup2(pipe_io[1][1],1); | |
| 197 dup2(pipe_io[1][1],2); | |
| 198 | |
| 199 closeall(3); | |
| 200 | |
| 201 // execute spamc | |
| 202 char* argv[3]; | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
203 argv[0] = (char*)spamc; |
|
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
204 argv[1] = (char*)"-c"; |
| 163 | 205 argv[2] = NULL; |
| 206 execvp(argv[0] , argv); // does not return! | |
| 207 _exit(1); // exec failed | |
| 208 break; | |
| 209 } | |
| 210 | |
| 211 // +++ PARENT +++ | |
| 212 | |
| 213 // close unused pipes | |
| 214 close(pipe_io[0][0]); | |
| 215 close(pipe_io[1][1]); | |
| 216 pipe_io[0][0] = -1; | |
| 217 pipe_io[1][1] = -1; | |
| 218 | |
| 219 // mark the pipes non-blocking | |
| 220 if (fcntl(pipe_io[0][1], F_SETFL, O_NONBLOCK) == -1) | |
| 221 error = true; | |
| 222 #if 0 /* don't really need to make the sink pipe nonblocking */ | |
| 223 if (fcntl(pipe_io[1][0], F_SETFL, O_NONBLOCK) == -1) | |
| 224 error = true; | |
| 225 #endif | |
| 226 | |
| 227 // we have to assume the client is running now. | |
| 228 running = true; | |
| 229 } | |
| 230 | |
| 231 | |
| 232 void SpamAssassin::output(const char* buffer, size_t size) | |
| 233 { | |
| 234 // if there are problems, fail. | |
| 235 if (error) return; | |
| 236 | |
| 237 if (!running) { | |
| 238 // buffer it | |
| 239 spamc_input.append(buffer, size); | |
| 240 return; | |
| 241 } | |
| 242 | |
| 243 // send to SpamAssassin | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
244 size_t total = 0; |
|
467
f5b394bec28c
allow checking names without A records on uribl lists; improve extraction of fake TLDs from our RPZ zone
Carl Byington <carl@five-ten-sg.com>
parents:
445
diff
changeset
|
245 ssize_t wsize = 0; |
| 163 | 246 string reason; |
| 247 int status; | |
| 248 do { | |
| 249 struct pollfd fds[2]; | |
| 250 int nfds = 2, nready; | |
| 251 fds[0].fd = pipe_io[0][1]; | |
| 252 fds[0].events = POLLOUT; | |
| 253 fds[0].revents = 0; | |
| 254 fds[1].fd = pipe_io[1][0]; | |
| 255 fds[1].events = POLLIN; | |
| 256 fds[1].revents = 0; | |
| 257 | |
| 258 nready = poll(fds, nfds, 1000); | |
| 259 if (nready == -1) { | |
| 260 my_syslog(priv, "poll failed"); | |
| 261 error = true; | |
| 262 return; | |
| 263 } | |
| 264 | |
|
240
536d59c3a7b8
pollnval might be triggered on the last pollin data chunk
Carl Byington <carl@five-ten-sg.com>
parents:
227
diff
changeset
|
265 if (fds[1].revents & (POLLERR|POLLNVAL)) { |
| 163 | 266 my_syslog(priv, "poll says my read pipe is busted"); |
| 267 error = true; | |
| 268 return; | |
| 269 } | |
| 270 | |
| 271 if (fds[0].revents & (POLLERR|POLLNVAL|POLLHUP)) { | |
| 272 my_syslog(priv, "poll says my write pipe is busted"); | |
| 273 error = true; | |
| 274 return; | |
| 275 } | |
| 276 | |
| 277 if (fds[1].revents & POLLIN) { | |
| 278 read_pipe(); | |
| 279 } | |
| 280 | |
| 281 if (fds[0].revents & POLLOUT) { | |
| 282 switch(wsize = write(pipe_io[0][1], (char *)buffer + total, size - total)) { | |
| 283 case -1: | |
| 284 if (errno == EAGAIN) continue; | |
| 285 reason = string(strerror(errno)); | |
| 286 // close the pipes | |
| 287 close(pipe_io[0][1]); | |
| 288 close(pipe_io[1][0]); | |
| 289 pipe_io[0][1] = -1; | |
| 290 pipe_io[1][0] = -1; | |
| 291 // Slaughter child | |
| 292 kill(pid, SIGKILL); | |
| 293 // wait for child to terminate | |
| 294 waitpid(pid, &status, 0); | |
| 295 my_syslog(priv, "write error: " + reason); | |
| 296 error = true; | |
| 297 running = false; | |
| 298 return; | |
| 299 default: | |
| 300 total += wsize; | |
| 301 break; | |
| 302 } | |
| 303 } | |
| 304 } while ( total < size ); | |
| 305 } | |
| 306 | |
| 307 | |
| 308 void SpamAssassin::output(const char* buffer) | |
| 309 { | |
| 310 output(buffer, strlen(buffer)); | |
| 311 } | |
| 312 | |
| 313 | |
| 314 void SpamAssassin::output(string buffer) | |
| 315 { | |
| 316 output(buffer.c_str(), buffer.size()); | |
| 317 } | |
| 318 | |
| 319 | |
| 320 void SpamAssassin::close_output() | |
| 321 { | |
| 322 if (close(pipe_io[0][1])) | |
| 323 my_syslog(priv, "close error: " + string(strerror(errno))); | |
| 324 pipe_io[0][1] = -1; | |
| 325 } | |
| 326 | |
| 327 | |
| 328 void SpamAssassin::input() | |
| 329 { | |
| 330 if (!running || error) return; | |
| 331 empty_and_close_pipe(); | |
| 332 if (running) { | |
| 333 // wait until child is dead | |
| 334 int status; | |
| 335 if (waitpid(pid, &status, 0) < 0) { | |
| 336 error = true; | |
| 337 }; | |
| 338 } | |
| 339 running = false; | |
| 340 } | |
| 341 | |
| 342 | |
| 343 int SpamAssassin::read_pipe() | |
| 344 { | |
| 345 long size; | |
| 346 int status; | |
| 347 char iobuff[1024]; | |
| 348 string reason; | |
| 349 | |
| 350 if (pipe_io[1][0] == -1) return 0; | |
| 351 | |
| 352 size = read(pipe_io[1][0], iobuff, 1024); | |
| 353 | |
| 354 if (size < 0) { | |
| 355 reason = string(strerror(errno)); | |
| 356 // Close remaining pipe. | |
| 357 close(pipe_io[1][0]); | |
| 358 pipe_io[1][0] = -1; | |
| 359 // Slaughter child | |
| 360 kill(pid, SIGKILL); | |
| 361 // wait for child to terminate | |
| 362 waitpid(pid, &status, 0); | |
| 363 my_syslog(priv, "read error: " + reason); | |
| 364 size = 0; | |
| 365 error = true; | |
| 366 running = false; | |
| 367 } else if (size == 0) { | |
| 368 // EOF. Close the pipe | |
| 369 if (close(pipe_io[1][0])) { | |
| 370 error = true; | |
| 371 my_syslog(priv, "close error: " + string(strerror(errno))); | |
| 372 } | |
| 373 pipe_io[1][0] = -1; | |
| 374 } else { | |
| 375 // append to mail buffer | |
| 376 spamc_output.append(iobuff, size); | |
| 377 } | |
| 378 return size; | |
| 379 } | |
| 380 | |
| 381 | |
| 382 void SpamAssassin::empty_and_close_pipe() | |
| 383 { | |
| 384 while (read_pipe()) | |
| 385 ; | |
| 386 } | |
| 387 | |
| 388 | |
| 389 void SpamAssassin::closeall(int fd) | |
| 390 { | |
| 391 int fdlimit = sysconf(_SC_OPEN_MAX); | |
| 392 while (fd < fdlimit) | |
| 393 close(fd++); | |
| 394 } | |
| 395 | |
| 396 | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
397 const char *SpamAssassin::getorwarnmacro(SMFICTX *ctx, const char *macro, const char *def, const char *scope) |
| 168 | 398 { |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
399 const char *rc = smfi_getsymval(ctx, (char*)macro); |
| 168 | 400 if (!rc) { |
| 401 rc = def; | |
| 402 warnmacro(macro, scope); | |
| 403 } | |
| 404 return rc; | |
| 405 } | |
| 406 | |
| 407 | |
|
214
82886d4dd71f
Fixes to compile on Fedora 9 and for const correctness.
Carl Byington <carl@five-ten-sg.com>
parents:
203
diff
changeset
|
408 void SpamAssassin::warnmacro(const char *macro, const char *scope) |
| 163 | 409 { |
| 410 if (warnedmacro) return; | |
| 411 char buf[maxlen]; | |
| 412 snprintf(buf, sizeof(buf), "Could not retrieve sendmail macro %s. Add it to confMILTER_MACROS_%s for better results.", macro, scope); | |
| 413 my_syslog(priv, buf); | |
| 414 warnedmacro = true; | |
| 415 } | |
| 416 |