3 * $Id: qmail-checkspam.c,v 1.2 2004/04/08 01:36:26 mdw Exp $
5 * Filter messages for spam
7 * (c) 2003 Mark Wooding
10 /*----- Licensing notice --------------------------------------------------*
12 * This program is free software; you can redistribute it and/or modify
13 * it under the terms of the GNU General Public License as published by
14 * the Free Software Foundation; either version 2 of the License, or
15 * (at your option) any later version.
17 * This program is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU General Public License for more details.
22 * You should have received a copy of the GNU General Public License
23 * along with this program; if not, write to the Free Software Foundation,
24 * Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
27 /*----- Header files ------------------------------------------------------*/
36 #include <sys/types.h>
37 #include <sys/unistd.h>
43 /*----- Main code ---------------------------------------------------------*/
45 static const char *strenv(const char *e, const char *d)
47 const char *p = getenv(e);
52 static double dblenv(const char *e, double d)
54 const char *p = getenv(e);
62 syslog(LOG_ERR, "bad floating value `%s' for %s'; ignoring", p, e);
69 static int intenv(const char *e, int d)
71 const char *p = getenv(e);
78 if (errno) return (d);
80 if (l < 0 || l > INT_MAX) {
81 syslog(LOG_ERR, "bad integer value `%s' for `%s'; ignoring", p, e);
87 static int safewrite(int fd, const void *p, size_t sz)
93 n = write(fd, pp, sz);
94 if (sz <= 0 && errno != EINTR && errno != EAGAIN)
102 static int shovel(int from, int to, const char *what)
108 n = read(from, buf, sizeof(buf));
109 if (n < 0 && errno != EINTR && errno != EAGAIN) {
110 syslog(LOG_ERR, "failed to read %s: %m", what);
115 if (safewrite(to, buf, n) < 0) {
116 syslog(LOG_ERR, "failed to write %s: %m", what);
127 char buf[ENVBUF + 1];
130 const char *rcpt[NRCPT];
134 static int readenv(struct envelope *e)
139 /* Read the raw envelope data. */
142 while (e->n < ENVBUF) {
143 n = read(1, e->buf + e->n, ENVBUF - e->n);
144 if (n < 0 && errno != EINTR && errno != EAGAIN) {
145 syslog(LOG_ERR, "failed to read envelope: %m");
153 /* Parse up the envelope data. */
154 p = e->buf; l = p + e->n;
157 e->send = "<invalid>";
158 syslog(LOG_ERR, "corrupt envelope (no sender marker)");
163 if (p >= l) goto trunc;
164 else if (!*p++) break;
168 if (p >= l) goto trunc;
170 else if (*p++ != 'T') {
171 syslog(LOG_ERR, "corrupt envelope (no recipient marker)");
174 if (p >= l) goto trunc;
175 if (e->nrcpt < NRCPT) e->rcpt[e->nrcpt++] = p;
177 if (p >= l) goto trunc;
178 else if (!*p++) break;
183 /* Failed to reach the final terminator; maybe there's more. This isn't a
184 * very bad situation.
191 static void real(void)
194 strenv("QMAIL_CHECKSPAM_QUEUE", "/var/qmail/bin/qmail-queue");
195 execlp(qmq, qmq, (char *)0);
196 syslog(LOG_ERR, "failed to exec %s: %m", qmq);
200 static int split(int *fd_m, int *fd_e)
205 if (pipe(pm) || pipe(pe)) return (-1);
206 if ((kid = fork()) < 0) return (-1);
208 dup2(pm[0], 0); close(pm[0]); close(pm[1]);
209 dup2(pe[0], 1); close(pe[0]); close(pe[1]);
212 close(pm[0]); *fd_m = pm[1];
213 close(pe[0]); *fd_e = pe[1];
218 static int writemsg(struct message *m, int fd_m, int shovelp)
220 if (message_write(fd_m, m) < 0) {
221 syslog(LOG_ERR, "failed to write message body: %m");
224 if (shovelp && shovel(0, fd_m, "message body"))
230 static int writeenv(struct envelope *e, int fd_e)
232 if (safewrite(fd_e, e->buf, e->n) < 0) {
233 syslog(LOG_ERR, "failed to write envelope: %m");
236 if ((e->f & EF_TRUNC) && shovel(1, fd_e, "envelope"))
242 static void logenv(struct envelope *e)
246 syslog(LOG_NOTICE, "sender = %s", e->send);
247 for (i = 0; i < e->nrcpt; i++)
248 syslog(LOG_NOTICE, "recipient %lu = %s", (unsigned long)i, e->rcpt[i]);
251 int main(int argc, char *argv[])
261 /* Set up logging output. */
262 if ((p = strrchr(argv[0], '/')) == 0) p = argv[0]; else p++;
263 openlog(p, LOG_PID, LOG_MAIL);
265 /* Read configuration from the environment. */
266 if (getenv("RELAYCLIENT")) real();
267 m.max_len = intenv("QMAIL_CHECKSPAM_MAXLEN", 2 * 1024 * 1024);
268 m.timeout = intenv("QMAIL_CHECKSPAM_TIMEOUT", 300);
270 /* Slurp an initial chunk of the message. */
271 rc = message_read(0, 0, &m);
276 /* We read the message body OK. Now try to slurp in the envelope and
277 * write a useful message.
279 if (readenv(&e)) return (54);
280 syslog(LOG_NOTICE, "scanning message");
283 /* Filter the message and see what happens. */
284 p = strenv("QMAIL_CHECKSPAM_SPAMDHOST", "localhost");
285 port = intenv("QMAIL_CHECKSPAM_SPAMDPORT", 783);
286 if (lookup_host(p, port, &sa)) {
287 syslog(LOG_ERR, "lookup failed for host `%s', port %d", p, port);
290 if (message_filter(&sa, "spamd", 0, &m)) {
291 syslog(LOG_ERR, "filter failed unexpectedly");
295 /* Check the resulting score. */
296 thresh = dblenv("QMAIL_CHECKSPAM_THRESH", m.threshold);
297 if (m.score >= thresh) {
298 syslog(LOG_NOTICE, "rejecting: score %g >= %g", m.score, thresh);
301 syslog(LOG_NOTICE, "accepting: score %g < %g", m.score, thresh);
303 /* Send the remaining stuff on to the real server. */
304 if (split(&fd_m, &fd_e)) return (56);
305 if (writemsg(&m, fd_m, 0) || writeenv(&e, fd_e)) _exit(127);
309 /* Message was too big. We must pass it all on to the real server
310 * before picking up the envelope.
312 syslog(LOG_NOTICE, "message too large for filtering");
313 if (split(&fd_m, &fd_e)) return (56);
314 if (writemsg(&m, fd_m, 1) || readenv(&e)) _exit(127);
316 if (writeenv(&e, fd_e)) _exit(127);
320 syslog(LOG_ERR, "failed to read message (rc = %d): %m", rc);
327 /*----- That's all, folks -------------------------------------------------*/