* Joe Loughry <loughry@uswest.net> sent a patch to handle multihomed machines.
* Changed mimedecode default to `off'; it seems that doing RFC2047 decoding
on headers throws away information that the MUA may need to see.
+* Change Received header parsing to no longer demand an embedded dot in
+ a mailhost address.
fetchmail-5.1.1 (Wed Sep 29 11:52:06 EDT 1999), 17827 lines:
* Added workaround, fetchmailconf warning, and FAQ about Novell GroupWise.
}
}
+/*
+ * Return zero on a syntactically invalid address, nz on a valid one.
+ *
+ * This used to be strchr(a, '.'), but it turns out that lines like this
+ *
+ * Received: from punt-1.mail.demon.net by mailstore for markb@ordern.com
+ * id 938765929:10:27223:2; Fri, 01 Oct 99 08:18:49 GMT
+ *
+ * are not uncommon. So now we just check that the following token is
+ * not itself an email address.
+ */
+#define VALID_ADDRESS(a) !strchr(rbuf, '@')
+
static char *parse_received(struct query *ctl, char *bufp)
/* try to extract real address from the Received line */
/* If a valid Received: line is found, we return the full address in
if (outlevel >= O_DEBUG)
report(stdout, _("analyzing Received line:\n%s"), bufp);
- /* search for whitepace-surrounded "by" followed by xxxx.yyyy */
+ /* search for whitepace-surrounded "by" followed by valid address */
for (base = bufp; ; base = ok + 2)
{
if (!(ok = strstr(base, "by")))
*tp++ = *sp;
*tp = '\0';
- /* look for embedded periods */
- if (strchr(rbuf, '.'))
+ /* look for valid address */
+ if (VALID_ADDRESS(rbuf))
break;
else
ok = sp - 1; /* arrange to skip this token */