Mercurial > dnsbl

comparison src/dnsbl.cpp @ 124:ea6f9c812faa stable-5-16

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
put hostname in smtp message for uribl style lookups
author carl
date Thu, 16 Mar 2006 15:20:37 -0800
parents ecd97e7eb1f0
children 05ae49d37896
comparison
equal deleted inserted replaced
123:ecd97e7eb1f0 124:ea6f9c812faa
656 656
657 657
658 //////////////////////////////////////////////// 658 ////////////////////////////////////////////////
659 // lookup the domain name part of a hostname on two lists 659 // lookup the domain name part of a hostname on two lists
660 // 660 //
661 bool uriblookup(mlfiPriv &priv ,char *hostname, char *top) ; 661 // if we find part of the hostname on the uribl, return
662 bool uriblookup(mlfiPriv &priv, char *hostname, char *top) { 662 // true and point found to the part of the hostname that we found.
663 // otherwise, return false and preserve the value of found.
664 //
665 bool uriblookup(mlfiPriv &priv ,char *hostname, char *top, char *&found) ;
666 bool uriblookup(mlfiPriv &priv, char *hostname, char *top, char *&found) {
663 // top is pointer to '.' char at end of base domain, or null for ip address form 667 // top is pointer to '.' char at end of base domain, or null for ip address form
664 // so for hostname of www.fred.mydomain.co.uk 668 // so for hostname of www.fred.mydomain.co.uk
665 // top points to-----------------------^ 669 // top points to-----------------------^
666 // and we end up looking at only mydomain.co.uk, ignoring the www.fred stuff 670 // and we end up looking at only mydomain.co.uk, ignoring the www.fred stuff
667 char buf[maxlen]; 671 char buf[maxlen];
668 char buf2[maxlen];
669 const char *uriblname[2] = { "multi.surbl.org", "multi.uribl.com" };
670
671 if (top) { 672 if (top) {
672 // add one more component 673 // add one more component
673 *top = '\0'; 674 *top = '\0';
674 char *x = strrchr(hostname, '.'); 675 char *x = strrchr(hostname, '.');
675 if (x) hostname = x+1; 676 if (x) hostname = x+1;
680 if (debug_syslog > 2) { 681 if (debug_syslog > 2) {
681 char tmp[maxlen]; 682 char tmp[maxlen];
682 snprintf(tmp, sizeof(tmp), "found %s on %s", hostname, priv.uribl_suffix); 683 snprintf(tmp, sizeof(tmp), "found %s on %s", hostname, priv.uribl_suffix);
683 my_syslog(tmp); 684 my_syslog(tmp);
684 } 685 }
686 found = hostname;
685 return true; 687 return true;
686 } 688 }
687 return false; 689 return false;
688 } 690 }
689 691
690 692
691 //////////////////////////////////////////////// 693 ////////////////////////////////////////////////
692 // uribl checker 694 // uribl checker
693 // ------------- 695 // -------------
694 // hostname MUST not have a trailing dot 696 // hostname MUST not have a trailing dot
695 // If tld, two level lookup. 697 // If tld, two level lookup.
696 // Else, look up three level domain. 698 // Else, look up three level domain.
697 bool check_uribl(mlfiPriv &priv, char *hostname) ; 699 //
698 bool check_uribl(mlfiPriv &priv, char *hostname) { 700 // if we find part of the hostname on the uribl, return
701 // true and point found to the part of the hostname that we found.
702 // otherwise, return false and preserve the value of found.
703 //
704 bool check_uribl(mlfiPriv &priv, char *hostname, char *&found) ;
705 bool check_uribl(mlfiPriv &priv, char *hostname, char *&found) {
699 in_addr ip; 706 in_addr ip;
700 if (inet_aton(hostname, &ip)) { 707 if (inet_aton(hostname, &ip)) {
701 const u_char *src = (const u_char *)&ip.s_addr; 708 const u_char *src = (const u_char *)&ip.s_addr;
702 char adr[sizeof "255.255.255.255"]; 709 static char adr[sizeof "255.255.255.255"];
703 snprintf(adr, sizeof(adr), "%u.%u.%u.%u", src[3], src[2], src[1], src[0]); 710 snprintf(adr, sizeof(adr), "%u.%u.%u.%u", src[3], src[2], src[1], src[0]);
704 return (uriblookup(priv, adr, NULL)); 711 return (uriblookup(priv, adr, NULL, found));
705 } 712 }
706 713
707 char *top, *top2, *top3; 714 char *top, *top2, *top3;
708 top = strrchr(hostname, '.'); 715 top = strrchr(hostname, '.');
709 if (top) { 716 if (top) {
713 720
714 if (top2) { 721 if (top2) {
715 string_set::iterator i = priv.memory->get_cctlds()->find(top2+1); 722 string_set::iterator i = priv.memory->get_cctlds()->find(top2+1);
716 string_set::iterator x = priv.memory->get_cctlds()->end(); 723 string_set::iterator x = priv.memory->get_cctlds()->end();
717 // if we have a 2-level-cctld, just look at top three levels of the name 724 // if we have a 2-level-cctld, just look at top three levels of the name
718 if (i != x) return uriblookup(priv, hostname, top2); 725 if (i != x) return uriblookup(priv, hostname, top2, found);
719 726
720 *top2 = '\0'; 727 *top2 = '\0';
721 top3 = strrchr(hostname, '.'); 728 top3 = strrchr(hostname, '.');
722 *top2 = '.'; 729 *top2 = '.';
723 730
724 // if we have more than 3 levels in the name, look at the top three levels of the name 731 // if we have more than 3 levels in the name, look at the top three levels of the name
725 if (top3 && uriblookup(priv, hostname, top2)) return true; 732 if (top3 && uriblookup(priv, hostname, top2, found)) return true;
726 // if that was not found, fall thru to looking at the top two levels 733 // if that was not found, fall thru to looking at the top two levels
727 } 734 }
728 // look at the top two levels of the name 735 // look at the top two levels of the name
729 return uriblookup(priv, hostname, top); 736 return uriblookup(priv, hostname, top, found);
730 } 737 }
731 return false; 738 return false;
732 } 739 }
733 740
734 741
735 //////////////////////////////////////////////// 742 ////////////////////////////////////////////////
736 // check the hosts from the body against the content filter and uribl dnsbls 743 // check the hosts from the body against the content filter and uribl dnsbls
737 // 744 //
738 bool check_hosts(mlfiPriv &priv, bool random, int limit, char *&msg, char *&host, int &ip); 745 //
739 bool check_hosts(mlfiPriv &priv, bool random, int limit, char *&msg, char *&host, int &ip) { 746 bool check_hosts(mlfiPriv &priv, bool random, int limit, char *&msg, char *&host, int &ip, char *&found);
747 bool check_hosts(mlfiPriv &priv, bool random, int limit, char *&msg, char *&host, int &ip, char *&found) {
748 found = NULL; // normally ip address style
740 if (!priv.content_suffix && !priv.uribl_suffix) return false; // nothing to check 749 if (!priv.content_suffix && !priv.uribl_suffix) return false; // nothing to check
741 CONFIG &dc = *priv.pc; 750 CONFIG &dc = *priv.pc;
742 string_set &hosts = priv.memory->get_hosts(); 751 string_set &hosts = priv.memory->get_hosts();
743 string_set &ignore = *priv.content_host_ignore; 752 string_set &ignore = *priv.content_host_ignore;
744 753
783 if (ip) { 792 if (ip) {
784 int_set::iterator i = ips.find(ip); 793 int_set::iterator i = ips.find(ip);
785 if (i == ips.end()) { 794 if (i == ips.end()) {
786 // we haven't looked this up yet 795 // we haven't looked this up yet
787 ips.insert(ip); 796 ips.insert(ip);
788 if (check_single(priv, ip, priv.content_suffix)) { 797 // check dnsbl style list
798 if (priv.content_suffix && check_single(priv, ip, priv.content_suffix)) {
789 msg = priv.content_message; 799 msg = priv.content_message;
790 return true; 800 return true;
791 } 801 }
792 // Check uribl & surbl 802 // Check uribl & surbl style list
793 if (check_uribl(priv, host)) { 803 if (priv.uribl_suffix && check_uribl(priv, host, found)) {
794 msg = priv.uribl_message; 804 msg = priv.uribl_message;
795 return true; 805 return true;
796 } 806 }
797 } 807 }
798 } 808 }
993 limit = max(limit, con.get_host_limit()); 1003 limit = max(limit, con.get_host_limit());
994 } 1004 }
995 } 1005 }
996 bool rejecting = alive.empty(); // if alive is empty, we must have set msg above in acceptable_content() 1006 bool rejecting = alive.empty(); // if alive is empty, we must have set msg above in acceptable_content()
997 if (!rejecting) { 1007 if (!rejecting) {
998 char *fmt; 1008 char *fmt, *found;
999 if (check_hosts(priv, random, limit, fmt, host, ip)) { 1009 if (check_hosts(priv, random, limit, fmt, host, ip, found)) {
1000 char adr[sizeof "255.255.255.255"]; 1010 if (found) {
1001 adr[0] = '\0'; 1011 // uribl style
1002 inet_ntop(AF_INET, (const u_char *)&ip, adr, sizeof(adr)); 1012 snprintf(buf, sizeof(buf), fmt, host, found);
1003 snprintf(buf, sizeof(buf), fmt, host, adr); 1013 }
1014 else {
1015 // dnsbl style
1016 char adr[sizeof "255.255.255.255"];
1017 adr[0] = '\0';
1018 inet_ntop(AF_INET, (const u_char *)&ip, adr, sizeof(adr));
1019 snprintf(buf, sizeof(buf), fmt, host, adr);
1020 }
1004 msg = buf; 1021 msg = buf;
1005 rejecting = true; 1022 rejecting = true;
1006 } 1023 }
1007 } 1024 }
1008 if (!rejecting) { 1025 if (!rejecting) {