Mercurial > dnsbl

comparison src/scanner.cpp @ 19:b8f5fa3dd5b8

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
fix problems in the state transitions causing impossible states
author carl
date Fri, 30 Apr 2004 22:44:56 -0700
parents 041ea016b684
children 06de5ab6a232
comparison
equal deleted inserted replaced
18:041ea016b684 19:b8f5fa3dd5b8
33 b_64, 33 b_64,
34 34
35 // counter for number of columns in the table 35 // counter for number of columns in the table
36 end_state, 36 end_state,
37 37
38 // temporary mime states 38 // temporary states
39 h_end, 39 h_end,
40 t_end, 40 t_end,
41 u_reco, 41 u_reco,
42 e_semi, 42 e_semi,
43 m_2, 43 m_2,
132 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4E N 132 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4E N
133 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4F O 133 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4F O
134 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x50 P 134 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x50 P
135 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x51 Q 135 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x51 Q
136 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x52 R 136 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x52 R
137 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x53 S 137 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x53 S
138 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x54 T 138 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x54 T
139 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x55 U 139 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x55 U
140 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x56 V 140 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x56 V
141 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x57 W 141 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x57 W
142 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x58 X 142 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x58 X
164 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6E n 164 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6E n
165 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6F o 165 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6F o
166 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x70 p 166 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x70 p
167 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x71 q 167 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x71 q
168 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x72 r 168 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x72 r
169 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x73 s 169 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x73 s
170 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x74 t 170 {h_host, h_host, t_init, t_disc, u_http, u_http, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x74 t
171 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x75 u 171 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x75 u
172 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x76 v 172 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x76 v
173 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x77 w 173 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x77 w
174 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x78 x 174 {h_host, h_host, t_init, t_disc, u_init, u_init, u_url, u_url, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x78 x
847 string_set *hosts; 847 string_set *hosts;
848 848
849 fsa(state init, fsa* next1_, fsa* next2_, string_set *hosts_); 849 fsa(state init, fsa* next1_, fsa* next2_, string_set *hosts_);
850 void push(u_char *buf, int len); 850 void push(u_char *buf, int len);
851 void pusher(); 851 void pusher();
852 void error(char *err);
852 }; 853 };
853 854
854 fsa::fsa(state init_, fsa *next1_, fsa *next2_, string_set *hosts_) { 855 fsa::fsa(state init_, fsa *next1_, fsa *next2_, string_set *hosts_) {
855 count = 0; 856 count = 0;
856 st = init_; 857 st = init_;
858 next1 = next1_; 859 next1 = next1_;
859 next2 = next2_; 860 next2 = next2_;
860 hosts = hosts_; 861 hosts = hosts_;
861 } 862 }
862 863
864 void fsa::error(char *err) {
865 count = 0;
866 st = init;
867 if (err) my_syslog(err);
868 }
869
863 void fsa::pusher() { 870 void fsa::pusher() {
864 if (next1) next1->push(pending, count); 871 if (next1) next1->push(pending, count);
865 if (next2) next2->push(pending, count); 872 if (next2) next2->push(pending, count);
866 count = 0; 873 count = 0;
867 } 874 }
868 875
869 void fsa::push(u_char *buf, int len) { 876 void fsa::push(u_char *buf, int len) {
870 for (int i=0; i<len; i++) { 877 for (int i=0; i<len; i++) {
878 if (count == (PENDING_LIMIT-1)) error(NULL);
879 if (st >= end_state) error("finite state machine impossible state");
871 u_char c = buf[i]; 880 u_char c = buf[i];
872 // guard against buffer overflow
873 if (count == (PENDING_LIMIT-1)) {
874 pusher();
875 st = init;
876 }
877 pending[count++] = c; 881 pending[count++] = c;
878 st = parse_table[c][st]; 882 st = parse_table[c][st];
879 switch (st) { 883 switch (st) {
880 884
881 ////////////////////////////// 885 //////////////////////////////
890 register_string(*hosts, (char*)pending); 894 register_string(*hosts, (char*)pending);
891 break; 895 break;
892 } 896 }
893 } 897 }
894 } 898 }
899 st = h_init;
895 } // fall thru 900 } // fall thru
896 901
897 case h_init: { 902 case h_init: {
898 count = 0; 903 count = 0;
899 } break; 904 } break;
913 } break; 918 } break;
914 919
915 ////////////////////////////// 920 //////////////////////////////
916 // url recognizer 921 // url recognizer
917 case u_sla: { 922 case u_sla: {
918 if ((count < 6) || (7 < count)) { 923 if ((count < 6) || (8 < count)) { // allow http:// or https://
919 count = 0; 924 count = 0;
920 st = u_init; 925 st = u_init;
921 } 926 }
922 } break; 927 } break;
923 928
924 case u_reco: { 929 case u_reco: {
925 if (count > 12) { 930 if (count > 13) { // need some minimal length host name after the protocol
926 pending[count-1] = 0; 931 pending[--count] = '\0'; // null terminate host name by overwriting the terminator
932 char *p = NULL;
927 if (strncasecmp((const char *)pending, "http://", 7) == 0) { 933 if (strncasecmp((const char *)pending, "http://", 7) == 0) {
928 char *p = (char *)pending + 7; 934 p = (char *)pending + 7;
929 if (strchr(p, '.')) register_string(*hosts, p); // require at least one . in a dns name
930 } 935 }
936 if (strncasecmp((const char *)pending, "https://", 8) == 0) {
937 p = (char *)pending + 8;
938 }
939 if (p && strchr(p, '.')) register_string(*hosts, p); // require at least one . in a dns name
931 } 940 }
941 st = u_init;
932 } // fall thru 942 } // fall thru
933 943
934 case u_init: { 944 case u_init: {
935 count = 0; // discard all characters 945 count = 0; // discard all characters
936 } break; 946 } break;