Mercurial > dnsbl

comparison src/scanner.cpp @ 12:6ac6d6b822ce stable-2-0

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
fix memory leak with duplicate url host names, document differences from sendmail.mc feature
author carl
date Fri, 23 Apr 2004 22:45:10 -0700
parents 2c206836b4cc
children 2ae8d953f1d0
comparison
equal deleted inserted replaced
11:2c206836b4cc 12:6ac6d6b822ce
818 0, // 0xff 818 0, // 0xff
819 }; 819 };
820 820
821 #define PENDING_LIMIT 100 821 #define PENDING_LIMIT 100
822 struct fsa { 822 struct fsa {
823 u_char pending[PENDING_LIMIT]; 823 u_char pending[PENDING_LIMIT];
824 int count; 824 int count;
825 state st; 825 state st;
826 state init; 826 state init;
827 fsa* next; 827 fsa* next;
828 string_set *urls; 828 string_set *hosts;
829 829
830 fsa(state init, fsa* next_, string_set *urls_); 830 fsa(state init, fsa* next_, string_set *hosts_);
831 void push(u_char *buf, int len); 831 void push(u_char *buf, int len);
832 }; 832 };
833 833
834 fsa::fsa(state init_, fsa *next_, string_set *urls_) { 834 fsa::fsa(state init_, fsa *next_, string_set *hosts_) {
835 count = 0; 835 count = 0;
836 st = init_; 836 st = init_;
837 init = init_; 837 init = init_;
838 next = next_; 838 next = next_;
839 urls = urls_; 839 hosts = hosts_;
840 } 840 }
841 841
842 void fsa::push(u_char *buf, int len) { 842 void fsa::push(u_char *buf, int len) {
843 for (int i=0; i<len; i++) { 843 for (int i=0; i<len; i++) {
844 u_char c = buf[i]; 844 u_char c = buf[i];
864 case u_reco: { 864 case u_reco: {
865 if (count > 12) { 865 if (count > 12) {
866 pending[count-1] = 0; 866 pending[count-1] = 0;
867 if (strncasecmp((const char *)pending, "http://", 7) == 0) { 867 if (strncasecmp((const char *)pending, "http://", 7) == 0) {
868 char *p = (char *)pending + 7; 868 char *p = (char *)pending + 7;
869 if (strchr(p, '.')) urls->insert(strdup(p)); // require at least one . in a dns name 869 if (strchr(p, '.')) register_string(*hosts, p); // require at least one . in a dns name
870 } 870 }
871 } 871 }
872 } // fall thru 872 } // fall thru
873 873
874 case u_init: { 874 case u_init: {
967 fsa *urls_parser; 967 fsa *urls_parser;
968 fsa *html_parser; 968 fsa *html_parser;
969 fsa *mime_parser; 969 fsa *mime_parser;
970 fsa *b64_parser; 970 fsa *b64_parser;
971 971
972 url_scanner(string_set *urls); 972 url_scanner(string_set *hosts);
973 ~url_scanner(); 973 ~url_scanner();
974 void scan(u_char *buffer, size_t length); 974 void scan(u_char *buffer, size_t length);
975 }; 975 };
976 976
977 url_scanner::url_scanner(string_set *urls) { 977 url_scanner::url_scanner(string_set *hosts) {
978 urls_parser = new fsa(u_init, NULL, urls); 978 urls_parser = new fsa(u_init, NULL, hosts);
979 html_parser = new fsa(e_init, urls_parser, NULL); 979 html_parser = new fsa(e_init, urls_parser, NULL);
980 mime_parser = new fsa(m_init, html_parser, NULL); 980 mime_parser = new fsa(m_init, html_parser, NULL);
981 b64_parser = new fsa(b_init, mime_parser, NULL); 981 b64_parser = new fsa(b_init, mime_parser, NULL);
982 } 982 }
983 983