Mercurial > dnsbl

comparison src/scanner.cpp @ 75:1142e46be550

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
start coding on new config syntax
author carl
date Wed, 13 Jul 2005 23:04:14 -0700
parents b7449114ebb0
children 81f1e400e8ab
comparison
equal deleted inserted replaced
74:b7449114ebb0 75:1142e46be550
7 */ 7 */
8 8
9 #include "includes.h" 9 #include "includes.h"
10 10
11 static char* scanner_version="$Id$"; 11 static char* scanner_version="$Id$";
12
13 ////////////////////////////////////////////////
14 // finite state machine
15 //
16 enum state {// host name recognizer states
17 h_init,
18 h_host,
19
20 // html tag discarder states
21 t_init,
22 t_tag1, // seen opening <
23 t_tag2, // not comment
24 t_com1, // seen !
25 t_com2, // seen first -
26 t_com3, // seen second -, looking for -->
27 t_com4, // seen first -
28 t_com5, // seen second -
29 t_disc, // looking for closing >
30
31 // url recognizer states
32 u_init,
33 u_http,
34 u_sla,
35 u_url,
36
37 // url decoder states %xx
38 d_init,
39 d_pcnt,
40 d_1,
41
42 // html entity decoder states &#nnn;
43 e_init,
44 e_amp,
45 e_num,
46
47 // mime decoder states =xx
48 m_init,
49 m_eq,
50 m_1,
51
52 // base64 decoder states
53 b_init,
54 b_lf,
55 b_lf2,
56 b_64,
57
58 // uuencoding decoder states
59 uu_init,
60 uu_lf,
61 uu_lf2,
62 uu_64,
63
64 // counter for number of columns in the table
65 end_state,
66
67 // temporary states
68 h_end,
69 t_bin,
70 t_end,
71 u_reco,
72 d_2,
73 e_semi,
74 m_2,
75 m_cr,
76 m_nl,
77 b_cr,
78 uu_cr
79 };
80
81 #define PENDING_LIMIT 100
82 class fsa {
83 u_char pending[PENDING_LIMIT];
84 int count;
85 state st;
86 state init;
87 fsa *next1;
88 fsa *next2;
89 recorder *memory;
90
91 public:
92 fsa(state init, fsa *next1_, fsa *next2_, recorder *memory_);
93 void push(u_char *buf, int len);
94 void pusher();
95 void error(char *err);
96 };
97
12 98
13 typedef state PARSE[end_state]; 99 typedef state PARSE[end_state];
14 100
15 static PARSE parse_table[256] = { 101 static PARSE parse_table[256] = {
16 // h_init, h_host, t_init, t_tag1, t_tag2, t_com1, t_com2, t_com3, t_com4, t_com5, t_disc, u_init, u_http, u_sla , u_url, d_init, d_pcnt, d_1, e_init, e_amp, e_num, m_init, m_eq, m_1, b_init, b_lf, b_lf2, b_64 uu_init, uu_lf, uu_lf2, uu_64 102 // h_init, h_host, t_init, t_tag1, t_tag2, t_com1, t_com2, t_com3, t_com4, t_com5, t_disc, u_init, u_http, u_sla , u_url, d_init, d_pcnt, d_1, e_init, e_amp, e_num, m_init, m_eq, m_1, b_init, b_lf, b_lf2, b_64 uu_init, uu_lf, uu_lf2, uu_64