Mercurial > dnsbl
annotate src/scanner.cpp @ 27:43a4f6b3e668 stable-2-3
add configurable host name limit and bad html tag limits.
| author | carl |
|---|---|
| date | Sat, 22 May 2004 22:30:45 -0700 |
| parents | 6176e7b2e8af |
| children | 33e1e3910506 |
| rev | line source |
|---|---|
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1 /* |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
2 |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
3 Copyright (c) 2004 Carl Byington - 510 Software Group, released under |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
4 the GPL version 2 or any later version at your choice available at |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
5 http://www.fsf.org/licenses/gpl.txt |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
6 |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
7 */ |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
8 |
| 8 | 9 static char* scanner_version="$Id$"; |
| 6 | 10 |
| 11 using namespace std; | |
| 12 | |
| 24 | 13 |
| 14 // object to record things we see in the body content | |
| 15 struct recorder | |
| 16 { | |
| 17 string_set *html_tags; // valid tags | |
| 18 string_set hosts; | |
| 19 int bad_html_tags; | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
20 int binary_tags; |
| 24 | 21 recorder(string_set *html_tags_); |
| 22 ~recorder(); | |
| 23 void empty(); | |
| 24 void new_url(char *host); | |
| 25 void new_tag(char *tag); | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
26 void binary(); |
| 24 | 27 }; |
| 28 recorder::recorder(string_set *html_tags_) { | |
| 29 html_tags = html_tags_; | |
| 30 bad_html_tags = 0; | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
31 binary_tags = 0; |
| 24 | 32 } |
| 33 recorder::~recorder() { | |
| 34 empty(); | |
| 35 } | |
| 36 void recorder::empty() { | |
| 37 bad_html_tags = 0; | |
| 38 discard(hosts); | |
| 39 } | |
| 40 void recorder::new_url(char *host) { | |
| 41 register_string(hosts, host); | |
| 42 } | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
43 void recorder::binary() { |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
44 binary_tags++; |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
45 } |
| 24 | 46 void recorder::new_tag(char *tag) { |
| 47 string_set::iterator i = html_tags->find(tag); | |
| 48 if (i == html_tags->end()) { | |
| 49 bad_html_tags++; | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
50 if (debug_syslog && (bad_html_tags < 10)) { |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
51 // only log the first 10 bad tags |
| 24 | 52 char buf[200]; |
| 53 snprintf(buf, sizeof(buf), "bad html tag %s", tag); | |
| 54 my_syslog(buf); | |
| 55 } | |
| 56 } | |
| 57 } | |
| 58 | |
| 59 | |
| 60 | |
| 16 | 61 enum state {// host name recognizer states |
| 62 h_init, | |
| 63 h_host, | |
| 64 | |
| 65 // html tag discarder states | |
| 66 t_init, | |
| 24 | 67 t_tag, |
| 16 | 68 t_disc, |
| 69 | |
| 70 // url recognizer states | |
| 7 | 71 u_init, |
| 72 u_http, | |
| 73 u_sla, | |
| 74 u_url, | |
| 6 | 75 |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
76 // url decoder states %xx |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
77 d_init, |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
78 d_pcnt, |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
79 d_1, |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
80 |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
81 // html entity decoder states &#nnn; |
| 7 | 82 e_init, |
| 83 e_amp, | |
| 84 e_num, | |
| 85 | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
86 // mime decoder states =xx |
| 7 | 87 m_init, |
| 6 | 88 m_eq, |
| 89 m_1, | |
| 90 | |
| 7 | 91 // base64 decoder states |
| 92 b_init, | |
| 6 | 93 b_lf, |
| 94 b_lf2, | |
| 95 b_64, | |
| 96 | |
| 7 | 97 // counter for number of columns in the table |
| 98 end_state, | |
| 6 | 99 |
|
19
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
100 // temporary states |
| 16 | 101 h_end, |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
102 t_bin, |
| 16 | 103 t_end, |
| 7 | 104 u_reco, |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
105 d_2, |
| 7 | 106 e_semi, |
| 107 m_2, | |
| 6 | 108 m_cr, |
| 109 m_nl, | |
| 110 b_cr, | |
| 111 }; | |
| 112 | |
| 113 typedef state PARSE[end_state]; | |
| 114 | |
| 115 static PARSE parse_table[256] = { | |
| 24 | 116 // h_init, h_host, t_init, t_tag, 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 |
| 6 | 117 |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
118 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x00 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
119 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x01 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
120 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x02 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
121 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x03 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
122 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x04 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
123 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x05 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
124 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x06 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
125 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x07 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
126 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x08 |
| 24 | 127 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x09 <tab> |
| 128 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_nl, m_init, b_lf, b_init, b_lf2, b_init, }, // 0x0a <lf> | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
129 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x0b |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
130 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x0c |
| 24 | 131 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_cr, m_init, b_init, b_init, b_init, b_cr, }, // 0x0d <cr> |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
132 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x0e |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
133 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x0f |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
134 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x10 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
135 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x11 xon char |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
136 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x12 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
137 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x13 xoff char |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
138 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x14 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
139 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x15 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
140 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x16 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
141 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x17 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
142 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x18 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
143 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x19 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
144 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x1a |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
145 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x1b |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
146 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x1c |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
147 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x1d |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
148 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x1e |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
149 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x1f |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
150 {h_init, h_end, t_init, t_end, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x20 space |
| 24 | 151 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x21 ! |
| 152 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x22 "" | |
| 153 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_num, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x23 # | |
| 154 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x24 $ | |
| 155 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_pcnt, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x25 % | |
| 156 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_amp, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x26 & | |
| 157 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x27 ' | |
| 158 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x28 ( | |
| 159 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x29 ) | |
| 160 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x2A * | |
| 161 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x2B + | |
| 162 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x2C , | |
| 163 {h_host, h_host, t_init, t_disc, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x2D - | |
| 164 {h_host, h_host, t_init, t_disc, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x2E . | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
165 {h_init, h_end, t_init, t_tag, t_disc, u_init, u_sla, u_sla, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x2F / |
| 24 | 166 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x30 0 |
| 167 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x31 1 | |
| 168 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x32 2 | |
| 169 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x33 3 | |
| 170 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x34 4 | |
| 171 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x35 5 | |
| 172 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x36 6 | |
| 173 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x37 7 | |
| 174 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x38 8 | |
| 175 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_num, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x39 9 | |
| 176 {h_init, h_end, t_init, t_disc, t_disc, u_http, u_http, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x3A : | |
| 177 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_semi, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x3B ; | |
| 178 {h_init, h_end, t_tag, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x3C < | |
| 179 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_eq, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x3D = | |
| 180 {h_init, h_end, t_init, t_end, t_end, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x3E > | |
| 181 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x3F ? | |
|
27
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
182 {h_init, h_host, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x40 @ |
| 24 | 183 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x41 A |
| 184 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x42 B | |
| 185 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x43 C | |
| 186 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x44 D | |
| 187 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x45 E | |
| 188 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x46 F | |
| 189 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x47 G | |
| 190 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x48 H | |
| 191 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x49 I | |
| 192 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4A J | |
| 193 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4B K | |
| 194 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4C L | |
| 195 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4D M | |
| 196 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4E N | |
| 197 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x4F O | |
| 198 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x50 P | |
| 199 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x51 Q | |
| 200 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x52 R | |
| 201 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x53 S | |
| 202 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x54 T | |
| 203 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x55 U | |
| 204 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x56 V | |
| 205 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x57 W | |
| 206 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x58 X | |
| 207 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x59 Y | |
| 208 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x5A Z | |
| 209 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x5B [ | |
| 210 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x5C brace | |
| 211 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x5D ] | |
| 212 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x5E ^ | |
| 213 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x5F _ | |
| 214 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x60 ` | |
| 215 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x61 a | |
| 216 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x62 b | |
| 217 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x63 c | |
| 218 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x64 d | |
| 219 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x65 e | |
| 220 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_1, d_2, e_init, e_init, e_init, m_init, m_1, m_2, b_init, b_64, b_64, b_64, }, // 0x66 f | |
| 221 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x67 g | |
| 222 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x68 h | |
| 223 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x69 i | |
| 224 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6A j | |
| 225 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6B k | |
| 226 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6C l | |
| 227 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6D m | |
| 228 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6E n | |
| 229 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x6F o | |
| 230 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x70 p | |
| 231 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x71 q | |
| 232 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x72 r | |
| 233 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x73 s | |
| 234 {h_host, h_host, t_init, t_tag, t_disc, u_http, u_http, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x74 t | |
| 235 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x75 u | |
| 236 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x76 v | |
| 237 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x77 w | |
| 238 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x78 x | |
| 239 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x79 y | |
| 240 {h_host, h_host, t_init, t_tag, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x7A z | |
| 241 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x7B { | |
| 242 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x7C | | |
| 243 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x7D } | |
| 244 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x7E ~ | |
| 245 {h_init, h_end, t_init, t_disc, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x7f | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
246 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x80 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
247 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x81 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
248 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x82 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
249 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x83 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
250 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x84 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
251 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x85 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
252 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x86 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
253 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x87 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
254 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x88 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
255 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x89 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
256 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x8a |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
257 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x8b |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
258 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x8c |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
259 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x8d |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
260 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x8e |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
261 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x8f |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
262 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x90 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
263 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x91 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
264 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x92 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
265 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x93 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
266 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x94 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
267 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x95 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
268 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x96 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
269 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x97 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
270 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x98 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
271 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x99 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
272 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x9a |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
273 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x9b |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
274 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x9c |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
275 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x9d |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
276 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x9e |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
277 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0x9f |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
278 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa0 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
279 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa1 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
280 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa2 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
281 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa3 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
282 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa4 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
283 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa5 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
284 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa6 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
285 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa7 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
286 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa8 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
287 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xa9 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
288 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xaa |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
289 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xab |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
290 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xac |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
291 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xad |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
292 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xae |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
293 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xaf |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
294 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb0 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
295 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb1 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
296 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb2 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
297 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb3 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
298 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb4 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
299 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb5 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
300 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb6 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
301 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb7 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
302 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb8 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
303 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xb9 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
304 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xba |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
305 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xbb |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
306 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xbc |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
307 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xbd |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
308 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xbe |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
309 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xbf |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
310 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc0 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
311 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc1 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
312 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc2 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
313 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc3 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
314 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc4 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
315 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc5 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
316 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc6 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
317 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc7 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
318 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc8 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
319 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xc9 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
320 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xca |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
321 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xcb |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
322 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xcc |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
323 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xcd |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
324 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xce |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
325 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xcf |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
326 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd0 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
327 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd1 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
328 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd2 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
329 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd3 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
330 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd4 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
331 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd5 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
332 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd6 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
333 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd7 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
334 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd8 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
335 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xd9 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
336 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xda |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
337 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xdb |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
338 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xdc |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
339 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xdd |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
340 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xde |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
341 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xdf |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
342 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe0 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
343 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe1 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
344 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe2 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
345 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe3 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
346 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe4 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
347 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe5 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
348 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe6 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
349 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe7 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
350 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe8 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
351 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xe9 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
352 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xea |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
353 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xeb |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
354 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xec |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
355 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xed |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
356 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xee |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
357 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xef |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
358 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf0 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
359 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf1 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
360 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf2 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
361 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf3 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
362 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf4 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
363 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf5 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
364 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf6 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
365 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf7 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
366 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf8 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
367 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xf9 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
368 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xfa |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
369 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xfb |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
370 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xfc |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
371 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xfd |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
372 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xfe |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
373 {h_init, h_end, t_init, t_bin, t_disc, u_init, u_init, u_init, u_reco, d_init, d_init, d_init, e_init, e_init, e_init, m_init, m_init, m_init, b_init, b_init, b_init, b_init, }, // 0xff |
| 6 | 374 }; |
| 375 | |
| 376 | |
| 16 | 377 char *tlds[] = { |
| 378 ".com", | |
| 379 ".net", | |
| 380 ".org", | |
| 381 ".biz", | |
|
27
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
382 ".info", |
| 16 | 383 NULL |
| 384 }; | |
| 385 | |
| 6 | 386 u_char hex_decode[256] = { |
| 387 0, // 0x00 | |
| 388 0, // 0x01 | |
| 389 0, // 0x02 | |
| 390 0, // 0x03 | |
| 391 0, // 0x04 | |
| 392 0, // 0x05 | |
| 393 0, // 0x06 | |
| 394 0, // 0x07 | |
| 395 0, // 0x08 | |
| 7 | 396 0, // 0x09 <tab> |
| 397 0, // 0x0a <lf> | |
| 6 | 398 0, // 0x0b |
| 399 0, // 0x0c | |
| 7 | 400 0, // 0x0d <cr> |
| 6 | 401 0, // 0x0e |
| 402 0, // 0x0f | |
| 403 0, // 0x10 | |
| 404 0, // 0x11 xon char | |
| 405 0, // 0x12 | |
| 406 0, // 0x13 xoff char | |
| 407 0, // 0x14 | |
| 408 0, // 0x15 | |
| 409 0, // 0x16 | |
| 410 0, // 0x17 | |
| 411 0, // 0x18 | |
| 412 0, // 0x19 | |
| 413 0, // 0x1a | |
| 414 0, // 0x1b | |
| 415 0, // 0x1c | |
| 416 0, // 0x1d | |
| 417 0, // 0x1e | |
| 418 0, // 0x1f | |
| 419 0, // 0x20 space | |
| 420 0, // 0x21 ! | |
| 421 0, // 0x22 "" | |
| 422 0, // 0x23 # | |
| 423 0, // 0x24 $ | |
| 424 0, // 0x25 % | |
| 425 0, // 0x26 & | |
| 426 0, // 0x27 ' | |
| 427 0, // 0x28 ( | |
| 428 0, // 0x29 ) | |
| 429 0, // 0x2A * | |
| 430 0, // 0x2B + | |
| 431 0, // 0x2C , | |
| 432 0, // 0x2D - | |
| 433 0, // 0x2E . | |
| 434 0, // 0x2F / | |
| 435 0, // 0x30 0 | |
| 436 1, // 0x31 1 | |
| 437 2, // 0x32 2 | |
| 438 3, // 0x33 3 | |
| 439 4, // 0x34 4 | |
| 440 5, // 0x35 5 | |
| 441 6, // 0x36 6 | |
| 442 7, // 0x37 7 | |
| 443 8, // 0x38 8 | |
| 444 9, // 0x39 9 | |
| 445 0, // 0x3A : | |
| 446 0, // 0x3B ; | |
| 447 0, // 0x3C < | |
| 448 0, // 0x3D = | |
| 449 0, // 0x3E > | |
| 450 0, // 0x3F ? | |
| 451 0, // 0x40 @ | |
| 452 10, // 0x41 A | |
| 453 11, // 0x42 B | |
| 454 12, // 0x43 C | |
| 455 13, // 0x44 D | |
| 456 14, // 0x45 E | |
| 457 15, // 0x46 F | |
| 458 0, // 0x47 G | |
| 459 0, // 0x48 H | |
| 460 0, // 0x49 I | |
| 461 0, // 0x4A J | |
| 462 0, // 0x4B K | |
| 463 0, // 0x4C L | |
| 464 0, // 0x4D M | |
| 465 0, // 0x4E N | |
| 466 0, // 0x4F O | |
| 467 0, // 0x50 P | |
| 468 0, // 0x51 Q | |
| 469 0, // 0x52 R | |
| 470 0, // 0x53 S | |
| 471 0, // 0x54 T | |
| 472 0, // 0x55 U | |
| 473 0, // 0x56 V | |
| 474 0, // 0x57 W | |
| 475 0, // 0x58 X | |
| 476 0, // 0x59 Y | |
| 477 0, // 0x5A Z | |
| 478 0, // 0x5B [ | |
| 479 0, // 0x5C brace | |
| 480 0, // 0x5D ] | |
| 481 0, // 0x5E ^ | |
| 482 0, // 0x5F _ | |
| 483 0, // 0x60 ` | |
| 484 10, // 0x61 a | |
| 485 11, // 0x62 b | |
| 486 12, // 0x63 c | |
| 487 13, // 0x64 d | |
| 488 14, // 0x65 e | |
| 489 15, // 0x66 f | |
| 490 0, // 0x67 g | |
| 491 0, // 0x68 h | |
| 492 0, // 0x69 i | |
| 493 0, // 0x6A j | |
| 494 0, // 0x6B k | |
| 495 0, // 0x6C l | |
| 496 0, // 0x6D m | |
| 497 0, // 0x6E n | |
| 498 0, // 0x6F o | |
| 499 0, // 0x70 p | |
| 500 0, // 0x71 q | |
| 501 0, // 0x72 r | |
| 502 0, // 0x73 s | |
| 503 0, // 0x74 t | |
| 504 0, // 0x75 u | |
| 505 0, // 0x76 v | |
| 506 0, // 0x77 w | |
| 507 0, // 0x78 x | |
| 508 0, // 0x79 y | |
| 509 0, // 0x7A z | |
| 510 0, // 0x7B { | |
| 511 0, // 0x7C | | |
| 512 0, // 0x7D } | |
| 513 0, // 0x7E ~ | |
| 514 0, // 0x7f | |
| 515 0, // 0x80 | |
| 516 0, // 0x81 | |
| 517 0, // 0x82 | |
| 518 0, // 0x83 | |
| 519 0, // 0x84 | |
| 520 0, // 0x85 | |
| 521 0, // 0x86 | |
| 522 0, // 0x87 | |
| 523 0, // 0x88 | |
| 524 0, // 0x89 | |
| 525 0, // 0x8a | |
| 526 0, // 0x8b | |
| 527 0, // 0x8c | |
| 528 0, // 0x8d | |
| 529 0, // 0x8e | |
| 530 0, // 0x8f | |
| 531 0, // 0x90 | |
| 532 0, // 0x91 | |
| 533 0, // 0x92 | |
| 534 0, // 0x93 | |
| 535 0, // 0x94 | |
| 536 0, // 0x95 | |
| 537 0, // 0x96 | |
| 538 0, // 0x97 | |
| 539 0, // 0x98 | |
| 540 0, // 0x99 | |
| 541 0, // 0x9a | |
| 542 0, // 0x9b | |
| 543 0, // 0x9c | |
| 544 0, // 0x9d | |
| 545 0, // 0x9e | |
| 546 0, // 0x9f | |
| 547 0, // 0xa0 | |
| 548 0, // 0xa1 | |
| 549 0, // 0xa2 | |
| 550 0, // 0xa3 | |
| 551 0, // 0xa4 | |
| 552 0, // 0xa5 | |
| 553 0, // 0xa6 | |
| 554 0, // 0xa7 | |
| 555 0, // 0xa8 | |
| 556 0, // 0xa9 | |
| 557 0, // 0xaa | |
| 558 0, // 0xab | |
| 559 0, // 0xac | |
| 560 0, // 0xad | |
| 561 0, // 0xae | |
| 562 0, // 0xaf | |
| 563 0, // 0xb0 | |
| 564 0, // 0xb1 | |
| 565 0, // 0xb2 | |
| 566 0, // 0xb3 | |
| 567 0, // 0xb4 | |
| 568 0, // 0xb5 | |
| 569 0, // 0xb6 | |
| 570 0, // 0xb7 | |
| 571 0, // 0xb8 | |
| 572 0, // 0xb9 | |
| 573 0, // 0xba | |
| 574 0, // 0xbb | |
| 575 0, // 0xbc | |
| 576 0, // 0xbd | |
| 577 0, // 0xbe | |
| 578 0, // 0xbf | |
| 579 0, // 0xc0 | |
| 580 0, // 0xc1 | |
| 581 0, // 0xc2 | |
| 582 0, // 0xc3 | |
| 583 0, // 0xc4 | |
| 584 0, // 0xc5 | |
| 585 0, // 0xc6 | |
| 586 0, // 0xc7 | |
| 587 0, // 0xc8 | |
| 588 0, // 0xc9 | |
| 589 0, // 0xca | |
| 590 0, // 0xcb | |
| 591 0, // 0xcc | |
| 592 0, // 0xcd | |
| 593 0, // 0xce | |
| 594 0, // 0xcf | |
| 595 0, // 0xd0 | |
| 596 0, // 0xd1 | |
| 597 0, // 0xd2 | |
| 598 0, // 0xd3 | |
| 599 0, // 0xd4 | |
| 600 0, // 0xd5 | |
| 601 0, // 0xd6 | |
| 602 0, // 0xd7 | |
| 603 0, // 0xd8 | |
| 604 0, // 0xd9 | |
| 605 0, // 0xda | |
| 606 0, // 0xdb | |
| 607 0, // 0xdc | |
| 608 0, // 0xdd | |
| 609 0, // 0xde | |
| 610 0, // 0xdf | |
| 611 0, // 0xe0 | |
| 612 0, // 0xe1 | |
| 613 0, // 0xe2 | |
| 614 0, // 0xe3 | |
| 615 0, // 0xe4 | |
| 616 0, // 0xe5 | |
| 617 0, // 0xe6 | |
| 618 0, // 0xe7 | |
| 619 0, // 0xe8 | |
| 620 0, // 0xe9 | |
| 621 0, // 0xea | |
| 622 0, // 0xeb | |
| 623 0, // 0xec | |
| 624 0, // 0xed | |
| 625 0, // 0xee | |
| 626 0, // 0xef | |
| 627 0, // 0xf0 | |
| 628 0, // 0xf1 | |
| 629 0, // 0xf2 | |
| 630 0, // 0xf3 | |
| 631 0, // 0xf4 | |
| 632 0, // 0xf5 | |
| 633 0, // 0xf6 | |
| 634 0, // 0xf7 | |
| 635 0, // 0xf8 | |
| 636 0, // 0xf9 | |
| 637 0, // 0xfa | |
| 638 0, // 0xfb | |
| 639 0, // 0xfc | |
| 640 0, // 0xfd | |
| 641 0, // 0xfe | |
| 642 0, // 0xff | |
| 643 }; | |
| 7 | 644 |
| 6 | 645 u_char b64_decode[256] = { |
| 646 0, // 0x00 | |
| 647 0, // 0x01 | |
| 648 0, // 0x02 | |
| 649 0, // 0x03 | |
| 650 0, // 0x04 | |
| 651 0, // 0x05 | |
| 652 0, // 0x06 | |
| 653 0, // 0x07 | |
| 654 0, // 0x08 | |
| 7 | 655 0, // 0x09 <tab> |
| 656 0, // 0x0a <lf> | |
| 6 | 657 0, // 0x0b |
| 658 0, // 0x0c | |
| 7 | 659 0, // 0x0d <cr> |
| 6 | 660 0, // 0x0e |
| 661 0, // 0x0f | |
| 662 0, // 0x10 | |
| 663 0, // 0x11 xon char | |
| 664 0, // 0x12 | |
| 665 0, // 0x13 xoff char | |
| 666 0, // 0x14 | |
| 667 0, // 0x15 | |
| 668 0, // 0x16 | |
| 669 0, // 0x17 | |
| 670 0, // 0x18 | |
| 671 0, // 0x19 | |
| 672 0, // 0x1a | |
| 673 0, // 0x1b | |
| 674 0, // 0x1c | |
| 675 0, // 0x1d | |
| 676 0, // 0x1e | |
| 677 0, // 0x1f | |
| 678 0, // 0x20 space | |
| 679 0, // 0x21 ! | |
| 680 0, // 0x22 "" | |
| 681 0, // 0x23 # | |
| 682 0, // 0x24 $ | |
| 683 0, // 0x25 % | |
| 684 0, // 0x26 & | |
| 685 0, // 0x27 ' | |
| 686 0, // 0x28 ( | |
| 687 0, // 0x29 ) | |
| 688 0, // 0x2A * | |
| 689 62, // 0x2B + | |
| 690 0, // 0x2C , | |
| 691 0, // 0x2D - | |
| 692 0, // 0x2E . | |
| 693 63, // 0x2F / | |
| 694 52, // 0x30 0 | |
| 695 53, // 0x31 1 | |
| 696 54, // 0x32 2 | |
| 697 55, // 0x33 3 | |
| 698 56, // 0x34 4 | |
| 699 57, // 0x35 5 | |
| 700 58, // 0x36 6 | |
| 701 59, // 0x37 7 | |
| 702 60, // 0x38 8 | |
| 703 61, // 0x39 9 | |
| 704 0, // 0x3A : | |
| 705 0, // 0x3B ; | |
| 706 0, // 0x3C < | |
| 707 0, // 0x3D = | |
| 708 0, // 0x3E > | |
| 709 0, // 0x3F ? | |
| 710 0, // 0x40 @ | |
| 7 | 711 0, // 0x41 A |
| 712 1, // 0x42 B | |
| 713 2, // 0x43 C | |
| 714 3, // 0x44 D | |
| 715 4, // 0x45 E | |
| 716 5, // 0x46 F | |
| 717 6, // 0x47 G | |
| 718 7, // 0x48 H | |
| 719 8, // 0x49 I | |
| 720 9, // 0x4A J | |
| 6 | 721 10, // 0x4B K |
| 722 11, // 0x4C L | |
| 723 12, // 0x4D M | |
| 724 13, // 0x4E N | |
| 725 14, // 0x4F O | |
| 726 15, // 0x50 P | |
| 727 16, // 0x51 Q | |
| 728 17, // 0x52 R | |
| 729 18, // 0x53 S | |
| 730 19, // 0x54 T | |
| 731 20, // 0x55 U | |
| 732 21, // 0x56 V | |
| 733 22, // 0x57 W | |
| 734 23, // 0x58 X | |
| 735 24, // 0x59 Y | |
| 736 25, // 0x5A Z | |
| 737 0, // 0x5B [ | |
| 738 0, // 0x5C brace | |
| 739 0, // 0x5D ] | |
| 740 0, // 0x5E ^ | |
| 741 0, // 0x5F _ | |
| 742 0, // 0x60 ` | |
| 743 26, // 0x61 a | |
| 744 27, // 0x62 b | |
| 745 28, // 0x63 c | |
| 746 29, // 0x64 d | |
| 747 30, // 0x65 e | |
| 748 31, // 0x66 f | |
| 749 32, // 0x67 g | |
| 750 33, // 0x68 h | |
| 751 34, // 0x69 i | |
| 752 35, // 0x6A j | |
| 753 36, // 0x6B k | |
| 754 37, // 0x6C l | |
| 755 38, // 0x6D m | |
| 756 39, // 0x6E n | |
| 757 40, // 0x6F o | |
| 758 41, // 0x70 p | |
| 759 42, // 0x71 q | |
| 760 43, // 0x72 r | |
| 761 44, // 0x73 s | |
| 762 45, // 0x74 t | |
| 763 46, // 0x75 u | |
| 764 47, // 0x76 v | |
| 765 48, // 0x77 w | |
| 766 49, // 0x78 x | |
| 767 50, // 0x79 y | |
| 768 51, // 0x7A z | |
| 769 0, // 0x7B { | |
| 770 0, // 0x7C | | |
| 771 0, // 0x7D } | |
| 772 0, // 0x7E ~ | |
| 773 0, // 0x7f | |
| 774 0, // 0x80 | |
| 775 0, // 0x81 | |
| 776 0, // 0x82 | |
| 777 0, // 0x83 | |
| 778 0, // 0x84 | |
| 779 0, // 0x85 | |
| 780 0, // 0x86 | |
| 781 0, // 0x87 | |
| 782 0, // 0x88 | |
| 783 0, // 0x89 | |
| 784 0, // 0x8a | |
| 785 0, // 0x8b | |
| 786 0, // 0x8c | |
| 787 0, // 0x8d | |
| 788 0, // 0x8e | |
| 789 0, // 0x8f | |
| 790 0, // 0x90 | |
| 791 0, // 0x91 | |
| 792 0, // 0x92 | |
| 793 0, // 0x93 | |
| 794 0, // 0x94 | |
| 795 0, // 0x95 | |
| 796 0, // 0x96 | |
| 797 0, // 0x97 | |
| 798 0, // 0x98 | |
| 799 0, // 0x99 | |
| 800 0, // 0x9a | |
| 801 0, // 0x9b | |
| 802 0, // 0x9c | |
| 803 0, // 0x9d | |
| 804 0, // 0x9e | |
| 805 0, // 0x9f | |
| 806 0, // 0xa0 | |
| 807 0, // 0xa1 | |
| 808 0, // 0xa2 | |
| 809 0, // 0xa3 | |
| 810 0, // 0xa4 | |
| 811 0, // 0xa5 | |
| 812 0, // 0xa6 | |
| 813 0, // 0xa7 | |
| 814 0, // 0xa8 | |
| 815 0, // 0xa9 | |
| 816 0, // 0xaa | |
| 817 0, // 0xab | |
| 818 0, // 0xac | |
| 819 0, // 0xad | |
| 820 0, // 0xae | |
| 821 0, // 0xaf | |
| 822 0, // 0xb0 | |
| 823 0, // 0xb1 | |
| 824 0, // 0xb2 | |
| 825 0, // 0xb3 | |
| 826 0, // 0xb4 | |
| 827 0, // 0xb5 | |
| 828 0, // 0xb6 | |
| 829 0, // 0xb7 | |
| 830 0, // 0xb8 | |
| 831 0, // 0xb9 | |
| 832 0, // 0xba | |
| 833 0, // 0xbb | |
| 834 0, // 0xbc | |
| 835 0, // 0xbd | |
| 836 0, // 0xbe | |
| 837 0, // 0xbf | |
| 838 0, // 0xc0 | |
| 839 0, // 0xc1 | |
| 840 0, // 0xc2 | |
| 841 0, // 0xc3 | |
| 842 0, // 0xc4 | |
| 843 0, // 0xc5 | |
| 844 0, // 0xc6 | |
| 845 0, // 0xc7 | |
| 846 0, // 0xc8 | |
| 847 0, // 0xc9 | |
| 848 0, // 0xca | |
| 849 0, // 0xcb | |
| 850 0, // 0xcc | |
| 851 0, // 0xcd | |
| 852 0, // 0xce | |
| 853 0, // 0xcf | |
| 854 0, // 0xd0 | |
| 855 0, // 0xd1 | |
| 856 0, // 0xd2 | |
| 857 0, // 0xd3 | |
| 858 0, // 0xd4 | |
| 859 0, // 0xd5 | |
| 860 0, // 0xd6 | |
| 861 0, // 0xd7 | |
| 862 0, // 0xd8 | |
| 863 0, // 0xd9 | |
| 864 0, // 0xda | |
| 865 0, // 0xdb | |
| 866 0, // 0xdc | |
| 867 0, // 0xdd | |
| 868 0, // 0xde | |
| 869 0, // 0xdf | |
| 870 0, // 0xe0 | |
| 871 0, // 0xe1 | |
| 872 0, // 0xe2 | |
| 873 0, // 0xe3 | |
| 874 0, // 0xe4 | |
| 875 0, // 0xe5 | |
| 876 0, // 0xe6 | |
| 877 0, // 0xe7 | |
| 878 0, // 0xe8 | |
| 879 0, // 0xe9 | |
| 880 0, // 0xea | |
| 881 0, // 0xeb | |
| 882 0, // 0xec | |
| 883 0, // 0xed | |
| 884 0, // 0xee | |
| 885 0, // 0xef | |
| 886 0, // 0xf0 | |
| 887 0, // 0xf1 | |
| 888 0, // 0xf2 | |
| 889 0, // 0xf3 | |
| 890 0, // 0xf4 | |
| 891 0, // 0xf5 | |
| 892 0, // 0xf6 | |
| 893 0, // 0xf7 | |
| 894 0, // 0xf8 | |
| 895 0, // 0xf9 | |
| 896 0, // 0xfa | |
| 897 0, // 0xfb | |
| 898 0, // 0xfc | |
| 899 0, // 0xfd | |
| 900 0, // 0xfe | |
| 901 0, // 0xff | |
| 902 }; | |
| 903 | |
| 8 | 904 #define PENDING_LIMIT 100 |
| 6 | 905 struct fsa { |
| 12 | 906 u_char pending[PENDING_LIMIT]; |
| 907 int count; | |
| 908 state st; | |
| 909 state init; | |
| 16 | 910 fsa* next1; |
| 911 fsa* next2; | |
| 24 | 912 recorder *memory; |
| 6 | 913 |
| 24 | 914 fsa(state init, fsa* next1_, fsa* next2_, recorder *memory_); |
| 6 | 915 void push(u_char *buf, int len); |
| 16 | 916 void pusher(); |
|
19
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
917 void error(char *err); |
| 6 | 918 }; |
| 919 | |
| 24 | 920 fsa::fsa(state init_, fsa *next1_, fsa *next2_, recorder *memory_) { |
| 921 count = 0; | |
| 922 st = init_; | |
| 923 init = init_; | |
| 924 next1 = next1_; | |
| 925 next2 = next2_; | |
| 926 memory = memory_; | |
| 6 | 927 } |
| 928 | |
|
19
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
929 void fsa::error(char *err) { |
|
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
930 count = 0; |
|
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
931 st = init; |
|
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
932 if (err) my_syslog(err); |
|
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
933 } |
|
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
934 |
| 16 | 935 void fsa::pusher() { |
| 936 if (next1) next1->push(pending, count); | |
| 937 if (next2) next2->push(pending, count); | |
| 938 count = 0; | |
| 939 } | |
| 940 | |
| 6 | 941 void fsa::push(u_char *buf, int len) { |
| 942 for (int i=0; i<len; i++) { | |
|
19
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
943 if (count == (PENDING_LIMIT-1)) error(NULL); |
|
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
944 if (st >= end_state) error("finite state machine impossible state"); |
| 6 | 945 u_char c = buf[i]; |
| 946 pending[count++] = c; | |
| 947 st = parse_table[c][st]; | |
| 948 switch (st) { | |
| 949 | |
| 950 ////////////////////////////// | |
| 16 | 951 // host name recognizer |
| 952 case h_end: { | |
| 18 | 953 pending[--count] = '\0'; // null terminate host name by overwriting the terminator |
|
27
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
954 if (!strchr((const char *)pending, '@')) { |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
955 // not an email address or message id |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
956 char *tld; |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
957 for (int i=0; (tld = tlds[i]); i++) { |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
958 int n = strlen(tld); |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
959 if (count > n) { |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
960 if (strncasecmp((const char *)(pending+count-n), tld, n) == 0) { |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
961 memory->new_url((char*)pending); |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
962 break; |
|
43a4f6b3e668
add configurable host name limit and bad html tag limits.
carl
parents:
25
diff
changeset
|
963 } |
| 16 | 964 } |
| 965 } | |
| 966 } | |
|
19
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
967 st = h_init; |
| 16 | 968 } // fall thru |
| 969 | |
| 970 case h_init: { | |
| 18 | 971 count = 0; |
| 16 | 972 } break; |
| 973 | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
974 |
| 16 | 975 ////////////////////////////// |
| 18 | 976 // html tag discarder |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
977 case t_bin: { |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
978 memory->binary(); |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
979 st = t_disc; |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
980 count = 0; // discard all characters |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
981 } break; |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
982 |
| 18 | 983 case t_end: { |
| 24 | 984 if (count > 2) { |
| 985 pending[--count] = '\0'; // null terminate html tag | |
| 986 for (int i=1; i<count; i++) pending[i] = tolower(pending[i]); | |
| 987 memory->new_tag((char*)pending+1); | |
| 988 } | |
|
25
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
989 st = (c == ' ') ? t_disc : t_init; |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
990 } // fall thru |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
991 |
|
6176e7b2e8af
better scanning for bad html tags, allow binary zip and gz files with random char sequences
carl
parents:
24
diff
changeset
|
992 case t_disc: { |
| 18 | 993 count = 0; // discard all characters |
| 16 | 994 } break; |
| 995 | |
| 996 case t_init: { | |
| 997 pusher(); | |
| 998 } break; | |
| 999 | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1000 |
| 16 | 1001 ////////////////////////////// |
| 7 | 1002 // url recognizer |
| 1003 case u_reco: { | |
|
19
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
1004 if (count > 13) { // need some minimal length host name after the protocol |
|
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
1005 pending[--count] = '\0'; // null terminate host name by overwriting the terminator |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1006 char *p = strrchr((const char *)pending, '/'); |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1007 if (p && // have a leading / |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1008 strchr(p, '.') && // require at least one . in a dns name |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1009 (strncasecmp((const char *)pending, "http", 4) == 0)) { // must start with protocol |
| 24 | 1010 memory->new_url(++p); // we seem to have a host name, skip the last / |
| 9 | 1011 } |
| 8 | 1012 } |
|
19
b8f5fa3dd5b8
fix problems in the state transitions causing impossible states
carl
parents:
18
diff
changeset
|
1013 st = u_init; |
| 16 | 1014 } // fall thru |
| 7 | 1015 |
| 1016 case u_init: { | |
| 1017 count = 0; // discard all characters | |
| 1018 } break; | |
| 1019 | |
| 1020 | |
| 1021 ////////////////////////////// | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1022 // url decoder %xx |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1023 case d_2: { |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1024 pending[0] = hex_decode[pending[1]] * 16 + hex_decode[pending[2]]; |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1025 count = 1; |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1026 st = d_init; |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1027 } // fall thru |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1028 |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1029 case d_init: { |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1030 pusher(); |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1031 } break; |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1032 |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1033 |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1034 ////////////////////////////// |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1035 // html entity decoder &#nnn; |
| 6 | 1036 case e_semi: { |
| 1037 pending[--count] = '\0'; // null terminate the digit string by overwriting the semicolon | |
| 1038 pending[0] = atoi((const char *)pending+2); | |
| 1039 count = 1; | |
| 1040 st = e_init; | |
| 1041 } // fall thru | |
| 1042 | |
| 1043 case e_init: { | |
| 16 | 1044 pusher(); |
| 1045 } break; | |
| 1046 | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1047 |
| 16 | 1048 ////////////////////////////// |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1049 // mime decoder =xx |
| 16 | 1050 case m_2: { |
| 1051 pending[0] = hex_decode[pending[1]] * 16 + hex_decode[pending[2]]; | |
| 1052 count = 1; | |
| 1053 st = m_init; | |
| 1054 } // fall thru | |
| 1055 | |
| 1056 case m_init: { | |
| 1057 pusher(); | |
| 1058 } break; | |
| 1059 | |
| 1060 case m_cr: { | |
| 1061 count = 1; | |
| 1062 st = m_eq; | |
| 1063 } break; | |
| 1064 | |
| 1065 case m_nl: { | |
| 6 | 1066 count = 0; |
| 16 | 1067 st = m_init; |
| 6 | 1068 } break; |
| 1069 | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1070 |
| 6 | 1071 ////////////////////////////// |
| 1072 // base64 decoder | |
| 1073 case b_lf2: { | |
| 1074 count--; | |
| 1075 } break; | |
| 1076 | |
| 1077 case b_cr: { | |
| 1078 int cnt = 0; | |
| 1079 if ((count % 4) == 1) { | |
| 1080 count--; | |
| 1081 // might have proper b64 data | |
| 1082 for (int i=0; i<count; i+=4) { | |
| 1083 unsigned long a1 = b64_decode[pending[i]]; | |
| 1084 unsigned long a2 = b64_decode[pending[i+1]]; | |
| 1085 unsigned long a3 = b64_decode[pending[i+2]]; | |
| 1086 unsigned long a4 = b64_decode[pending[i+3]]; | |
| 1087 unsigned long a = (a1 << 18) | (a2 << 12) | (a3 << 6) | a4; | |
| 1088 pending[cnt++] = (a & 0x00ff0000) >> 16; | |
| 1089 pending[cnt++] = (a & 0x0000ff00) >> 8; | |
| 1090 pending[cnt++] = (a & 0x000000ff); | |
| 1091 if ((char)pending[i+3] == '=') cnt--; | |
| 1092 if ((char)pending[i+2] == '=') cnt--; | |
| 1093 } | |
| 1094 count = cnt; | |
| 1095 st = b_lf2; | |
| 1096 } | |
| 1097 else st = b_init; | |
| 1098 } // fall thru | |
| 1099 | |
| 1100 case b_lf: | |
| 1101 case b_init: { | |
| 16 | 1102 pusher(); |
| 6 | 1103 } break; |
| 1104 | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1105 |
| 6 | 1106 ////////////////////////////// |
| 1107 // states that just accumulate characters in the pending buffer | |
| 16 | 1108 case h_host: |
| 24 | 1109 case t_tag: |
| 7 | 1110 case u_http: |
| 1111 case u_url: | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1112 case u_sla: |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1113 case d_pcnt: |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1114 case d_1: |
| 6 | 1115 case e_amp: |
| 1116 case e_num: | |
| 1117 case m_eq: | |
| 1118 case m_1: | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1119 case b_64: |
| 6 | 1120 default: { |
| 1121 } break; | |
| 1122 } | |
| 1123 } | |
| 1124 } | |
| 1125 | |
| 8 | 1126 struct url_scanner { |
| 16 | 1127 fsa *host_parser; |
| 1128 fsa *tags_parser; | |
| 8 | 1129 fsa *urls_parser; |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1130 fsa *urld_parser; |
| 8 | 1131 fsa *html_parser; |
| 1132 fsa *mime_parser; | |
| 1133 fsa *b64_parser; | |
| 6 | 1134 |
| 24 | 1135 url_scanner(recorder *memory); |
| 8 | 1136 ~url_scanner(); |
| 1137 void scan(u_char *buffer, size_t length); | |
| 1138 }; | |
| 1139 | |
| 24 | 1140 url_scanner::url_scanner(recorder *memory) { |
| 1141 host_parser = new fsa(h_init, NULL, NULL, memory); | |
| 1142 tags_parser = new fsa(t_init, host_parser, NULL, memory); | |
| 1143 urls_parser = new fsa(u_init, NULL, NULL, memory); | |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1144 urld_parser = new fsa(d_init, urls_parser, tags_parser, NULL); |
|
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1145 html_parser = new fsa(e_init, urld_parser, NULL, NULL); |
| 16 | 1146 mime_parser = new fsa(m_init, html_parser, NULL, NULL); |
| 1147 b64_parser = new fsa(b_init, mime_parser, NULL, NULL); | |
| 6 | 1148 } |
| 1149 | |
| 8 | 1150 url_scanner::~url_scanner() { |
| 16 | 1151 delete host_parser; |
| 1152 delete tags_parser; | |
| 8 | 1153 delete urls_parser; |
|
23
06de5ab6a232
add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames
carl
parents:
19
diff
changeset
|
1154 delete urld_parser; |
| 8 | 1155 delete html_parser; |
| 1156 delete mime_parser; | |
| 1157 delete b64_parser; | |
| 6 | 1158 } |
| 1159 | |
| 8 | 1160 void url_scanner::scan(u_char *buffer, size_t length) { |
| 1161 b64_parser->push(buffer, length); | |
| 6 | 1162 } |
| 1163 |