dnsbl: src/scanner.cpp annotate

annotate src/scanner.cpp @ 58:7bb8bbf79285

changes to handle 5iantlavalamp.com

author	carl
date	Thu, 28 Oct 2004 22:54:34 -0700
parents	a84752107aca
children	a39c813e8f7b

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 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	9 static char* scanner_version="$Id$";
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	10
cea50d98a6cf start work on content url scanner carl parents: diff changeset	11 using namespace std;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	12
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	13
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	14 // object to record things we see in the body content
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	15 struct recorder
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	16 {
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	17 mlfiPriv *priv; // needed for syslog
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	18 string_set *html_tags; // valid tags
28 33e1e3910506 add configurable list of tlds carl parents: 27 diff changeset	19 string_set *tlds; // valid tlds
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	20 string_set hosts;
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	21 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	22 int binary_tags;
42 afcf403709ef updates for 3.2, try to drop root privileges carl parents: 41 diff changeset	23 recorder(mlfiPriv priv_, string_set html_tags_, string_set *tlds_);
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	24 ~recorder();
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	25 void empty();
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	26 void new_url(char *host);
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	27 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	28 void binary();
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	29 };
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	30 recorder::recorder(mlfiPriv priv_, string_set html_tags_, string_set *tlds_) {
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	31 priv = priv_;
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	32 html_tags = html_tags_;
28 33e1e3910506 add configurable list of tlds carl parents: 27 diff changeset	33 tlds = tlds_;
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	34 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	35 binary_tags = 0;
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	36 }
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	37 recorder::~recorder() {
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	38 empty();
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	39 }
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	40 void recorder::empty() {
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	41 bad_html_tags = 0;
28 33e1e3910506 add configurable list of tlds carl parents: 27 diff changeset	42 binary_tags = 0;
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	43 discard(hosts);
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	44 }
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	45 void recorder::new_url(char *host) {
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	46 register_string(hosts, host);
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	47 }
25 6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	48 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	49 binary_tags++;
6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	50 }
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	51 void recorder::new_tag(char *tag) {
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	52 string_set::iterator i = html_tags->find(tag);
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	53 if (i == html_tags->end()) {
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	54 bad_html_tags++;
50 05957b18b2e6 final changes for 3.3 carl parents: 44 diff changeset	55 if (debug_syslog && (bad_html_tags < 10)) {
25 6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	56 // only log the first 10 bad tags
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	57 char buf[200];
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	58 snprintf(buf, sizeof(buf), "bad html tag %s", tag);
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	59 my_syslog(priv, buf);
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	60 }
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	61 }
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	62 }
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	63
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	64
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	65
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	66 enum state {// host name recognizer states
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	67 h_init,
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	68 h_host,
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	69
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	70 // html tag discarder states
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	71 t_init,
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	72 t_tag1, // seen opening <
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	73 t_tag2, // not comment
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	74 t_com1, // seen !
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	75 t_com2, // seen first -
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	76 t_com3, // seen second -, looking for -->
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	77 t_com4, // seen first -
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	78 t_com5, // seen second -
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	79 t_disc, // looking for closing >
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	80
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	81 // url recognizer states
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	82 u_init,
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	83 u_http,
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	84 u_sla,
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	85 u_url,
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	86
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	87 // url decoder states %xx
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	88 d_init,
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	89 d_pcnt,
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	90 d_1,
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	91
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	92 // html entity decoder states &#nnn;
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	93 e_init,
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	94 e_amp,
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	95 e_num,
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	96
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	97 // mime decoder states =xx
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	98 m_init,
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	99 m_eq,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	100 m_1,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	101
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	102 // base64 decoder states
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	103 b_init,
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	104 b_lf,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	105 b_lf2,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	106 b_64,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	107
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	108 // counter for number of columns in the table
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	109 end_state,
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	110
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	111 // temporary states
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	112 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	113 t_bin,
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	114 t_end,
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	115 u_reco,
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	116 d_2,
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	117 e_semi,
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	118 m_2,
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	119 m_cr,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	120 m_nl,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	121 b_cr,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	122 };
cea50d98a6cf start work on content url scanner carl parents: diff changeset	123
cea50d98a6cf start work on content url scanner carl parents: diff changeset	124 typedef state PARSE[end_state];
cea50d98a6cf start work on content url scanner carl parents: diff changeset	125
cea50d98a6cf start work on content url scanner carl parents: diff changeset	126 static PARSE parse_table[256] = {
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	127 // h_init, h_host, t_init, t_tag1, t_tag2, t_com1, t_com2, t_com3, t_com4, t_com5, t_disc, u_init, u_http, u_sla , u_url, d_init, d_pcnt, d_1, e_init, e_amp, e_num, m_init, m_eq, m_1, b_init, b_lf, b_lf2, b_64
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	128
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	129 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	130 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	131 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	132 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	133 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	134 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	135 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	136 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	137 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	138 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com5, 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>
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	139 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com5, 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>
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	140 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	141 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	142 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com5, 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>
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	143 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	144 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	145 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	146 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	147 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	148 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	149 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	150 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	151 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	152 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	153 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	154 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	155 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	156 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	157 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	158 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	159 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	160 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	161 {h_init, h_end, t_init, t_end, t_end, t_end, t_end, t_com3, t_com3, t_com5, 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
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	162 {h_init, h_end, t_init, t_com1, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 !
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	163 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 ""
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	164 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 #
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	165 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 $
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	166 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 %
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	167 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 &
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	168 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 '
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	169 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 (
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	170 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 )
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	171 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 *
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	172 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 +
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	173 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 ,
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	174 {h_host, h_host, t_init, t_disc, t_disc, t_com2, t_com3, t_com4, t_com5, t_com3, 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 -
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	175 {h_host, h_host, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 .
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	176 {h_init, h_end, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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 /
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	177 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	178 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	179 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	180 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	181 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	182 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	183 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	184 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	185 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	186 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	187 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 :
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	188 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 ;
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	189 {h_init, h_end, t_tag1, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 <
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	190 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 =
37 ccf2c2651c1d add -pthread compiler option, fix error in scanning for end of html comment carl parents: 33 diff changeset	191 {h_init, h_end, t_init, t_end, t_end, t_end, t_end, t_com3, t_com3, 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 >
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	192 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 ?
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	193 {h_init, h_host, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 @
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	194 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x41 A
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	195 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x42 B
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	196 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x43 C
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	197 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x44 D
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	198 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x45 E
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	199 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x46 F
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	200 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	201 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	202 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	203 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	204 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	205 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	206 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	207 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	208 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	209 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	210 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	211 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	212 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	213 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	214 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	215 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	216 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	217 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_num, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x58 X
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	218 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	219 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	220 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 [
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	221 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	222 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 ]
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	223 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 ^
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	224 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 _
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	225 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 `
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	226 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x61 a
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	227 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x62 b
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	228 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x63 c
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	229 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x64 d
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	230 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x65 e
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	231 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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, }, // 0x66 f
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	232 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	233 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	234 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	235 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	236 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	237 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	238 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	239 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	240 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	241 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	242 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	243 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	244 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	245 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	246 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	247 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	248 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	249 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, t_disc, u_init, u_init, u_url, u_url, d_init, d_init, d_init, e_init, e_init, e_num, m_init, m_init, m_init, b_init, b_64, b_64, b_64, }, // 0x78 x
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	250 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	251 {h_host, h_host, t_init, t_tag2, t_tag2, t_tag2, t_tag2, t_com3, t_com3, t_com3, 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
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	252 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 {
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	253 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 \|
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	254 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 }
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	255 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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 ~
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	256 {h_init, h_end, t_init, t_disc, t_disc, t_disc, t_disc, t_com3, t_com3, t_com3, 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
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	257 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	258 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	259 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	260 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	261 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	262 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	263 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	264 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	265 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	266 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	267 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	268 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	269 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	270 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	271 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	272 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	273 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	274 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	275 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	276 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	277 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	278 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	279 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	280 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	281 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	282 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	283 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	284 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	285 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	286 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	287 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	288 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	289 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	290 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	291 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	292 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	293 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	294 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	295 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	296 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	297 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	298 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	299 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	300 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	301 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	302 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	303 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	304 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	305 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	306 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	307 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	308 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	309 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	310 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	311 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	312 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	313 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	314 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	315 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	316 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	317 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	318 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	319 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	320 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	321 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	322 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	323 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	324 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	325 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	326 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	327 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	328 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	329 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	330 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	331 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	332 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	333 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	334 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	335 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	336 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	337 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	338 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	339 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	340 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	341 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	342 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	343 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	344 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	345 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	346 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	347 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	348 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	349 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	350 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	351 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	352 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	353 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	354 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	355 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	356 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	357 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	358 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	359 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	360 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	361 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	362 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	363 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	364 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	365 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	366 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	367 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	368 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	369 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	370 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	371 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	372 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	373 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	374 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	375 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	376 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	377 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	378 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	379 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	380 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	381 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	382 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	383 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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
d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	384 {h_init, h_end, t_init, t_bin, t_bin, t_bin, t_bin, t_com3, t_com3, t_com3, 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 cea50d98a6cf start work on content url scanner carl parents: diff changeset	385 };
cea50d98a6cf start work on content url scanner carl parents: diff changeset	386
cea50d98a6cf start work on content url scanner carl parents: diff changeset	387
cea50d98a6cf start work on content url scanner carl parents: diff changeset	388 u_char hex_decode[256] = {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	389 0, // 0x00
cea50d98a6cf start work on content url scanner carl parents: diff changeset	390 0, // 0x01
cea50d98a6cf start work on content url scanner carl parents: diff changeset	391 0, // 0x02
cea50d98a6cf start work on content url scanner carl parents: diff changeset	392 0, // 0x03
cea50d98a6cf start work on content url scanner carl parents: diff changeset	393 0, // 0x04
cea50d98a6cf start work on content url scanner carl parents: diff changeset	394 0, // 0x05
cea50d98a6cf start work on content url scanner carl parents: diff changeset	395 0, // 0x06
cea50d98a6cf start work on content url scanner carl parents: diff changeset	396 0, // 0x07
cea50d98a6cf start work on content url scanner carl parents: diff changeset	397 0, // 0x08
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	398 0, // 0x09 <tab>
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	399 0, // 0x0a <lf>
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	400 0, // 0x0b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	401 0, // 0x0c
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	402 0, // 0x0d <cr>
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	403 0, // 0x0e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	404 0, // 0x0f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	405 0, // 0x10
cea50d98a6cf start work on content url scanner carl parents: diff changeset	406 0, // 0x11 xon char
cea50d98a6cf start work on content url scanner carl parents: diff changeset	407 0, // 0x12
cea50d98a6cf start work on content url scanner carl parents: diff changeset	408 0, // 0x13 xoff char
cea50d98a6cf start work on content url scanner carl parents: diff changeset	409 0, // 0x14
cea50d98a6cf start work on content url scanner carl parents: diff changeset	410 0, // 0x15
cea50d98a6cf start work on content url scanner carl parents: diff changeset	411 0, // 0x16
cea50d98a6cf start work on content url scanner carl parents: diff changeset	412 0, // 0x17
cea50d98a6cf start work on content url scanner carl parents: diff changeset	413 0, // 0x18
cea50d98a6cf start work on content url scanner carl parents: diff changeset	414 0, // 0x19
cea50d98a6cf start work on content url scanner carl parents: diff changeset	415 0, // 0x1a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	416 0, // 0x1b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	417 0, // 0x1c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	418 0, // 0x1d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	419 0, // 0x1e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	420 0, // 0x1f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	421 0, // 0x20 space
cea50d98a6cf start work on content url scanner carl parents: diff changeset	422 0, // 0x21 !
cea50d98a6cf start work on content url scanner carl parents: diff changeset	423 0, // 0x22 ""
cea50d98a6cf start work on content url scanner carl parents: diff changeset	424 0, // 0x23 #
cea50d98a6cf start work on content url scanner carl parents: diff changeset	425 0, // 0x24 $
cea50d98a6cf start work on content url scanner carl parents: diff changeset	426 0, // 0x25 %
cea50d98a6cf start work on content url scanner carl parents: diff changeset	427 0, // 0x26 &
cea50d98a6cf start work on content url scanner carl parents: diff changeset	428 0, // 0x27 '
cea50d98a6cf start work on content url scanner carl parents: diff changeset	429 0, // 0x28 (
cea50d98a6cf start work on content url scanner carl parents: diff changeset	430 0, // 0x29 )
cea50d98a6cf start work on content url scanner carl parents: diff changeset	431 0, // 0x2A *
cea50d98a6cf start work on content url scanner carl parents: diff changeset	432 0, // 0x2B +
cea50d98a6cf start work on content url scanner carl parents: diff changeset	433 0, // 0x2C ,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	434 0, // 0x2D -
cea50d98a6cf start work on content url scanner carl parents: diff changeset	435 0, // 0x2E .
cea50d98a6cf start work on content url scanner carl parents: diff changeset	436 0, // 0x2F /
cea50d98a6cf start work on content url scanner carl parents: diff changeset	437 0, // 0x30 0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	438 1, // 0x31 1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	439 2, // 0x32 2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	440 3, // 0x33 3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	441 4, // 0x34 4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	442 5, // 0x35 5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	443 6, // 0x36 6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	444 7, // 0x37 7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	445 8, // 0x38 8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	446 9, // 0x39 9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	447 0, // 0x3A :
cea50d98a6cf start work on content url scanner carl parents: diff changeset	448 0, // 0x3B ;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	449 0, // 0x3C <
cea50d98a6cf start work on content url scanner carl parents: diff changeset	450 0, // 0x3D =
cea50d98a6cf start work on content url scanner carl parents: diff changeset	451 0, // 0x3E >
cea50d98a6cf start work on content url scanner carl parents: diff changeset	452 0, // 0x3F ?
cea50d98a6cf start work on content url scanner carl parents: diff changeset	453 0, // 0x40 @
cea50d98a6cf start work on content url scanner carl parents: diff changeset	454 10, // 0x41 A
cea50d98a6cf start work on content url scanner carl parents: diff changeset	455 11, // 0x42 B
cea50d98a6cf start work on content url scanner carl parents: diff changeset	456 12, // 0x43 C
cea50d98a6cf start work on content url scanner carl parents: diff changeset	457 13, // 0x44 D
cea50d98a6cf start work on content url scanner carl parents: diff changeset	458 14, // 0x45 E
cea50d98a6cf start work on content url scanner carl parents: diff changeset	459 15, // 0x46 F
cea50d98a6cf start work on content url scanner carl parents: diff changeset	460 0, // 0x47 G
cea50d98a6cf start work on content url scanner carl parents: diff changeset	461 0, // 0x48 H
cea50d98a6cf start work on content url scanner carl parents: diff changeset	462 0, // 0x49 I
cea50d98a6cf start work on content url scanner carl parents: diff changeset	463 0, // 0x4A J
cea50d98a6cf start work on content url scanner carl parents: diff changeset	464 0, // 0x4B K
cea50d98a6cf start work on content url scanner carl parents: diff changeset	465 0, // 0x4C L
cea50d98a6cf start work on content url scanner carl parents: diff changeset	466 0, // 0x4D M
cea50d98a6cf start work on content url scanner carl parents: diff changeset	467 0, // 0x4E N
cea50d98a6cf start work on content url scanner carl parents: diff changeset	468 0, // 0x4F O
cea50d98a6cf start work on content url scanner carl parents: diff changeset	469 0, // 0x50 P
cea50d98a6cf start work on content url scanner carl parents: diff changeset	470 0, // 0x51 Q
cea50d98a6cf start work on content url scanner carl parents: diff changeset	471 0, // 0x52 R
cea50d98a6cf start work on content url scanner carl parents: diff changeset	472 0, // 0x53 S
cea50d98a6cf start work on content url scanner carl parents: diff changeset	473 0, // 0x54 T
cea50d98a6cf start work on content url scanner carl parents: diff changeset	474 0, // 0x55 U
cea50d98a6cf start work on content url scanner carl parents: diff changeset	475 0, // 0x56 V
cea50d98a6cf start work on content url scanner carl parents: diff changeset	476 0, // 0x57 W
cea50d98a6cf start work on content url scanner carl parents: diff changeset	477 0, // 0x58 X
cea50d98a6cf start work on content url scanner carl parents: diff changeset	478 0, // 0x59 Y
cea50d98a6cf start work on content url scanner carl parents: diff changeset	479 0, // 0x5A Z
cea50d98a6cf start work on content url scanner carl parents: diff changeset	480 0, // 0x5B [
cea50d98a6cf start work on content url scanner carl parents: diff changeset	481 0, // 0x5C brace
cea50d98a6cf start work on content url scanner carl parents: diff changeset	482 0, // 0x5D ]
cea50d98a6cf start work on content url scanner carl parents: diff changeset	483 0, // 0x5E ^
cea50d98a6cf start work on content url scanner carl parents: diff changeset	484 0, // 0x5F _
cea50d98a6cf start work on content url scanner carl parents: diff changeset	485 0, // 0x60 `
cea50d98a6cf start work on content url scanner carl parents: diff changeset	486 10, // 0x61 a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	487 11, // 0x62 b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	488 12, // 0x63 c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	489 13, // 0x64 d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	490 14, // 0x65 e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	491 15, // 0x66 f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	492 0, // 0x67 g
cea50d98a6cf start work on content url scanner carl parents: diff changeset	493 0, // 0x68 h
cea50d98a6cf start work on content url scanner carl parents: diff changeset	494 0, // 0x69 i
cea50d98a6cf start work on content url scanner carl parents: diff changeset	495 0, // 0x6A j
cea50d98a6cf start work on content url scanner carl parents: diff changeset	496 0, // 0x6B k
cea50d98a6cf start work on content url scanner carl parents: diff changeset	497 0, // 0x6C l
cea50d98a6cf start work on content url scanner carl parents: diff changeset	498 0, // 0x6D m
cea50d98a6cf start work on content url scanner carl parents: diff changeset	499 0, // 0x6E n
cea50d98a6cf start work on content url scanner carl parents: diff changeset	500 0, // 0x6F o
cea50d98a6cf start work on content url scanner carl parents: diff changeset	501 0, // 0x70 p
cea50d98a6cf start work on content url scanner carl parents: diff changeset	502 0, // 0x71 q
cea50d98a6cf start work on content url scanner carl parents: diff changeset	503 0, // 0x72 r
cea50d98a6cf start work on content url scanner carl parents: diff changeset	504 0, // 0x73 s
cea50d98a6cf start work on content url scanner carl parents: diff changeset	505 0, // 0x74 t
cea50d98a6cf start work on content url scanner carl parents: diff changeset	506 0, // 0x75 u
cea50d98a6cf start work on content url scanner carl parents: diff changeset	507 0, // 0x76 v
cea50d98a6cf start work on content url scanner carl parents: diff changeset	508 0, // 0x77 w
cea50d98a6cf start work on content url scanner carl parents: diff changeset	509 0, // 0x78 x
cea50d98a6cf start work on content url scanner carl parents: diff changeset	510 0, // 0x79 y
cea50d98a6cf start work on content url scanner carl parents: diff changeset	511 0, // 0x7A z
cea50d98a6cf start work on content url scanner carl parents: diff changeset	512 0, // 0x7B {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	513 0, // 0x7C \|
cea50d98a6cf start work on content url scanner carl parents: diff changeset	514 0, // 0x7D }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	515 0, // 0x7E ~
cea50d98a6cf start work on content url scanner carl parents: diff changeset	516 0, // 0x7f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	517 0, // 0x80
cea50d98a6cf start work on content url scanner carl parents: diff changeset	518 0, // 0x81
cea50d98a6cf start work on content url scanner carl parents: diff changeset	519 0, // 0x82
cea50d98a6cf start work on content url scanner carl parents: diff changeset	520 0, // 0x83
cea50d98a6cf start work on content url scanner carl parents: diff changeset	521 0, // 0x84
cea50d98a6cf start work on content url scanner carl parents: diff changeset	522 0, // 0x85
cea50d98a6cf start work on content url scanner carl parents: diff changeset	523 0, // 0x86
cea50d98a6cf start work on content url scanner carl parents: diff changeset	524 0, // 0x87
cea50d98a6cf start work on content url scanner carl parents: diff changeset	525 0, // 0x88
cea50d98a6cf start work on content url scanner carl parents: diff changeset	526 0, // 0x89
cea50d98a6cf start work on content url scanner carl parents: diff changeset	527 0, // 0x8a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	528 0, // 0x8b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	529 0, // 0x8c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	530 0, // 0x8d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	531 0, // 0x8e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	532 0, // 0x8f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	533 0, // 0x90
cea50d98a6cf start work on content url scanner carl parents: diff changeset	534 0, // 0x91
cea50d98a6cf start work on content url scanner carl parents: diff changeset	535 0, // 0x92
cea50d98a6cf start work on content url scanner carl parents: diff changeset	536 0, // 0x93
cea50d98a6cf start work on content url scanner carl parents: diff changeset	537 0, // 0x94
cea50d98a6cf start work on content url scanner carl parents: diff changeset	538 0, // 0x95
cea50d98a6cf start work on content url scanner carl parents: diff changeset	539 0, // 0x96
cea50d98a6cf start work on content url scanner carl parents: diff changeset	540 0, // 0x97
cea50d98a6cf start work on content url scanner carl parents: diff changeset	541 0, // 0x98
cea50d98a6cf start work on content url scanner carl parents: diff changeset	542 0, // 0x99
cea50d98a6cf start work on content url scanner carl parents: diff changeset	543 0, // 0x9a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	544 0, // 0x9b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	545 0, // 0x9c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	546 0, // 0x9d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	547 0, // 0x9e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	548 0, // 0x9f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	549 0, // 0xa0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	550 0, // 0xa1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	551 0, // 0xa2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	552 0, // 0xa3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	553 0, // 0xa4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	554 0, // 0xa5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	555 0, // 0xa6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	556 0, // 0xa7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	557 0, // 0xa8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	558 0, // 0xa9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	559 0, // 0xaa
cea50d98a6cf start work on content url scanner carl parents: diff changeset	560 0, // 0xab
cea50d98a6cf start work on content url scanner carl parents: diff changeset	561 0, // 0xac
cea50d98a6cf start work on content url scanner carl parents: diff changeset	562 0, // 0xad
cea50d98a6cf start work on content url scanner carl parents: diff changeset	563 0, // 0xae
cea50d98a6cf start work on content url scanner carl parents: diff changeset	564 0, // 0xaf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	565 0, // 0xb0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	566 0, // 0xb1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	567 0, // 0xb2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	568 0, // 0xb3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	569 0, // 0xb4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	570 0, // 0xb5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	571 0, // 0xb6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	572 0, // 0xb7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	573 0, // 0xb8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	574 0, // 0xb9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	575 0, // 0xba
cea50d98a6cf start work on content url scanner carl parents: diff changeset	576 0, // 0xbb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	577 0, // 0xbc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	578 0, // 0xbd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	579 0, // 0xbe
cea50d98a6cf start work on content url scanner carl parents: diff changeset	580 0, // 0xbf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	581 0, // 0xc0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	582 0, // 0xc1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	583 0, // 0xc2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	584 0, // 0xc3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	585 0, // 0xc4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	586 0, // 0xc5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	587 0, // 0xc6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	588 0, // 0xc7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	589 0, // 0xc8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	590 0, // 0xc9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	591 0, // 0xca
cea50d98a6cf start work on content url scanner carl parents: diff changeset	592 0, // 0xcb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	593 0, // 0xcc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	594 0, // 0xcd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	595 0, // 0xce
cea50d98a6cf start work on content url scanner carl parents: diff changeset	596 0, // 0xcf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	597 0, // 0xd0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	598 0, // 0xd1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	599 0, // 0xd2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	600 0, // 0xd3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	601 0, // 0xd4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	602 0, // 0xd5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	603 0, // 0xd6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	604 0, // 0xd7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	605 0, // 0xd8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	606 0, // 0xd9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	607 0, // 0xda
cea50d98a6cf start work on content url scanner carl parents: diff changeset	608 0, // 0xdb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	609 0, // 0xdc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	610 0, // 0xdd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	611 0, // 0xde
cea50d98a6cf start work on content url scanner carl parents: diff changeset	612 0, // 0xdf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	613 0, // 0xe0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	614 0, // 0xe1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	615 0, // 0xe2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	616 0, // 0xe3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	617 0, // 0xe4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	618 0, // 0xe5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	619 0, // 0xe6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	620 0, // 0xe7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	621 0, // 0xe8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	622 0, // 0xe9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	623 0, // 0xea
cea50d98a6cf start work on content url scanner carl parents: diff changeset	624 0, // 0xeb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	625 0, // 0xec
cea50d98a6cf start work on content url scanner carl parents: diff changeset	626 0, // 0xed
cea50d98a6cf start work on content url scanner carl parents: diff changeset	627 0, // 0xee
cea50d98a6cf start work on content url scanner carl parents: diff changeset	628 0, // 0xef
cea50d98a6cf start work on content url scanner carl parents: diff changeset	629 0, // 0xf0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	630 0, // 0xf1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	631 0, // 0xf2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	632 0, // 0xf3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	633 0, // 0xf4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	634 0, // 0xf5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	635 0, // 0xf6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	636 0, // 0xf7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	637 0, // 0xf8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	638 0, // 0xf9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	639 0, // 0xfa
cea50d98a6cf start work on content url scanner carl parents: diff changeset	640 0, // 0xfb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	641 0, // 0xfc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	642 0, // 0xfd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	643 0, // 0xfe
cea50d98a6cf start work on content url scanner carl parents: diff changeset	644 0, // 0xff
cea50d98a6cf start work on content url scanner carl parents: diff changeset	645 };
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	646
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	647 u_char b64_decode[256] = {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	648 0, // 0x00
cea50d98a6cf start work on content url scanner carl parents: diff changeset	649 0, // 0x01
cea50d98a6cf start work on content url scanner carl parents: diff changeset	650 0, // 0x02
cea50d98a6cf start work on content url scanner carl parents: diff changeset	651 0, // 0x03
cea50d98a6cf start work on content url scanner carl parents: diff changeset	652 0, // 0x04
cea50d98a6cf start work on content url scanner carl parents: diff changeset	653 0, // 0x05
cea50d98a6cf start work on content url scanner carl parents: diff changeset	654 0, // 0x06
cea50d98a6cf start work on content url scanner carl parents: diff changeset	655 0, // 0x07
cea50d98a6cf start work on content url scanner carl parents: diff changeset	656 0, // 0x08
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	657 0, // 0x09 <tab>
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	658 0, // 0x0a <lf>
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	659 0, // 0x0b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	660 0, // 0x0c
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	661 0, // 0x0d <cr>
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	662 0, // 0x0e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	663 0, // 0x0f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	664 0, // 0x10
cea50d98a6cf start work on content url scanner carl parents: diff changeset	665 0, // 0x11 xon char
cea50d98a6cf start work on content url scanner carl parents: diff changeset	666 0, // 0x12
cea50d98a6cf start work on content url scanner carl parents: diff changeset	667 0, // 0x13 xoff char
cea50d98a6cf start work on content url scanner carl parents: diff changeset	668 0, // 0x14
cea50d98a6cf start work on content url scanner carl parents: diff changeset	669 0, // 0x15
cea50d98a6cf start work on content url scanner carl parents: diff changeset	670 0, // 0x16
cea50d98a6cf start work on content url scanner carl parents: diff changeset	671 0, // 0x17
cea50d98a6cf start work on content url scanner carl parents: diff changeset	672 0, // 0x18
cea50d98a6cf start work on content url scanner carl parents: diff changeset	673 0, // 0x19
cea50d98a6cf start work on content url scanner carl parents: diff changeset	674 0, // 0x1a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	675 0, // 0x1b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	676 0, // 0x1c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	677 0, // 0x1d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	678 0, // 0x1e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	679 0, // 0x1f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	680 0, // 0x20 space
cea50d98a6cf start work on content url scanner carl parents: diff changeset	681 0, // 0x21 !
cea50d98a6cf start work on content url scanner carl parents: diff changeset	682 0, // 0x22 ""
cea50d98a6cf start work on content url scanner carl parents: diff changeset	683 0, // 0x23 #
cea50d98a6cf start work on content url scanner carl parents: diff changeset	684 0, // 0x24 $
cea50d98a6cf start work on content url scanner carl parents: diff changeset	685 0, // 0x25 %
cea50d98a6cf start work on content url scanner carl parents: diff changeset	686 0, // 0x26 &
cea50d98a6cf start work on content url scanner carl parents: diff changeset	687 0, // 0x27 '
cea50d98a6cf start work on content url scanner carl parents: diff changeset	688 0, // 0x28 (
cea50d98a6cf start work on content url scanner carl parents: diff changeset	689 0, // 0x29 )
cea50d98a6cf start work on content url scanner carl parents: diff changeset	690 0, // 0x2A *
cea50d98a6cf start work on content url scanner carl parents: diff changeset	691 62, // 0x2B +
cea50d98a6cf start work on content url scanner carl parents: diff changeset	692 0, // 0x2C ,
cea50d98a6cf start work on content url scanner carl parents: diff changeset	693 0, // 0x2D -
cea50d98a6cf start work on content url scanner carl parents: diff changeset	694 0, // 0x2E .
cea50d98a6cf start work on content url scanner carl parents: diff changeset	695 63, // 0x2F /
cea50d98a6cf start work on content url scanner carl parents: diff changeset	696 52, // 0x30 0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	697 53, // 0x31 1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	698 54, // 0x32 2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	699 55, // 0x33 3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	700 56, // 0x34 4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	701 57, // 0x35 5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	702 58, // 0x36 6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	703 59, // 0x37 7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	704 60, // 0x38 8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	705 61, // 0x39 9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	706 0, // 0x3A :
cea50d98a6cf start work on content url scanner carl parents: diff changeset	707 0, // 0x3B ;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	708 0, // 0x3C <
cea50d98a6cf start work on content url scanner carl parents: diff changeset	709 0, // 0x3D =
cea50d98a6cf start work on content url scanner carl parents: diff changeset	710 0, // 0x3E >
cea50d98a6cf start work on content url scanner carl parents: diff changeset	711 0, // 0x3F ?
cea50d98a6cf start work on content url scanner carl parents: diff changeset	712 0, // 0x40 @
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	713 0, // 0x41 A
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	714 1, // 0x42 B
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	715 2, // 0x43 C
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	716 3, // 0x44 D
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	717 4, // 0x45 E
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	718 5, // 0x46 F
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	719 6, // 0x47 G
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	720 7, // 0x48 H
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	721 8, // 0x49 I
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	722 9, // 0x4A J
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	723 10, // 0x4B K
cea50d98a6cf start work on content url scanner carl parents: diff changeset	724 11, // 0x4C L
cea50d98a6cf start work on content url scanner carl parents: diff changeset	725 12, // 0x4D M
cea50d98a6cf start work on content url scanner carl parents: diff changeset	726 13, // 0x4E N
cea50d98a6cf start work on content url scanner carl parents: diff changeset	727 14, // 0x4F O
cea50d98a6cf start work on content url scanner carl parents: diff changeset	728 15, // 0x50 P
cea50d98a6cf start work on content url scanner carl parents: diff changeset	729 16, // 0x51 Q
cea50d98a6cf start work on content url scanner carl parents: diff changeset	730 17, // 0x52 R
cea50d98a6cf start work on content url scanner carl parents: diff changeset	731 18, // 0x53 S
cea50d98a6cf start work on content url scanner carl parents: diff changeset	732 19, // 0x54 T
cea50d98a6cf start work on content url scanner carl parents: diff changeset	733 20, // 0x55 U
cea50d98a6cf start work on content url scanner carl parents: diff changeset	734 21, // 0x56 V
cea50d98a6cf start work on content url scanner carl parents: diff changeset	735 22, // 0x57 W
cea50d98a6cf start work on content url scanner carl parents: diff changeset	736 23, // 0x58 X
cea50d98a6cf start work on content url scanner carl parents: diff changeset	737 24, // 0x59 Y
cea50d98a6cf start work on content url scanner carl parents: diff changeset	738 25, // 0x5A Z
cea50d98a6cf start work on content url scanner carl parents: diff changeset	739 0, // 0x5B [
cea50d98a6cf start work on content url scanner carl parents: diff changeset	740 0, // 0x5C brace
cea50d98a6cf start work on content url scanner carl parents: diff changeset	741 0, // 0x5D ]
cea50d98a6cf start work on content url scanner carl parents: diff changeset	742 0, // 0x5E ^
cea50d98a6cf start work on content url scanner carl parents: diff changeset	743 0, // 0x5F _
cea50d98a6cf start work on content url scanner carl parents: diff changeset	744 0, // 0x60 `
cea50d98a6cf start work on content url scanner carl parents: diff changeset	745 26, // 0x61 a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	746 27, // 0x62 b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	747 28, // 0x63 c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	748 29, // 0x64 d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	749 30, // 0x65 e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	750 31, // 0x66 f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	751 32, // 0x67 g
cea50d98a6cf start work on content url scanner carl parents: diff changeset	752 33, // 0x68 h
cea50d98a6cf start work on content url scanner carl parents: diff changeset	753 34, // 0x69 i
cea50d98a6cf start work on content url scanner carl parents: diff changeset	754 35, // 0x6A j
cea50d98a6cf start work on content url scanner carl parents: diff changeset	755 36, // 0x6B k
cea50d98a6cf start work on content url scanner carl parents: diff changeset	756 37, // 0x6C l
cea50d98a6cf start work on content url scanner carl parents: diff changeset	757 38, // 0x6D m
cea50d98a6cf start work on content url scanner carl parents: diff changeset	758 39, // 0x6E n
cea50d98a6cf start work on content url scanner carl parents: diff changeset	759 40, // 0x6F o
cea50d98a6cf start work on content url scanner carl parents: diff changeset	760 41, // 0x70 p
cea50d98a6cf start work on content url scanner carl parents: diff changeset	761 42, // 0x71 q
cea50d98a6cf start work on content url scanner carl parents: diff changeset	762 43, // 0x72 r
cea50d98a6cf start work on content url scanner carl parents: diff changeset	763 44, // 0x73 s
cea50d98a6cf start work on content url scanner carl parents: diff changeset	764 45, // 0x74 t
cea50d98a6cf start work on content url scanner carl parents: diff changeset	765 46, // 0x75 u
cea50d98a6cf start work on content url scanner carl parents: diff changeset	766 47, // 0x76 v
cea50d98a6cf start work on content url scanner carl parents: diff changeset	767 48, // 0x77 w
cea50d98a6cf start work on content url scanner carl parents: diff changeset	768 49, // 0x78 x
cea50d98a6cf start work on content url scanner carl parents: diff changeset	769 50, // 0x79 y
cea50d98a6cf start work on content url scanner carl parents: diff changeset	770 51, // 0x7A z
cea50d98a6cf start work on content url scanner carl parents: diff changeset	771 0, // 0x7B {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	772 0, // 0x7C \|
cea50d98a6cf start work on content url scanner carl parents: diff changeset	773 0, // 0x7D }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	774 0, // 0x7E ~
cea50d98a6cf start work on content url scanner carl parents: diff changeset	775 0, // 0x7f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	776 0, // 0x80
cea50d98a6cf start work on content url scanner carl parents: diff changeset	777 0, // 0x81
cea50d98a6cf start work on content url scanner carl parents: diff changeset	778 0, // 0x82
cea50d98a6cf start work on content url scanner carl parents: diff changeset	779 0, // 0x83
cea50d98a6cf start work on content url scanner carl parents: diff changeset	780 0, // 0x84
cea50d98a6cf start work on content url scanner carl parents: diff changeset	781 0, // 0x85
cea50d98a6cf start work on content url scanner carl parents: diff changeset	782 0, // 0x86
cea50d98a6cf start work on content url scanner carl parents: diff changeset	783 0, // 0x87
cea50d98a6cf start work on content url scanner carl parents: diff changeset	784 0, // 0x88
cea50d98a6cf start work on content url scanner carl parents: diff changeset	785 0, // 0x89
cea50d98a6cf start work on content url scanner carl parents: diff changeset	786 0, // 0x8a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	787 0, // 0x8b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	788 0, // 0x8c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	789 0, // 0x8d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	790 0, // 0x8e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	791 0, // 0x8f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	792 0, // 0x90
cea50d98a6cf start work on content url scanner carl parents: diff changeset	793 0, // 0x91
cea50d98a6cf start work on content url scanner carl parents: diff changeset	794 0, // 0x92
cea50d98a6cf start work on content url scanner carl parents: diff changeset	795 0, // 0x93
cea50d98a6cf start work on content url scanner carl parents: diff changeset	796 0, // 0x94
cea50d98a6cf start work on content url scanner carl parents: diff changeset	797 0, // 0x95
cea50d98a6cf start work on content url scanner carl parents: diff changeset	798 0, // 0x96
cea50d98a6cf start work on content url scanner carl parents: diff changeset	799 0, // 0x97
cea50d98a6cf start work on content url scanner carl parents: diff changeset	800 0, // 0x98
cea50d98a6cf start work on content url scanner carl parents: diff changeset	801 0, // 0x99
cea50d98a6cf start work on content url scanner carl parents: diff changeset	802 0, // 0x9a
cea50d98a6cf start work on content url scanner carl parents: diff changeset	803 0, // 0x9b
cea50d98a6cf start work on content url scanner carl parents: diff changeset	804 0, // 0x9c
cea50d98a6cf start work on content url scanner carl parents: diff changeset	805 0, // 0x9d
cea50d98a6cf start work on content url scanner carl parents: diff changeset	806 0, // 0x9e
cea50d98a6cf start work on content url scanner carl parents: diff changeset	807 0, // 0x9f
cea50d98a6cf start work on content url scanner carl parents: diff changeset	808 0, // 0xa0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	809 0, // 0xa1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	810 0, // 0xa2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	811 0, // 0xa3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	812 0, // 0xa4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	813 0, // 0xa5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	814 0, // 0xa6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	815 0, // 0xa7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	816 0, // 0xa8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	817 0, // 0xa9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	818 0, // 0xaa
cea50d98a6cf start work on content url scanner carl parents: diff changeset	819 0, // 0xab
cea50d98a6cf start work on content url scanner carl parents: diff changeset	820 0, // 0xac
cea50d98a6cf start work on content url scanner carl parents: diff changeset	821 0, // 0xad
cea50d98a6cf start work on content url scanner carl parents: diff changeset	822 0, // 0xae
cea50d98a6cf start work on content url scanner carl parents: diff changeset	823 0, // 0xaf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	824 0, // 0xb0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	825 0, // 0xb1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	826 0, // 0xb2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	827 0, // 0xb3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	828 0, // 0xb4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	829 0, // 0xb5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	830 0, // 0xb6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	831 0, // 0xb7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	832 0, // 0xb8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	833 0, // 0xb9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	834 0, // 0xba
cea50d98a6cf start work on content url scanner carl parents: diff changeset	835 0, // 0xbb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	836 0, // 0xbc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	837 0, // 0xbd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	838 0, // 0xbe
cea50d98a6cf start work on content url scanner carl parents: diff changeset	839 0, // 0xbf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	840 0, // 0xc0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	841 0, // 0xc1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	842 0, // 0xc2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	843 0, // 0xc3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	844 0, // 0xc4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	845 0, // 0xc5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	846 0, // 0xc6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	847 0, // 0xc7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	848 0, // 0xc8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	849 0, // 0xc9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	850 0, // 0xca
cea50d98a6cf start work on content url scanner carl parents: diff changeset	851 0, // 0xcb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	852 0, // 0xcc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	853 0, // 0xcd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	854 0, // 0xce
cea50d98a6cf start work on content url scanner carl parents: diff changeset	855 0, // 0xcf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	856 0, // 0xd0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	857 0, // 0xd1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	858 0, // 0xd2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	859 0, // 0xd3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	860 0, // 0xd4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	861 0, // 0xd5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	862 0, // 0xd6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	863 0, // 0xd7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	864 0, // 0xd8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	865 0, // 0xd9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	866 0, // 0xda
cea50d98a6cf start work on content url scanner carl parents: diff changeset	867 0, // 0xdb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	868 0, // 0xdc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	869 0, // 0xdd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	870 0, // 0xde
cea50d98a6cf start work on content url scanner carl parents: diff changeset	871 0, // 0xdf
cea50d98a6cf start work on content url scanner carl parents: diff changeset	872 0, // 0xe0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	873 0, // 0xe1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	874 0, // 0xe2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	875 0, // 0xe3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	876 0, // 0xe4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	877 0, // 0xe5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	878 0, // 0xe6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	879 0, // 0xe7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	880 0, // 0xe8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	881 0, // 0xe9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	882 0, // 0xea
cea50d98a6cf start work on content url scanner carl parents: diff changeset	883 0, // 0xeb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	884 0, // 0xec
cea50d98a6cf start work on content url scanner carl parents: diff changeset	885 0, // 0xed
cea50d98a6cf start work on content url scanner carl parents: diff changeset	886 0, // 0xee
cea50d98a6cf start work on content url scanner carl parents: diff changeset	887 0, // 0xef
cea50d98a6cf start work on content url scanner carl parents: diff changeset	888 0, // 0xf0
cea50d98a6cf start work on content url scanner carl parents: diff changeset	889 0, // 0xf1
cea50d98a6cf start work on content url scanner carl parents: diff changeset	890 0, // 0xf2
cea50d98a6cf start work on content url scanner carl parents: diff changeset	891 0, // 0xf3
cea50d98a6cf start work on content url scanner carl parents: diff changeset	892 0, // 0xf4
cea50d98a6cf start work on content url scanner carl parents: diff changeset	893 0, // 0xf5
cea50d98a6cf start work on content url scanner carl parents: diff changeset	894 0, // 0xf6
cea50d98a6cf start work on content url scanner carl parents: diff changeset	895 0, // 0xf7
cea50d98a6cf start work on content url scanner carl parents: diff changeset	896 0, // 0xf8
cea50d98a6cf start work on content url scanner carl parents: diff changeset	897 0, // 0xf9
cea50d98a6cf start work on content url scanner carl parents: diff changeset	898 0, // 0xfa
cea50d98a6cf start work on content url scanner carl parents: diff changeset	899 0, // 0xfb
cea50d98a6cf start work on content url scanner carl parents: diff changeset	900 0, // 0xfc
cea50d98a6cf start work on content url scanner carl parents: diff changeset	901 0, // 0xfd
cea50d98a6cf start work on content url scanner carl parents: diff changeset	902 0, // 0xfe
cea50d98a6cf start work on content url scanner carl parents: diff changeset	903 0, // 0xff
cea50d98a6cf start work on content url scanner carl parents: diff changeset	904 };
cea50d98a6cf start work on content url scanner carl parents: diff changeset	905
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	906 #define PENDING_LIMIT 100
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	907 struct fsa {
12 6ac6d6b822ce fix memory leak with duplicate url host names, carl parents: 11 diff changeset	908 u_char pending[PENDING_LIMIT];
6ac6d6b822ce fix memory leak with duplicate url host names, carl parents: 11 diff changeset	909 int count;
6ac6d6b822ce fix memory leak with duplicate url host names, carl parents: 11 diff changeset	910 state st;
6ac6d6b822ce fix memory leak with duplicate url host names, carl parents: 11 diff changeset	911 state init;
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	912 fsa* next1;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	913 fsa* next2;
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	914 recorder *memory;
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	915
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	916 fsa(state init, fsa* next1_, fsa* next2_, recorder *memory_);
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	917 void push(u_char *buf, int len);
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	918 void pusher();
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	919 void error(char *err);
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	920 };
cea50d98a6cf start work on content url scanner carl parents: diff changeset	921
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	922 fsa::fsa(state init_, fsa next1_, fsa next2_, recorder *memory_) {
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	923 count = 0;
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	924 st = init_;
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	925 init = init_;
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	926 next1 = next1_;
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	927 next2 = next2_;
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	928 memory = memory_;
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	929 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	930
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	931 void fsa::error(char *err) {
b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	932 count = 0;
b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	933 st = init;
42 afcf403709ef updates for 3.2, try to drop root privileges carl parents: 41 diff changeset	934 if (err) my_syslog(memory->priv, err);
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	935 }
b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	936
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	937 void fsa::pusher() {
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	938 if (next1) next1->push(pending, count);
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	939 if (next2) next2->push(pending, count);
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	940 count = 0;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	941 }
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	942
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	943 void fsa::push(u_char *buf, int len) {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	944 for (int i=0; i<len; i++) {
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	945 if (count == (PENDING_LIMIT-1)) error(NULL);
b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	946 if (st >= end_state) error("finite state machine impossible state");
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	947 u_char c = buf[i];
cea50d98a6cf start work on content url scanner carl parents: diff changeset	948 pending[count++] = c;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	949 st = parse_table[c][st];
cea50d98a6cf start work on content url scanner carl parents: diff changeset	950 switch (st) {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	951
cea50d98a6cf start work on content url scanner carl parents: diff changeset	952 //////////////////////////////
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	953 // host name recognizer
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	954 case h_end: {
18 041ea016b684 add scanning for bare hostnames carl parents: 17 diff changeset	955 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	956 if (!strchr((const char *)pending, '@')) {
43a4f6b3e668 add configurable host name limit and bad html tag limits. carl parents: 25 diff changeset	957 // not an email address or message id
28 33e1e3910506 add configurable list of tlds carl parents: 27 diff changeset	958 char p1 = strchr((const char )pending, '.');
33e1e3910506 add configurable list of tlds carl parents: 27 diff changeset	959 char p2 = strrchr((const char )pending, '.');
52 a84752107aca host names cannot have consecutive periods. carl parents: 50 diff changeset	960 char p3 = strstr((const char )pending, "..");
a84752107aca host names cannot have consecutive periods. carl parents: 50 diff changeset	961 if (p1 && (p1 != p2) & !p3) {
a84752107aca host names cannot have consecutive periods. carl parents: 50 diff changeset	962 // have two periods, so at least three components, and no empty components
44 6b79046b18c2 changes for 3.2 carl parents: 42 diff changeset	963 for (int i=0; i<count; i++) pending[i] = tolower(pending[i]);
28 33e1e3910506 add configurable list of tlds carl parents: 27 diff changeset	964 // is last component a tld?
29 4dfdf33f1db0 add syslog msg freeing memory, use bare tld names without leading period carl parents: 28 diff changeset	965 string_set::iterator i = memory->tlds->find(p2+1);
28 33e1e3910506 add configurable list of tlds carl parents: 27 diff changeset	966 if (i != memory->tlds->end()) memory->new_url((char*)pending);
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	967 }
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	968 }
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	969 st = h_init;
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	970 } // fall thru
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	971
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	972 case h_init: {
18 041ea016b684 add scanning for bare hostnames carl parents: 17 diff changeset	973 count = 0;
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	974 } break;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	975
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	976
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	977 //////////////////////////////
18 041ea016b684 add scanning for bare hostnames carl parents: 17 diff changeset	978 // 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	979 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	980 memory->binary();
6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	981 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	982 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	983 } break;
6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	984
18 041ea016b684 add scanning for bare hostnames carl parents: 17 diff changeset	985 case t_end: {
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	986 if (count > 1) {
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	987 pending[--count] = '\0'; // null terminate html tag
33 ce229348cdbe failed to lowercase the first char of html tags carl parents: 31 diff changeset	988 for (int i=0; i<count; i++) pending[i] = tolower(pending[i]);
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	989 memory->new_tag((char*)pending);
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	990 }
25 6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	991 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	992 } // fall thru
6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	993
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	994 case t_tag1:
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	995 case t_com2:
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	996 case t_com3:
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	997 case t_com4:
8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	998 case t_com5:
25 6176e7b2e8af better scanning for bad html tags, allow binary zip and gz files with random char sequences carl parents: 24 diff changeset	999 case t_disc: {
18 041ea016b684 add scanning for bare hostnames carl parents: 17 diff changeset	1000 count = 0; // discard all characters
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1001 } break;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1002
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1003 case t_init: {
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1004 pusher();
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1005 } break;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1006
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1007
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1008 //////////////////////////////
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1009 // url recognizer
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1010 case u_reco: {
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	1011 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	1012 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	1013 char p = strrchr((const char )pending, '/');
52 a84752107aca host names cannot have consecutive periods. carl parents: 50 diff changeset	1014 if (p && // have a leading /
a84752107aca host names cannot have consecutive periods. carl parents: 50 diff changeset	1015 strchr(p, '.') && // require at least one . in a dns name
a84752107aca host names cannot have consecutive periods. carl parents: 50 diff changeset	1016 !strstr(p, "..") && // no empty components in the dns name
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1017 (strncasecmp((const char *)pending, "http", 4) == 0)) { // must start with protocol
44 6b79046b18c2 changes for 3.2 carl parents: 42 diff changeset	1018 // we seem to have a host name
6b79046b18c2 changes for 3.2 carl parents: 42 diff changeset	1019 p++; // skip the last /
6b79046b18c2 changes for 3.2 carl parents: 42 diff changeset	1020 int c = strlen(p);
6b79046b18c2 changes for 3.2 carl parents: 42 diff changeset	1021 for (int i=0; i<c; i++) p[i] = tolower(p[i]);
6b79046b18c2 changes for 3.2 carl parents: 42 diff changeset	1022 memory->new_url(p); // record it
9 8c65411cd7ab integration work on url scanner carl parents: 8 diff changeset	1023 }
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1024 }
19 b8f5fa3dd5b8 fix problems in the state transitions causing impossible states carl parents: 18 diff changeset	1025 st = u_init;
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1026 } // fall thru
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1027
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1028 case u_init: {
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1029 count = 0; // discard all characters
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1030 } break;
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1031
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1032
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1033 //////////////////////////////
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1034 // url decoder %xx
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1035 case d_2: {
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1036 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	1037 count = 1;
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1038 st = d_init;
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1039 } // fall thru
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1040
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1041 case d_init: {
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1042 pusher();
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1043 } break;
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1044
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1045
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1046 //////////////////////////////
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1047 // html entity decoder &#nnn; &#xnnn;
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1048 case e_semi: {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1049 pending[--count] = '\0'; // null terminate the digit string by overwriting the semicolon
41 d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1050 u_char f = pending[2];
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1051 if ((f == 'x') \|\| (f == 'X')) {
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1052 pending[0] = strtol((const char )pending+2, (char *)NULL, 16);
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1053 }
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1054 else {
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1055 pending[0] = atoi((const char *)pending+2);
d95af8129dfa updates for 3.2, changing file layout, add queueid to messages carl parents: 37 diff changeset	1056 }
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1057 count = 1;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1058 st = e_init;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1059 } // fall thru
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1060
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1061 case e_init: {
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1062 pusher();
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1063 } break;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1064
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1065
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1066 //////////////////////////////
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1067 // mime decoder =xx
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1068 case m_2: {
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1069 pending[0] = hex_decode[pending[1]] * 16 + hex_decode[pending[2]];
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1070 count = 1;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1071 st = m_init;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1072 } // fall thru
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1073
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1074 case m_init: {
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1075 pusher();
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1076 } break;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1077
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1078 case m_cr: {
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1079 count = 1;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1080 st = m_eq;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1081 } break;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1082
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1083 case m_nl: {
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1084 count = 0;
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1085 st = m_init;
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1086 } break;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1087
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1088
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1089 //////////////////////////////
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1090 // base64 decoder
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1091 case b_lf2: {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1092 count--;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1093 } break;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1094
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1095 case b_cr: {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1096 int cnt = 0;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1097 if ((count % 4) == 1) {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1098 count--;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1099 // might have proper b64 data
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1100 for (int i=0; i<count; i+=4) {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1101 unsigned long a1 = b64_decode[pending[i]];
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1102 unsigned long a2 = b64_decode[pending[i+1]];
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1103 unsigned long a3 = b64_decode[pending[i+2]];
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1104 unsigned long a4 = b64_decode[pending[i+3]];
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1105 unsigned long a = (a1 << 18) \| (a2 << 12) \| (a3 << 6) \| a4;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1106 pending[cnt++] = (a & 0x00ff0000) >> 16;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1107 pending[cnt++] = (a & 0x0000ff00) >> 8;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1108 pending[cnt++] = (a & 0x000000ff);
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1109 if ((char)pending[i+3] == '=') cnt--;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1110 if ((char)pending[i+2] == '=') cnt--;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1111 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1112 count = cnt;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1113 st = b_lf2;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1114 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1115 else st = b_init;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1116 } // fall thru
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1117
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1118 case b_lf:
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1119 case b_init: {
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1120 pusher();
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1121 } break;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1122
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1123
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1124 //////////////////////////////
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1125 // states that just accumulate characters in the pending buffer
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1126 case h_host:
30 8f43f8fa1ad7 ignore text including bad html tags inside comments carl parents: 29 diff changeset	1127 case t_tag2:
31 d16b27371533 consider <!tag> to be bad html tag carl parents: 30 diff changeset	1128 case t_com1:
7 93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1129 case u_http:
93ff6d1ef647 stable standalone scanner module carl parents: 6 diff changeset	1130 case u_url:
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1131 case u_sla:
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1132 case d_pcnt:
06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1133 case d_1:
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1134 case e_amp:
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1135 case e_num:
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1136 case m_eq:
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1137 case m_1:
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1138 case b_64:
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1139 default: {
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1140 } break;
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1141 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1142 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1143 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1144
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1145 struct url_scanner {
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1146 fsa *host_parser;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1147 fsa *tags_parser;
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1148 fsa *urls_parser;
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1149 fsa *urld_parser;
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1150 fsa *html_parser;
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1151 fsa *mime_parser;
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1152 fsa *b64_parser;
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1153
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	1154 url_scanner(recorder *memory);
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1155 ~url_scanner();
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1156 void scan(u_char *buffer, size_t length);
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1157 };
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1158
24 2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	1159 url_scanner::url_scanner(recorder *memory) {
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	1160 host_parser = new fsa(h_init, NULL, NULL, memory);
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	1161 tags_parser = new fsa(t_init, host_parser, NULL, memory);
2e23b7184d2b start coding for bad html tag detection carl parents: 23 diff changeset	1162 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	1163 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	1164 html_parser = new fsa(e_init, urld_parser, NULL, NULL);
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1165 mime_parser = new fsa(m_init, html_parser, NULL, NULL);
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1166 b64_parser = new fsa(b_init, mime_parser, NULL, NULL);
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1167 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1168
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1169 url_scanner::~url_scanner() {
16 2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1170 delete host_parser;
2ae8d953f1d0 add scanning for bare hostnames carl parents: 12 diff changeset	1171 delete tags_parser;
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1172 delete urls_parser;
23 06de5ab6a232 add url decoding stage, allow http:/ single / in yahoo redirector, allow ip address hostnames carl parents: 19 diff changeset	1173 delete urld_parser;
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1174 delete html_parser;
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1175 delete mime_parser;
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1176 delete b64_parser;
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1177 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1178
8 dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1179 void url_scanner::scan(u_char *buffer, size_t length) {
dbe18921f741 integration work on url scanner carl parents: 7 diff changeset	1180 b64_parser->push(buffer, length);
6 cea50d98a6cf start work on content url scanner carl parents: diff changeset	1181 }
cea50d98a6cf start work on content url scanner carl parents: diff changeset	1182

Mercurial > dnsbl

annotate src/scanner.cpp @ 58:7bb8bbf79285