|
1
|
<?php
|
|
2
|
/**
|
|
3
|
*
|
|
4
|
* @category modules
|
|
5
|
* @package output_filter
|
|
6
|
* @author Christian Sommer, WB-Project, Werner v.d. Decken
|
|
7
|
* @copyright 2011, Website Baker Org. e.V.
|
|
8
|
* @link http://www.websitebaker2.org/
|
|
9
|
* @license http://www.gnu.org/licenses/gpl.html
|
|
10
|
* @platform WebsiteBaker 2.8.2
|
|
11
|
* @requirements PHP 5.2.2 and higher
|
|
12
|
* @version $Id: filter-routines.php 1534 2011-12-08 12:22:16Z darkviper $
|
|
13
|
* @filesource $HeadURL: svn://isteam.dynxs.de/wb-archiv/branches/2.8.x/wb/modules/output_filter/filter-routines.php $
|
|
14
|
* @lastmodified $Date: 2011-12-08 13:22:16 +0100 (Thu, 08 Dec 2011) $
|
|
15
|
*
|
|
16
|
*/
|
|
17
|
/* -------------------------------------------------------- */
|
|
18
|
// Must include code to stop this file being accessed directly
|
|
19
|
require_once( dirname(dirname(dirname(__FILE__))).'/framework/globalExceptionHandler.php');
|
|
20
|
if(!defined('WB_PATH')) { throw new IllegalFileException(); }
|
|
21
|
/* -------------------------------------------------------- */
|
|
22
|
|
|
23
|
/* ************************************************************************** */
|
|
24
|
/**
|
|
25
|
* execute the frontend output filter to modify email addresses
|
|
26
|
* @param string actual content
|
|
27
|
* @return string modified content
|
|
28
|
*/
|
|
29
|
function executeFrontendOutputFilter($content) {
|
|
30
|
// get output filter settings from database
|
|
31
|
$filter_settings = getOutputFilterSettings();
|
|
32
|
$output_filter_mode = 0;
|
|
33
|
$output_filter_mode |= ($filter_settings['email_filter'] * pow(2, 0)); // n | 2^0
|
|
34
|
$output_filter_mode |= ($filter_settings['mailto_filter'] * pow(2, 1)); // n | 2^1
|
|
35
|
define('OUTPUT_FILTER_AT_REPLACEMENT', $filter_settings['at_replacement']);
|
|
36
|
define('OUTPUT_FILTER_DOT_REPLACEMENT', $filter_settings['dot_replacement']);
|
|
37
|
/* ### filter type: full qualified URLs ##################################### */
|
|
38
|
if($filter_settings['sys_rel'] == 1){
|
|
39
|
$content = _doFilterRelUrl($content);
|
|
40
|
}
|
|
41
|
/* ### filter type: protect email addresses ################################# */
|
|
42
|
if( ($output_filter_mode & pow(2, 0)) || ($output_filter_mode & pow(2, 1)) ) {
|
|
43
|
$content = _doFilterEmail($content, $output_filter_mode);
|
|
44
|
}
|
|
45
|
/* ### end of filters ####################################################### */
|
|
46
|
return $content;
|
|
47
|
}
|
|
48
|
/* ************************************************************************** */
|
|
49
|
/**
|
|
50
|
* function to read the current filter settings
|
|
51
|
* @global object $database
|
|
52
|
* @global object $admin
|
|
53
|
* @param void
|
|
54
|
* @return array contains all settings
|
|
55
|
*/
|
|
56
|
function getOutputFilterSettings() {
|
|
57
|
global $database, $admin;
|
|
58
|
// set default values
|
|
59
|
$settings = array(
|
|
60
|
'sys_rel' => 0,
|
|
61
|
'email_filter' => 0,
|
|
62
|
'mailto_filter' => 0,
|
|
63
|
'at_replacement' => '(at)',
|
|
64
|
'dot_replacement' => '(dot)'
|
|
65
|
);
|
|
66
|
// be sure field 'sys_rel' is in table
|
|
67
|
$database->field_add( TABLE_PREFIX.'mod_output_filter', 'sys_rel', 'INT NOT NULL DEFAULT \'0\' FIRST');
|
|
68
|
// request settings from database
|
|
69
|
$sql = 'SELECT * FROM `'.TABLE_PREFIX.'mod_output_filter';
|
|
70
|
if(($res = $database->query($sql))) {
|
|
71
|
if(($rec = $res->fetchRow())) {
|
|
72
|
$settings = $rec;
|
|
73
|
$settings['at_replacement'] = $admin->strip_slashes($settings['at_replacement']);
|
|
74
|
$settings['dot_replacement'] = $admin->strip_slashes($settings['dot_replacement']);
|
|
75
|
}
|
|
76
|
}
|
|
77
|
// return array with filter settings
|
|
78
|
return $settings;
|
|
79
|
}
|
|
80
|
/* ************************************************************************** */
|
|
81
|
/**
|
|
82
|
* Convert full qualified, local URLs into relative URLs
|
|
83
|
* @param string $content
|
|
84
|
* @return string
|
|
85
|
*/
|
|
86
|
function _doFilterRelUrl($content) {
|
|
87
|
$content = preg_replace_callback(
|
|
88
|
'/((?:href|src)\s*=\s*")([^\"]*?)(")/iU',
|
|
89
|
create_function('$matches',
|
|
90
|
'$retval = $matches[0]; '.
|
|
91
|
'$h = parse_url($matches[2], PHP_URL_HOST); '.
|
|
92
|
'if(isset($h) && $h != \'\') { '.
|
|
93
|
'if(stripos(WB_URL, $h) !== false) { '.
|
|
94
|
'$a = parse_url($matches[2]); '.
|
|
95
|
'$p = (isset($a[\'path\']) ? $a[\'path\'] : \'\'); '.
|
|
96
|
'$q = (isset($a[\'query\']) ? \'?\'.$a[\'query\'] : \'\'); '.
|
|
97
|
'$f = (isset($a[\'fragment\']) ? \'#\'.$a[\'fragment\'] : \'\'); '.
|
|
98
|
'$p .= ($q.$f); '.
|
|
99
|
'$retval = $matches[1]."/".(isset($p) ? ltrim(str_replace("//", "/", $p), "/") : "").$matches[3]; '.
|
|
100
|
'}} return $retval;'),
|
|
101
|
$content);
|
|
102
|
return $content;
|
|
103
|
}
|
|
104
|
/* ************************************************************************** */
|
|
105
|
/**
|
|
106
|
* protect email addresses (replace '@' and '.' and obfuscate address
|
|
107
|
* @param string $content
|
|
108
|
* @return string
|
|
109
|
*/
|
|
110
|
function _doFilterEmail($content, $output_filter_mode) {
|
|
111
|
// test if js-decryption is installed
|
|
112
|
if( preg_match('/<head.*<.*src=\".*\/mdcr.js.*>.*<\/head/siU', $content) ) {
|
|
113
|
$output_filter_mode |= pow(2, 2); // n | 2^2
|
|
114
|
}else {
|
|
115
|
// try to insert js-decrypt into <head> if available
|
|
116
|
$script = str_replace('\\', '/', dirname(__FILE__)).'/js/mdcr.js';
|
|
117
|
if(is_readable($script)) {
|
|
118
|
$scriptLink = '<script src="'.$script.'" type="text/javascript"></script>';
|
|
119
|
$regex = '/(.*)(<\s*?\/\s*?head\s*>.*)/isU';
|
|
120
|
$replace = '$1'.$scriptLink.'$2';
|
|
121
|
$content = preg_replace ($regex, $replace, $content);
|
|
122
|
$output_filter_mode |= pow(2, 2); // n | 2^2
|
|
123
|
}
|
|
124
|
}
|
|
125
|
// define some constants so we do not call the database in the callback functions again
|
|
126
|
define('OUTPUT_FILTER_MODE', (int)$output_filter_mode);
|
|
127
|
/* *** obfuscate mailto addresses by js:mdcr *** */
|
|
128
|
// work out the defined output filter mode: possible output filter modes: [0], 1, 2, 3, 6, 7
|
|
129
|
// 2^0 * (0.. disable, 1.. enable) filtering of mail addresses in text
|
|
130
|
// 2^1 * (0.. disable, 1.. enable) filtering of mail addresses in mailto links
|
|
131
|
// 2^2 * (0.. disable, 1.. enable) Javascript mailto encryption (only if mailto filtering enabled)
|
|
132
|
|
|
133
|
// first search part to find all mailto email addresses
|
|
134
|
$pattern = '#(<a[^<]*href\s*?=\s*?"\s*?mailto\s*?:\s*?)([A-Z0-9._%+-]+@(?:[A-Z0-9-]+\.)+[A-Z]{2,4})([^"]*?)"([^>]*>)(.*?)</a>';
|
|
135
|
// second part to find all non mailto email addresses
|
|
136
|
$pattern .= '|(value\s*=\s*"|\')??\b([A-Z0-9._%+-]+@(?:[A-Z0-9-]+\.)+[A-Z]{2,4})\b#i';
|
|
137
|
/*
|
|
138
|
Sub 1:\b(<a.[^<]*href\s*?=\s*?"\s*?mailto\s*?:\s*?) --> "<a id="yyy" class="xxx" href = " mailto :" ignoring white spaces
|
|
139
|
Sub 2:([A-Z0-9._%+-]+@(?:[A-Z0-9-]+\.)+[A-Z]{2,4}) --> the email address in the mailto: part of the mail link
|
|
140
|
Sub 3:([^"]*?)" --> possible ?Subject&cc... stuff attached to the mail address
|
|
141
|
Sub 4:([^>]*>) --> all class or id statements after the mailto but before closing ..>
|
|
142
|
Sub 5:(.*?)</a>\b --> the mailto text; all characters between >xxxxx</a>
|
|
143
|
Sub 6:|\b([A-Z0-9._%+-]+@(?:[A-Z0-9-]+\.)+[A-Z]{2,4})\b --> email addresses which may appear in the text (require word boundaries)
|
|
144
|
*/
|
|
145
|
// find all email addresses embedded in the content and filter them using a callback function
|
|
146
|
$content = preg_replace_callback($pattern, '_cbDoExecuteFilter', $content);
|
|
147
|
return $content;
|
|
148
|
}
|
|
149
|
/* ************************************************************************** */
|
|
150
|
/**
|
|
151
|
* callback-function for function _doFilterEmail() to proceed search results
|
|
152
|
* @param array results from preg_replace
|
|
153
|
* @return string proceeded replacement string
|
|
154
|
*/
|
|
155
|
function _cbDoExecuteFilter($match) {
|
|
156
|
// check if required arguments are defined
|
|
157
|
if(!( defined('OUTPUT_FILTER_MODE')
|
|
158
|
&& defined('OUTPUT_FILTER_AT_REPLACEMENT')
|
|
159
|
&& defined('OUTPUT_FILTER_MODE')
|
|
160
|
) ) {
|
|
161
|
return $match[0];
|
|
162
|
}
|
|
163
|
$search = array('@', '.');
|
|
164
|
$replace = array(OUTPUT_FILTER_AT_REPLACEMENT ,OUTPUT_FILTER_DOT_REPLACEMENT);
|
|
165
|
// check if the match contains the expected number of subpatterns (6|8)
|
|
166
|
switch (count($match)) {
|
|
167
|
case 8:
|
|
168
|
/** OUTPUT FILTER FOR EMAIL ADDRESSES EMBEDDED IN TEXT **/
|
|
169
|
// 1.. text mails only, 3.. text mails + mailto (no JS), 7 text mails + mailto (JS)
|
|
170
|
if(!in_array(OUTPUT_FILTER_MODE, array(1,3,7))) return $match[0];
|
|
171
|
// do not filter mail addresses included in input tags (<input ... value = "test@mail)
|
|
172
|
if (strpos($match[6], 'value') !== false) return $match[0];
|
|
173
|
// filtering of non mailto email addresses enabled
|
|
174
|
return str_replace($search, $replace, $match[0]);
|
|
175
|
break;
|
|
176
|
case 6:
|
|
177
|
/** OUTPUT FILTER FOR EMAIL ADDRESSES EMBEDDED IN MAILTO LINKS **/
|
|
178
|
// 2.. mailto only (no JS), 3.. text mails + mailto (no JS), 6.. mailto only (JS), 7.. all filters active
|
|
179
|
if(!in_array(OUTPUT_FILTER_MODE, array(2,3,6,7))) return $match[0];
|
|
180
|
// check if last part of the a href link: >xxxx</a> contains a email address we need to filter
|
|
181
|
$pattern = '#[A-Z0-9._%+-]+@(?:[A-Z0-9-]+\.)+[A-Z]{2,4}#i';
|
|
182
|
if(preg_match_all($pattern, $match[5], $matches)) {
|
|
183
|
foreach($matches as $submatch) {
|
|
184
|
foreach($submatch as $value) {
|
|
185
|
// replace all . and all @ in email address parts by (dot) and (at) strings
|
|
186
|
$match[5] = str_replace($value, str_replace($search, $replace, $value), $match[5]);
|
|
187
|
}
|
|
188
|
}
|
|
189
|
}
|
|
190
|
// check if Javascript encryption routine is enabled
|
|
191
|
if(in_array(OUTPUT_FILTER_MODE, array(6,7))) {
|
|
192
|
/** USE JAVASCRIPT ENCRYPTION FOR MAILTO LINKS **/
|
|
193
|
// extract possible class and id attribute from ahref link
|
|
194
|
preg_match('/class\s*?=\s*?("|\')(.*?)\1/ix', $match[0], $class_attr);
|
|
195
|
$class_attr = empty($class_attr) ? '' : 'class="' . $class_attr[2] . '" ';
|
|
196
|
preg_match('/id\s*?=\s*?("|\')(.*?)\1/ix', $match[0], $id_attr);
|
|
197
|
$id_attr = empty($id_attr) ? '' : 'id="' . $id_attr[2] . '" ';
|
|
198
|
// preprocess mailto link parts for further usage
|
|
199
|
$search = array('@', '.', '_', '-'); $replace = array('F', 'Z', 'X', 'K');
|
|
200
|
$email_address = str_replace($search, $replace, strtolower($match[2]));
|
|
201
|
$email_subject = rawurlencode(html_entity_decode($match[3]));
|
|
202
|
// create a random encryption key for the Caesar cipher
|
|
203
|
mt_srand((double)microtime()*1000000); // (PHP < 4.2.0)
|
|
204
|
$shift = mt_rand(1, 25);
|
|
205
|
// encrypt the email using an adapted Caesar cipher
|
|
206
|
$encrypted_email = "";
|
|
207
|
for($i = strlen($email_address) -1; $i > -1; $i--) {
|
|
208
|
if(preg_match('#[FZXK0-9]#', $email_address[$i], $characters)) {
|
|
209
|
$encrypted_email .= $email_address[$i];
|
|
210
|
} else {
|
|
211
|
$encrypted_email .= chr((ord($email_address[$i]) -97 + $shift) % 26 + 97);
|
|
212
|
}
|
|
213
|
}
|
|
214
|
$encrypted_email .= chr($shift + 97);
|
|
215
|
// build the encrypted Javascript mailto link
|
|
216
|
$mailto_link = "<a {$class_attr}{$id_attr}href=\"javascript:mdcr('$encrypted_email','$email_subject')\">" .$match[5] ."</a>";
|
|
217
|
return $mailto_link;
|
|
218
|
} else {
|
|
219
|
/** DO NOT USE JAVASCRIPT ENCRYPTION FOR MAILTO LINKS **/
|
|
220
|
// as minimum protection, replace @ in the mailto part by (at)
|
|
221
|
// dots are not transformed as this would transform my.name@domain.com into: my(dot)name(at)domain(dot)com
|
|
222
|
// rebuild the mailto link from the subpatterns (at the missing characters " and </a>")
|
|
223
|
return $match[1] .str_replace('@', OUTPUT_FILTER_AT_REPLACEMENT, $match[2]) .$match[3] .'"' .$match[4] .$match[5] .'</a>';
|
|
224
|
// if you want to protect both, @ and dots, comment out the line above and remove the comment from the line below
|
|
225
|
// return $match[1] .str_replace($search, $replace, $match[2]) .$match[3] .'"' .$match[4] .$match[5] .'</a>';
|
|
226
|
}
|
|
227
|
break;
|
|
228
|
default:
|
|
229
|
// number of subpatterns do not match the requirements ... do nothing
|
|
230
|
return $match[0];
|
|
231
|
break;
|
|
232
|
}
|
|
233
|
}
|
|
234
|
/* ************************************************************************** */
|