|
1
|
|
|
<?php |
|
2
|
|
|
namespace Limonte; |
|
3
|
|
|
|
|
4
|
|
|
class AdblockRule |
|
5
|
|
|
{ |
|
6
|
|
|
private $rule; |
|
7
|
|
|
|
|
8
|
|
|
private $regex; |
|
9
|
|
|
|
|
10
|
|
|
private $isComment = false; |
|
11
|
|
|
|
|
12
|
|
|
private $isHtml = false; |
|
13
|
|
|
|
|
14
|
|
|
private $isException = false; |
|
15
|
|
|
|
|
16
|
|
|
public function __construct($rule) |
|
17
|
|
|
{ |
|
18
|
|
|
$this->rule = $rule; |
|
19
|
|
|
|
|
20
|
|
|
if ($this->startsWith($this->rule, '@@')) { |
|
21
|
|
|
$this->isException = true; |
|
22
|
|
|
$this->rule = mb_substr($this->rule, 2); |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
// comment |
|
26
|
|
|
if ($this->startsWith($rule, '!') || $this->startsWith($rule, '[Adblock')) { |
|
27
|
|
|
$this->isComment = true; |
|
28
|
|
|
|
|
29
|
|
|
// HTML rule |
|
30
|
|
|
} elseif ($this->contains($rule, '##') || $this->contains($rule, '#@#')) { |
|
31
|
|
|
$this->isHtml = true; |
|
32
|
|
|
|
|
33
|
|
|
// URI rule |
|
34
|
|
|
} else { |
|
35
|
|
|
$this->makeRegex(); |
|
36
|
|
|
} |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @param string $url |
|
41
|
|
|
* |
|
42
|
|
|
* @return boolean |
|
43
|
|
|
*/ |
|
44
|
|
|
public function matchUrl($url) |
|
45
|
|
|
{ |
|
46
|
|
|
return (boolean)preg_match( |
|
47
|
|
|
'/' . $this->getRegex() . '/', |
|
48
|
|
|
$url |
|
49
|
|
|
); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* @return string |
|
54
|
|
|
*/ |
|
55
|
|
|
public function getRegex() |
|
56
|
|
|
{ |
|
57
|
|
|
return $this->regex; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
/** |
|
61
|
|
|
* @return boolean |
|
62
|
|
|
*/ |
|
63
|
|
|
public function isComment() |
|
64
|
|
|
{ |
|
65
|
|
|
return $this->isComment; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* @return boolean |
|
70
|
|
|
*/ |
|
71
|
|
|
public function isHtml() |
|
72
|
|
|
{ |
|
73
|
|
|
return $this->isHtml; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* @return boolean |
|
78
|
|
|
*/ |
|
79
|
|
|
public function isException() |
|
80
|
|
|
{ |
|
81
|
|
|
return $this->isException; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
private function makeRegex() |
|
85
|
|
|
{ |
|
86
|
|
|
if (empty($this->rule)) { |
|
87
|
|
|
throw new InvalidRuleException("Empty rule"); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
$regex = $this->rule; |
|
91
|
|
|
|
|
92
|
|
|
// Check if the rule isn't already regexp |
|
93
|
|
|
if ($this->startsWith($regex, '/') && $this->endsWith($regex, '/')) { |
|
94
|
|
|
$this->regex = mb_substr($this->rule, 1, mb_strlen($this->rule) - 2); |
|
95
|
|
|
|
|
96
|
|
|
if (empty($this->regex)) { |
|
97
|
|
|
throw new InvalidRuleException("Empty rule"); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
return; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
// escape special regex characters |
|
104
|
|
|
$regex = preg_replace("/([\\\.\$\+\?\{\}\(\)\[\]\/])/", "\\\\$1", $this->rule); |
|
105
|
|
|
|
|
106
|
|
|
// Separator character ^ matches anything but a letter, a digit, or |
|
107
|
|
|
// one of the following: _ - . %. The end of the address is also |
|
108
|
|
|
// accepted as separator. |
|
109
|
|
|
$regex = str_replace("^", "([^\w\d_\-.%]|$)", $regex); |
|
110
|
|
|
|
|
111
|
|
|
// * symbol |
|
112
|
|
|
$regex = str_replace("*", ".*", $regex); |
|
113
|
|
|
|
|
114
|
|
|
// | in the end means the end of the address |
|
115
|
|
|
if ($this->endsWith($regex, '|')) { |
|
116
|
|
|
$regex = mb_substr($regex, 0, mb_strlen($regex) - 1) . '$'; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
// || in the beginning means beginning of the domain name |
|
120
|
|
|
if ($this->startsWith($regex, '||')) { |
|
121
|
|
|
if (mb_strlen($regex) > 2) { |
|
122
|
|
|
// http://tools.ietf.org/html/rfc3986#appendix-B |
|
123
|
|
|
$regex = "^([^:\/?#]+:)?(\/\/([^\/?#]*\.)?)?" . mb_substr($regex, 2); |
|
124
|
|
|
} |
|
125
|
|
|
// | in the beginning means start of the address |
|
126
|
|
|
} elseif ($this->startsWith($regex, '|')) { |
|
127
|
|
|
$regex = '^' . mb_substr($regex, 1); |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
// other | symbols should be escaped |
|
131
|
|
|
$regex = preg_replace("/\|(?![\$])/", "\|$1", $regex); |
|
132
|
|
|
|
|
133
|
|
|
$this->regex = $regex; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
private function startsWith($haystack, $needle) |
|
137
|
|
|
{ |
|
138
|
|
|
$length = mb_strlen($needle); |
|
139
|
|
|
return (mb_substr($haystack, 0, $length) === $needle); |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
private function endsWith($haystack, $needle) |
|
143
|
|
|
{ |
|
144
|
|
|
$length = mb_strlen($needle); |
|
145
|
|
|
if ($length == 0) { |
|
146
|
|
|
return true; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
return (mb_substr($haystack, -$length) === $needle); |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
private function contains($haystack, $needle) |
|
153
|
|
|
{ |
|
154
|
|
|
return strpos($haystack, $needle) !== false; |
|
155
|
|
|
} |
|
156
|
|
|
} |
|
157
|
|
|
|
limonte /
php-adblock-parser
