|
1
|
|
|
<?php |
|
2
|
|
|
namespace Limonte; |
|
3
|
|
|
|
|
4
|
|
|
class AdblockParser |
|
5
|
|
|
{ |
|
6
|
|
|
/** |
|
7
|
|
|
* @var AdblockRule[] |
|
8
|
|
|
*/ |
|
9
|
|
|
private $rules; |
|
10
|
|
|
|
|
11
|
|
|
private $cacheFolder; |
|
12
|
10 |
|
|
|
13
|
|
|
private $cacheExpire = 1; // 1 day |
|
14
|
10 |
|
|
|
15
|
10 |
|
public function __construct(array $rules = []) |
|
16
|
10 |
|
{ |
|
17
|
|
|
$this->initRules($rules); |
|
18
|
|
|
} |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
10 |
|
* @param array $rules |
|
22
|
|
|
*/ |
|
23
|
10 |
|
public function initRules(array $rules = []) |
|
24
|
|
|
{ |
|
25
|
9 |
|
$this->rules = []; |
|
26
|
9 |
|
$this->addRules($rules); |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @param string[] $rules |
|
31
|
|
|
*/ |
|
32
|
10 |
|
public function addRules($rules) |
|
33
|
4 |
|
{ |
|
34
|
10 |
|
foreach ($rules as $rule) { |
|
35
|
10 |
|
try { |
|
36
|
|
|
$this->rules[] = new AdblockRule($rule); |
|
37
|
|
|
} catch (InvalidRuleException $e) { |
|
38
|
|
|
// Skip invalid rules |
|
39
|
|
|
} |
|
40
|
3 |
|
} |
|
41
|
|
|
|
|
42
|
|
|
// Sort rules, exceptions first |
|
43
|
3 |
|
usort($this->rules, function (AdblockRule $a, AdblockRule $b) { |
|
44
|
3 |
|
return (int)$a->isException() < (int)$b->isException(); |
|
45
|
2 |
|
}); |
|
46
|
|
|
} |
|
47
|
1 |
|
|
|
48
|
|
|
/** |
|
49
|
3 |
|
* @param string|array $path |
|
50
|
3 |
|
*/ |
|
51
|
3 |
|
public function loadRules($path) |
|
52
|
|
|
{ |
|
53
|
|
|
// single resource |
|
54
|
2 |
|
if (is_string($path)) { |
|
55
|
2 |
|
if (filter_var($path, FILTER_VALIDATE_URL)) { |
|
56
|
2 |
|
$content = $this->getCachedResource($path); |
|
57
|
|
|
} else { |
|
58
|
|
|
$content = @file_get_contents($path); |
|
59
|
3 |
|
} |
|
60
|
|
|
if ($content) { |
|
61
|
|
|
$rules = preg_split("/(\r\n|\n|\r)/", $content); |
|
62
|
|
|
$this->addRules($rules); |
|
63
|
|
|
} |
|
64
|
1 |
|
// array of resources |
|
65
|
|
|
} elseif (is_array($path)) { |
|
66
|
1 |
|
foreach ($path as $item) { |
|
67
|
|
|
$this->loadRules($item); |
|
68
|
|
|
} |
|
69
|
|
|
} |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* @return array |
|
74
|
10 |
|
*/ |
|
75
|
|
|
public function getRules() |
|
76
|
10 |
|
{ |
|
77
|
|
|
return $this->rules; |
|
78
|
10 |
|
} |
|
79
|
1 |
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* @param string $url |
|
82
|
9 |
|
* @return string[] |
|
83
|
9 |
|
* @throws \Exception |
|
84
|
4 |
|
*/ |
|
85
|
|
|
public function shouldBlock($url) |
|
86
|
|
|
{ |
|
87
|
9 |
|
$rules = []; |
|
88
|
9 |
|
$url = trim($url); |
|
89
|
2 |
|
|
|
90
|
|
|
if (filter_var($url, FILTER_VALIDATE_URL) === false) { |
|
91
|
9 |
|
throw new \Exception("Invalid URL"); |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
foreach ($this->rules as $rule) { |
|
95
|
6 |
|
if ($rule->isComment() || $rule->isHtml() || $rule->isException()) { |
|
96
|
|
|
continue; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
if ($rule->matchUrl($url)) { |
|
100
|
|
|
$rules[] = $rule->getRule(); |
|
101
|
|
|
} |
|
102
|
|
|
} |
|
103
|
7 |
|
|
|
104
|
|
|
return $rules; |
|
105
|
7 |
|
} |
|
106
|
|
|
|
|
107
|
|
|
/** |
|
108
|
|
|
* @param string $url |
|
109
|
|
|
* |
|
110
|
|
|
* @return boolean |
|
111
|
|
|
*/ |
|
112
|
|
|
public function shouldNotBlock($url) |
|
113
|
1 |
|
{ |
|
114
|
|
|
return !$this->shouldBlock($url); |
|
115
|
1 |
|
} |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* Get cache folder |
|
119
|
|
|
* |
|
120
|
|
|
* @return string |
|
121
|
|
|
*/ |
|
122
|
|
|
public function getCacheFolder() |
|
123
|
1 |
|
{ |
|
124
|
|
|
return $this->cacheFolder; |
|
125
|
1 |
|
} |
|
126
|
1 |
|
|
|
127
|
|
|
/** |
|
128
|
|
|
* Set cache folder |
|
129
|
|
|
* |
|
130
|
|
|
* @param string $cacheFolder |
|
131
|
|
|
*/ |
|
132
|
|
|
public function setCacheFolder($cacheFolder) |
|
133
|
1 |
|
{ |
|
134
|
|
|
$this->cacheFolder = rtrim($cacheFolder, DIRECTORY_SEPARATOR) . DIRECTORY_SEPARATOR; |
|
135
|
1 |
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* Get cache expire (in days) |
|
139
|
|
|
* |
|
140
|
|
|
* @return integer |
|
141
|
|
|
*/ |
|
142
|
|
|
public function getCacheExpire() |
|
143
|
1 |
|
{ |
|
144
|
|
|
return $this->cacheExpire; |
|
145
|
1 |
|
} |
|
146
|
1 |
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Set cache expire (in days) |
|
149
|
|
|
* |
|
150
|
|
|
* @param integer $expireInDays |
|
151
|
1 |
|
*/ |
|
152
|
|
|
public function setCacheExpire($expireInDays) |
|
153
|
1 |
|
{ |
|
154
|
1 |
|
$this->cacheExpire = $expireInDays; |
|
155
|
1 |
|
} |
|
156
|
|
|
|
|
157
|
|
|
/** |
|
158
|
1 |
|
* Clear external resources cache |
|
159
|
|
|
*/ |
|
160
|
|
|
public function clearCache() |
|
161
|
|
|
{ |
|
162
|
|
|
if ($this->cacheFolder) { |
|
163
|
|
|
foreach (glob($this->cacheFolder . '*') as $file) { |
|
164
|
|
|
unlink($file); |
|
165
|
2 |
|
} |
|
166
|
|
|
} |
|
167
|
2 |
|
} |
|
168
|
1 |
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* @param string $url |
|
171
|
1 |
|
* |
|
172
|
|
|
* @return string |
|
173
|
1 |
|
*/ |
|
174
|
|
|
private function getCachedResource($url) |
|
175
|
|
|
{ |
|
176
|
1 |
|
if (!$this->cacheFolder) { |
|
177
|
|
|
return @file_get_contents($url); |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
1 |
|
$cacheFile = $this->cacheFolder . basename($url) . md5($url); |
|
181
|
1 |
|
|
|
182
|
1 |
|
if (file_exists($cacheFile) && (filemtime($cacheFile) > (time() - 60 * 24 * $this->cacheExpire))) { |
|
183
|
|
|
// Cache file is less than five minutes old. |
|
184
|
|
|
// Don't bother refreshing, just use the file as-is. |
|
185
|
|
|
$content = @file_get_contents($cacheFile); |
|
186
|
1 |
|
} else { |
|
187
|
|
|
// Our cache is out-of-date, so load the data from our remote server, |
|
188
|
|
|
// and also save it over our cache for next time. |
|
189
|
|
|
$content = @file_get_contents($url); |
|
190
|
|
|
if ($content) { |
|
191
|
|
|
file_put_contents($cacheFile, $content, LOCK_EX); |
|
192
|
|
|
} |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
return $content; |
|
196
|
|
|
} |
|
197
|
|
|
} |
|
198
|
|
|
|
limonte /
php-adblock-parser
