1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* vipnytt/RobotsTxtParser |
4
|
|
|
* |
5
|
|
|
* @link https://github.com/VIPnytt/RobotsTxtParser |
6
|
|
|
* @license https://github.com/VIPnytt/RobotsTxtParser/blob/master/LICENSE The MIT License (MIT) |
7
|
|
|
*/ |
8
|
|
|
|
9
|
|
|
namespace vipnytt\RobotsTxtParser\Parser\Directives; |
10
|
|
|
|
11
|
|
|
use vipnytt\RobotsTxtParser\Client\Directives\AllowClient; |
12
|
|
|
use vipnytt\RobotsTxtParser\Handler\RenderHandler; |
13
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* Class AllowParser |
17
|
|
|
* |
18
|
|
|
* @package vipnytt\RobotsTxtParser\Parser\Directives |
19
|
|
|
*/ |
20
|
|
|
class AllowParser implements ParserInterface, RobotsTxtInterface |
21
|
|
|
{ |
22
|
|
|
/** |
23
|
|
|
* Directive |
24
|
|
|
* @var string |
25
|
|
|
*/ |
26
|
|
|
private $directive; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* Path |
30
|
|
|
* @var array |
31
|
|
|
*/ |
32
|
|
|
private $path = []; |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* Sort result |
36
|
|
|
* @var bool |
37
|
|
|
*/ |
38
|
|
|
private $sort = false; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* AllowParser constructor |
42
|
|
|
* |
43
|
|
|
* @param string $directive |
44
|
|
|
*/ |
45
|
|
|
public function __construct($directive) |
46
|
|
|
{ |
47
|
|
|
$this->directive = $directive; |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* Add |
52
|
|
|
* |
53
|
|
|
* @param string $line |
54
|
|
|
* @return bool |
55
|
|
|
*/ |
56
|
|
|
public function add($line) |
57
|
|
|
{ |
58
|
|
|
$line = $this->normalize($line); |
59
|
|
|
if (substr($line, 0, 1) == '/' && |
60
|
|
|
!in_array($line, $this->path) |
61
|
|
|
) { |
62
|
|
|
$this->path[] = $line; |
63
|
|
|
return true; |
64
|
|
|
} |
65
|
|
|
return false; |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* Normalize rules |
70
|
|
|
* |
71
|
|
|
* @param $line |
72
|
|
|
* @return string |
73
|
|
|
*/ |
74
|
|
|
private function normalize($line) |
75
|
|
|
{ |
76
|
|
|
// Prepend slash if starting with an wildcard |
77
|
|
|
if (substr($line, 0, 1) == '*') { |
78
|
|
|
$line = '/' . $line; |
79
|
|
|
} |
80
|
|
|
// Remove unnecessary characters after an end anchor |
81
|
|
|
if (($pos = mb_strpos($line, '$')) !== false) { |
82
|
|
|
$line = mb_substr($line, 0, $pos + 1); |
83
|
|
|
} |
84
|
|
|
// Remove unnecessary wildcards |
85
|
|
|
$line = rtrim($line, '*'); |
86
|
|
|
return $line; |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* Render |
91
|
|
|
* |
92
|
|
|
* @param RenderHandler $handler |
93
|
|
|
* @return bool |
94
|
|
|
*/ |
95
|
|
|
public function render(RenderHandler $handler) |
96
|
|
|
{ |
97
|
|
|
if ($this->directive === self::DIRECTIVE_DISALLOW && |
98
|
|
|
count($this->path) === 0 && |
99
|
|
|
$handler->getLevel() == 2 |
100
|
|
|
) { |
101
|
|
|
$handler->add($this->directive, ''); |
102
|
|
|
return true; |
103
|
|
|
} |
104
|
|
|
$this->sort(); |
105
|
|
|
foreach ($this->path as $path) { |
106
|
|
|
$handler->add($this->directive, $path); |
107
|
|
|
} |
108
|
|
|
return true; |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* Sort by length |
113
|
|
|
* |
114
|
|
|
* @return bool |
115
|
|
|
*/ |
116
|
|
|
private function sort() |
117
|
|
|
{ |
118
|
|
|
if ($this->sort) { |
119
|
|
|
return $this->sort; |
120
|
|
|
}; |
121
|
|
|
return $this->sort = rsort($this->path) && usort($this->path, function ($a, $b) { |
122
|
|
|
// PHP 7: Switch to the <=> "Spaceship" operator |
123
|
|
|
return mb_strlen($a) - mb_strlen($b); |
124
|
|
|
}); |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* Client |
129
|
|
|
* |
130
|
|
|
* @return AllowClient |
131
|
|
|
*/ |
132
|
|
|
public function client() |
133
|
|
|
{ |
134
|
|
|
$this->sort(); |
135
|
|
|
return new AllowClient($this->path); |
136
|
|
|
} |
137
|
|
|
} |
138
|
|
|
|