|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Bee4\RobotsTxt; |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* Class Rule |
|
7
|
|
|
* Represent a Ruleset inside a Robots.txt file |
|
8
|
|
|
* |
|
9
|
|
|
* @copyright Bee4 2015 |
|
10
|
|
|
* @author Stephane HULARD <[email protected]> |
|
11
|
|
|
*/ |
|
12
|
|
|
class Rule |
|
13
|
|
|
{ |
|
14
|
|
|
const COMPILED = 'compiled'; |
|
15
|
|
|
const DIRTY = 'dirty'; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* User agent on which the rule apply |
|
19
|
|
|
* @var string |
|
20
|
|
|
*/ |
|
21
|
|
|
private $ua; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* Rule status (compiled or dirty) |
|
25
|
|
|
* @var string |
|
26
|
|
|
*/ |
|
27
|
|
|
private $state; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* Expression collection with allow / disallow segments |
|
31
|
|
|
* @var array |
|
32
|
|
|
*/ |
|
33
|
|
|
private $exp = [ |
|
34
|
|
|
'allow' => [], |
|
35
|
|
|
'disallow' => [] |
|
36
|
|
|
]; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Compiled regex pattern with allow / disallow segments |
|
40
|
|
|
* @var array |
|
41
|
|
|
*/ |
|
42
|
|
|
private $patterns = [ |
|
43
|
|
|
'allow' => '', |
|
44
|
|
|
'disallow' => '' |
|
45
|
|
|
]; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* @param string $ua |
|
49
|
|
|
*/ |
|
50
|
|
|
public function __construct($ua) |
|
51
|
|
|
{ |
|
52
|
1 |
|
$this->ua = $ua; |
|
53
|
1 |
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* Retrieve rule's UserAgent |
|
57
|
|
|
* @return string |
|
58
|
|
|
*/ |
|
59
|
|
|
public function getUserAgent() |
|
60
|
|
|
{ |
|
61
|
1 |
|
return $this->ua; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* Add a pattern to match in the current rule by allowing |
|
66
|
|
|
* @param string $pattern |
|
67
|
|
|
* @return Rule |
|
68
|
|
|
*/ |
|
69
|
|
|
public function allow($pattern) |
|
70
|
|
|
{ |
|
71
|
1 |
|
return $this->addExpression(new Expression($pattern), 'allow'); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* Add a pattern to match in the current rule by disallowing |
|
76
|
|
|
* @param string $pattern |
|
77
|
|
|
* @return Rule |
|
78
|
|
|
*/ |
|
79
|
|
|
public function disallow($pattern) |
|
80
|
|
|
{ |
|
81
|
1 |
|
return $this->addExpression(new Expression($pattern), 'disallow'); |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* Add an expression in the current rule |
|
86
|
|
|
* @param Expression $exp Expression to be added |
|
87
|
|
|
* @param string $mode Expression mode (allow / disallow) |
|
88
|
|
|
* @return Rule |
|
89
|
|
|
*/ |
|
90
|
|
|
private function addExpression(Expression $exp, $mode) |
|
91
|
|
|
{ |
|
92
|
1 |
|
$this->state = self::DIRTY; |
|
93
|
1 |
|
$this->exp[$mode][] = $exp; |
|
94
|
1 |
|
return $this; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* Compile expressions to a global pattern |
|
99
|
|
|
* @return boolean|null |
|
100
|
|
|
*/ |
|
101
|
|
|
private function compile() |
|
102
|
|
|
{ |
|
103
|
1 |
|
if (self::COMPILED === $this->state) { |
|
104
|
1 |
|
return true; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
1 |
|
$process = function (array &$patterns) { |
|
108
|
1 |
|
usort($patterns, function (Expression $a, Expression $b) { |
|
109
|
1 |
|
return strlen($a->getRaw()) < strlen($b->getRaw()); |
|
110
|
1 |
|
}); |
|
111
|
|
|
|
|
112
|
1 |
|
return '/^(('.implode(')|(', $patterns).'))$/'; |
|
113
|
1 |
|
}; |
|
114
|
1 |
|
$this->patterns['allow'] = $process($this->exp['allow']); |
|
115
|
1 |
|
$this->patterns['disallow'] = $process($this->exp['disallow']); |
|
116
|
1 |
|
$this->state = self::COMPILED; |
|
117
|
1 |
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Check if the URL is allowed or not |
|
121
|
|
|
* @param string $url |
|
122
|
|
|
* @return boolean |
|
123
|
|
|
*/ |
|
124
|
|
|
public function match($url) |
|
125
|
|
|
{ |
|
126
|
1 |
|
$this->compile(); |
|
127
|
|
|
|
|
128
|
1 |
|
if (0 < count($this->exp['disallow']) && |
|
129
|
1 |
|
1 === preg_match($this->patterns['disallow'], $url, $disallowed) ) { |
|
130
|
1 |
|
if (0 < count($this->exp['allow']) && |
|
131
|
1 |
|
1 === preg_match($this->patterns['allow'], $url, $allowed) |
|
132
|
|
|
) { |
|
133
|
1 |
|
$a = $this->lastFilledIndex($allowed); |
|
134
|
1 |
|
$d = $this->lastFilledIndex($disallowed); |
|
135
|
|
|
return |
|
136
|
1 |
|
strlen($this->exp['allow'][$a-2]->getRaw()) >= |
|
137
|
1 |
|
strlen($this->exp['disallow'][$d-2]->getRaw()); |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
1 |
|
return false; |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
1 |
|
return true; |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
/** |
|
147
|
|
|
* Retrieve the last filled index in a given array |
|
148
|
|
|
* @param array $data |
|
149
|
|
|
* @return integer |
|
150
|
|
|
*/ |
|
151
|
|
|
private function lastFilledIndex(array $data) |
|
152
|
|
|
{ |
|
153
|
1 |
|
return key(array_slice(array_filter($data), -1, 1, true)); |
|
154
|
|
|
} |
|
155
|
|
|
} |
|
156
|
|
|
|
bee4 /
robots.txt
