|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* This file is part of the bisarca/robots-txt package. |
|
5
|
|
|
* |
|
6
|
|
|
* (c) Emanuele Minotto <[email protected]> |
|
7
|
|
|
* |
|
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
9
|
|
|
* file that was distributed with this source code. |
|
10
|
|
|
*/ |
|
11
|
|
|
|
|
12
|
|
|
namespace Bisarca\RobotsTxt\Directive; |
|
13
|
|
|
|
|
14
|
|
|
use Bisarca\RobotsTxt\Exception\InvalidDirectiveException; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* "User-agent" directive element. |
|
18
|
|
|
*/ |
|
19
|
|
|
class UserAgent implements StartOfGroupInterface, GroupMemberInterface |
|
20
|
|
|
{ |
|
21
|
|
|
/** |
|
22
|
|
|
* When the asterisk is used, than all the user-agents are included. |
|
23
|
|
|
* |
|
24
|
|
|
* @var string |
|
25
|
|
|
*/ |
|
26
|
|
|
const ALL_AGENTS = '*'; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Directive value. |
|
30
|
|
|
* |
|
31
|
|
|
* @var string |
|
32
|
|
|
*/ |
|
33
|
|
|
private $value = ''; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* {@inheritdoc} |
|
37
|
|
|
*/ |
|
38
|
|
|
public function __construct(string $raw) |
|
39
|
|
|
{ |
|
40
|
|
|
if (!preg_match('/^user-agent:\s*([^# ]+).*/i', $raw, $matches)) { |
|
41
|
|
|
throw InvalidDirectiveException::create($raw); |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
$this->value = trim($matches[1]); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* {@inheritdoc} |
|
49
|
|
|
*/ |
|
50
|
|
|
public static function getField(): string |
|
51
|
|
|
{ |
|
52
|
|
|
return 'user-agent'; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* {@inheritdoc} |
|
57
|
|
|
*/ |
|
58
|
|
|
public function getValue(): string |
|
59
|
|
|
{ |
|
60
|
|
|
return $this->value; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* {@inheritdoc} |
|
65
|
|
|
*/ |
|
66
|
|
|
public function __toString(): string |
|
67
|
|
|
{ |
|
68
|
|
|
return sprintf('User-agent: %s', $this->value); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Checks if a user-agent is matching current user agent. |
|
73
|
|
|
* |
|
74
|
|
|
* @param string $userAgent |
|
75
|
|
|
* |
|
76
|
|
|
* @return bool |
|
77
|
|
|
*/ |
|
78
|
|
|
public function isMatching(string $userAgent): bool |
|
79
|
|
|
{ |
|
80
|
|
|
// "*" matches any sequence of characters (zero or more) |
|
81
|
|
|
$value = str_replace('*', '.*', $this->value); |
|
82
|
|
|
|
|
83
|
|
|
return 1 === preg_match('#^'.$value.'#', trim($userAgent)); |
|
84
|
|
|
} |
|
85
|
|
|
} |
|
86
|
|
|
|
bisarca /
robots-txt
