|
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\Parser\UriParser; |
|
12
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Class HostParser |
|
16
|
|
|
* |
|
17
|
|
|
* @link http://tools.ietf.org/html/rfc952 |
|
18
|
|
|
* |
|
19
|
|
|
* @package vipnytt\RobotsTxtParser\Parser\Directives |
|
20
|
|
|
*/ |
|
21
|
|
|
abstract class HostParserCore implements ParserInterface, RobotsTxtInterface |
|
22
|
|
|
{ |
|
23
|
|
|
/** |
|
24
|
|
|
* Base uri |
|
25
|
|
|
* @var string |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $base; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* Effective uri |
|
31
|
|
|
* @var string |
|
32
|
|
|
*/ |
|
33
|
|
|
protected $effective; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Host values |
|
37
|
|
|
* @var string[] |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $host = []; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* HostParser constructor. |
|
43
|
|
|
* |
|
44
|
|
|
* @param string $base |
|
45
|
|
|
* @param string $effective |
|
46
|
|
|
*/ |
|
47
|
|
|
public function __construct($base, $effective) |
|
48
|
|
|
{ |
|
49
|
|
|
$this->base = $base; |
|
50
|
|
|
$this->effective = $effective; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* Add |
|
55
|
|
|
* |
|
56
|
|
|
* @param string $line |
|
57
|
|
|
* @return bool |
|
58
|
|
|
*/ |
|
59
|
|
|
public function add($line) |
|
60
|
|
|
{ |
|
61
|
|
|
$host = $this->parse($line); |
|
62
|
|
|
if ($host === false || |
|
63
|
|
|
$host !== $line || |
|
64
|
|
|
in_array($host, $this->host) |
|
65
|
|
|
) { |
|
66
|
|
|
return false; |
|
67
|
|
|
} |
|
68
|
|
|
$this->host[] = $line; |
|
69
|
|
|
return true; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* Parse |
|
74
|
|
|
* |
|
75
|
|
|
* @param string $line |
|
76
|
|
|
* @return string|false |
|
77
|
|
|
*/ |
|
78
|
|
|
private function parse($line) |
|
79
|
|
|
{ |
|
80
|
|
|
$uriParser = new UriParser($line); |
|
81
|
|
|
$line = $uriParser->encode(); |
|
82
|
|
|
if ($uriParser->validateIP() || |
|
83
|
|
|
!$uriParser->validateHost() || |
|
84
|
|
|
( |
|
85
|
|
|
parse_url($line, PHP_URL_SCHEME) !== null && |
|
86
|
|
|
!$uriParser->validateScheme() |
|
87
|
|
|
) |
|
88
|
|
|
) { |
|
89
|
|
|
return false; |
|
90
|
|
|
} |
|
91
|
|
|
$parts = $this->getParts($line); |
|
92
|
|
|
return $parts['scheme'] . $parts['host'] . $parts['port']; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* Get URI parts |
|
97
|
|
|
* |
|
98
|
|
|
* @param string $uri |
|
99
|
|
|
* @return string[]|false |
|
100
|
|
|
*/ |
|
101
|
|
|
private function getParts($uri) |
|
102
|
|
|
{ |
|
103
|
|
|
return ($parsed = parse_url($uri)) === false ? false : [ |
|
104
|
|
|
'scheme' => isset($parsed['scheme']) ? $parsed['scheme'] . '://' : '', |
|
105
|
|
|
'host' => isset($parsed['host']) ? $parsed['host'] : $parsed['path'], |
|
106
|
|
|
'port' => isset($parsed['port']) ? ':' . $parsed['port'] : '', |
|
107
|
|
|
]; |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
|
|
|