|
1
|
|
|
<?php |
|
2
|
|
|
namespace vipnytt\RobotsTxtParser\Parser\Directives; |
|
3
|
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Parser\RobotsTxtInterface; |
|
5
|
|
|
use vipnytt\RobotsTxtParser\Parser\UrlParser; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* Class Host |
|
9
|
|
|
* |
|
10
|
|
|
* @package vipnytt\RobotsTxtParser\Parser\Directives |
|
11
|
|
|
*/ |
|
12
|
|
|
class Host implements DirectiveInterface, RobotsTxtInterface |
|
13
|
|
|
{ |
|
14
|
|
|
use UrlParser; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Directive |
|
18
|
|
|
*/ |
|
19
|
|
|
const DIRECTIVE = self::DIRECTIVE_HOST; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* Host array |
|
23
|
|
|
* @var string[] |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $array = []; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Host constructor. |
|
29
|
|
|
*/ |
|
30
|
|
|
public function __construct() |
|
31
|
|
|
{ |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* Add |
|
36
|
|
|
* |
|
37
|
|
|
* @param string $line |
|
38
|
|
|
* @return bool |
|
39
|
|
|
*/ |
|
40
|
|
|
public function add($line) |
|
41
|
|
|
{ |
|
42
|
|
|
$host = $this->parse($line); |
|
43
|
|
|
if ( |
|
44
|
|
|
$host === false || |
|
45
|
|
|
$line !== $host || |
|
46
|
|
|
in_array($host, $this->array) |
|
47
|
|
|
) { |
|
48
|
|
|
return false; |
|
49
|
|
|
} |
|
50
|
|
|
$this->array[] = $line; |
|
51
|
|
|
return true; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* Parse |
|
56
|
|
|
* |
|
57
|
|
|
* @param string $line |
|
58
|
|
|
* @return string|false |
|
59
|
|
|
*/ |
|
60
|
|
|
protected function parse($line) |
|
61
|
|
|
{ |
|
62
|
|
|
if (($parsed = parse_url(($line = $this->urlEncode(mb_strtolower($line))))) === false) { |
|
63
|
|
|
return false; |
|
64
|
|
|
} |
|
65
|
|
|
$line = isset($parsed['host']) ? $parsed['host'] : $parsed['path']; |
|
66
|
|
|
if ( |
|
67
|
|
|
!$this->urlValidateHost($line) || |
|
68
|
|
|
( |
|
69
|
|
|
isset($parsed['scheme']) && |
|
70
|
|
|
!$this->urlValidateScheme($parsed['scheme']) |
|
71
|
|
|
) |
|
72
|
|
|
) { |
|
73
|
|
|
return false; |
|
74
|
|
|
} |
|
75
|
|
|
$scheme = isset($parsed['scheme']) ? $parsed['scheme'] . '://' : ''; |
|
76
|
|
|
$port = isset($parsed['port']) ? ':' . $parsed['port'] : ''; |
|
77
|
|
|
return $scheme . $line . $port; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* Check |
|
82
|
|
|
* |
|
83
|
|
|
* @param string $url |
|
84
|
|
|
* @return bool |
|
85
|
|
|
*/ |
|
86
|
|
|
public function check($url) |
|
87
|
|
|
{ |
|
88
|
|
|
if (empty($this->array)) { |
|
89
|
|
|
return false; |
|
90
|
|
|
} |
|
91
|
|
|
$url = mb_strtolower($this->urlEncode($url)); |
|
92
|
|
|
$parts = [ |
|
93
|
|
|
'scheme' => parse_url($url, PHP_URL_SCHEME), |
|
94
|
|
|
'host' => parse_url($url, PHP_URL_HOST), |
|
95
|
|
|
]; |
|
96
|
|
|
$parts['port'] = is_int($port = parse_url($url, PHP_URL_PORT)) ? $port : getservbyname($parts['scheme'], 'tcp'); |
|
97
|
|
|
$cases = [ |
|
98
|
|
|
$parts['host'], |
|
99
|
|
|
$parts['host'] . ':' . $parts['port'], |
|
100
|
|
|
$parts['scheme'] . '://' . $parts['host'], |
|
101
|
|
|
$parts['scheme'] . '://' . $parts['host'] . ':' . $parts['port'] |
|
102
|
|
|
]; |
|
103
|
|
|
if (in_array($this->array[0], $cases)) { |
|
104
|
|
|
return true; |
|
105
|
|
|
} |
|
106
|
|
|
return false; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* Export rules |
|
111
|
|
|
* |
|
112
|
|
|
* @return string[][] |
|
113
|
|
|
*/ |
|
114
|
|
|
public function export() |
|
115
|
|
|
{ |
|
116
|
|
|
return empty($this->array) ? [] : [self::DIRECTIVE => $this->array]; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Render |
|
121
|
|
|
* |
|
122
|
|
|
* @return string[] |
|
123
|
|
|
*/ |
|
124
|
|
|
public function render() |
|
125
|
|
|
{ |
|
126
|
|
|
$result = []; |
|
127
|
|
|
foreach ($this->array as $value) { |
|
128
|
|
|
$result[] = self::DIRECTIVE . ':' . $value; |
|
129
|
|
|
} |
|
130
|
|
|
return $result; |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
|
VIPnytt /
RobotsTxtParser
