|
1
|
|
|
<?php |
|
2
|
|
|
namespace vipnytt\RobotsTxtParser\Client\Directives; |
|
3
|
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Parser\UriParser; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* Class HostClient |
|
8
|
|
|
* |
|
9
|
|
|
* @see https://github.com/VIPnytt/RobotsTxtParser/blob/master/docs/methods/HostClient.md for documentation |
|
10
|
|
|
* @package vipnytt\RobotsTxtParser\Client\Directives |
|
11
|
|
|
*/ |
|
12
|
|
|
class HostClient implements ClientInterface |
|
13
|
|
|
{ |
|
14
|
|
|
/** |
|
15
|
|
|
* Base uri |
|
16
|
|
|
* @var string |
|
17
|
|
|
*/ |
|
18
|
|
|
private $base; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Effective uri |
|
22
|
|
|
* @var string |
|
23
|
|
|
*/ |
|
24
|
|
|
private $effective; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Host values |
|
28
|
|
|
* @var string[] |
|
29
|
|
|
*/ |
|
30
|
|
|
private $host; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* Parent directive |
|
34
|
|
|
* @var string|null |
|
35
|
|
|
*/ |
|
36
|
|
|
private $parent; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* HostClient constructor. |
|
40
|
|
|
* |
|
41
|
|
|
* @param string $base |
|
42
|
|
|
* @param string $effective |
|
43
|
|
|
* @param string[] $host |
|
44
|
|
|
* @param string|null $parentDirective |
|
45
|
|
|
*/ |
|
46
|
|
|
public function __construct($base, $effective, array $host, $parentDirective = null) |
|
47
|
|
|
{ |
|
48
|
|
|
$this->base = $base; |
|
49
|
|
|
$this->effective = $effective; |
|
50
|
|
|
$this->host = $host; |
|
51
|
|
|
$this->parent = $parentDirective; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* Is preferred host |
|
56
|
|
|
* |
|
57
|
|
|
* @return bool |
|
58
|
|
|
*/ |
|
59
|
|
|
public function isPreferred() |
|
60
|
|
|
{ |
|
61
|
|
|
if (($host = $this->get()) === null) { |
|
62
|
|
|
return $this->base === $this->effective; |
|
63
|
|
|
} |
|
64
|
|
|
$parsed = parse_url($host); |
|
65
|
|
|
$new = [ |
|
66
|
|
|
'scheme' => isset($parsed['scheme']) ? $parsed['scheme'] : parse_url($this->base, PHP_URL_SCHEME), |
|
67
|
|
|
'host' => isset($parsed['host']) ? $parsed['host'] : $parsed['path'], |
|
68
|
|
|
]; |
|
69
|
|
|
$new['port'] = isset($parsed['port']) ? $parsed['port'] : getservbyname($new['scheme'], 'tcp'); |
|
70
|
|
|
return $this->base == $new['scheme'] . '://' . $new['host'] . ':' . $new['port']; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* Get Host |
|
75
|
|
|
* |
|
76
|
|
|
* @return string|null |
|
77
|
|
|
*/ |
|
78
|
|
|
public function get() |
|
79
|
|
|
{ |
|
80
|
|
|
return isset($this->host[0]) ? $this->host[0] : null; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* Get Host, falls back to Effective Request URI if not found |
|
85
|
|
|
* |
|
86
|
|
|
* @return string |
|
87
|
|
|
*/ |
|
88
|
|
|
public function getWithFallback() |
|
89
|
|
|
{ |
|
90
|
|
|
if (($get = $this->get()) !== null) { |
|
91
|
|
|
// Host defined by the Host directive |
|
92
|
|
|
return $get; |
|
93
|
|
|
} elseif ( |
|
94
|
|
|
$this->base !== $this->effective && |
|
95
|
|
|
parse_url($this->base, PHP_URL_HOST) === ($host = parse_url($this->effective, PHP_URL_HOST)) |
|
96
|
|
|
) { |
|
97
|
|
|
// Host is the same, but Scheme or Port is different |
|
98
|
|
|
return getservbyname($scheme = parse_url($this->effective, PHP_URL_SCHEME), 'tcp') === parse_url($this->effective, PHP_URL_PORT) ? $scheme . '://' . $host : $this->effective; |
|
99
|
|
|
} |
|
100
|
|
|
// Return Host name only |
|
101
|
|
|
return parse_url($this->effective, PHP_URL_HOST); |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
/** |
|
105
|
|
|
* Is host listed by directive |
|
106
|
|
|
* |
|
107
|
|
|
* @param string $uri |
|
108
|
|
|
* @return bool |
|
109
|
|
|
*/ |
|
110
|
|
|
public function isUriListed($uri) |
|
111
|
|
|
{ |
|
112
|
|
|
$uriParser = new UriParser($uri); |
|
113
|
|
|
$uri = mb_strtolower($uriParser->encode()); |
|
114
|
|
|
$parts = [ |
|
115
|
|
|
'scheme' => parse_url($uri, PHP_URL_SCHEME), |
|
116
|
|
|
'host' => parse_url($uri, PHP_URL_HOST), |
|
117
|
|
|
]; |
|
118
|
|
|
$parts['port'] = is_int($port = parse_url($uri, PHP_URL_PORT)) ? $port : getservbyname($parts['scheme'], 'tcp'); |
|
119
|
|
|
$cases = [ |
|
120
|
|
|
$parts['host'], |
|
121
|
|
|
$parts['host'] . ':' . $parts['port'], |
|
122
|
|
|
$parts['scheme'] . '://' . $parts['host'], |
|
123
|
|
|
$parts['scheme'] . '://' . $parts['host'] . ':' . $parts['port'] |
|
124
|
|
|
]; |
|
125
|
|
|
foreach ($this->host as $host) { |
|
126
|
|
|
if (in_array($host, $cases)) { |
|
127
|
|
|
return true; |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
return false; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* Export |
|
135
|
|
|
* |
|
136
|
|
|
* @return string[]|string|null |
|
137
|
|
|
*/ |
|
138
|
|
|
public function export() |
|
139
|
|
|
{ |
|
140
|
|
|
if ($this->parent === null) { |
|
141
|
|
|
return isset($this->host[0]) ? $this->host[0] : null; |
|
142
|
|
|
} |
|
143
|
|
|
return $this->host; |
|
144
|
|
|
} |
|
145
|
|
|
} |
|
146
|
|
|
|
VIPnytt /
RobotsTxtParser
