|
1
|
|
|
<?php |
|
2
|
|
|
namespace vipnytt\RobotsTxtParser\Client\Directives; |
|
3
|
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Exceptions\ClientException; |
|
5
|
|
|
use vipnytt\RobotsTxtParser\Parser\UriParser; |
|
6
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* Class AllowClient |
|
10
|
|
|
* |
|
11
|
|
|
* @see https://github.com/VIPnytt/RobotsTxtParser/blob/master/docs/methods/AllowClient.md for documentation |
|
12
|
|
|
* @package vipnytt\RobotsTxtParser\Client\Directives |
|
13
|
|
|
*/ |
|
14
|
|
|
class AllowClient implements ClientInterface, RobotsTxtInterface |
|
15
|
|
|
{ |
|
16
|
|
|
use DirectiveClientCommons; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Paths |
|
20
|
|
|
* @var array |
|
21
|
|
|
*/ |
|
22
|
|
|
private $paths; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Host |
|
26
|
|
|
* @var HostClient |
|
27
|
|
|
*/ |
|
28
|
|
|
private $host; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* Clean-param |
|
32
|
|
|
* @var CleanParamClient |
|
33
|
|
|
*/ |
|
34
|
|
|
private $cleanParam; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* AllowClient constructor. |
|
38
|
|
|
* |
|
39
|
|
|
* @param array $paths |
|
40
|
|
|
* @param HostClient $host |
|
41
|
|
|
* @param CleanParamClient $cleanParam |
|
42
|
|
|
*/ |
|
43
|
|
|
public function __construct(array $paths, HostClient $host, CleanParamClient $cleanParam) |
|
44
|
|
|
{ |
|
45
|
|
|
$this->host = $host; |
|
46
|
|
|
$this->paths = $paths; |
|
47
|
|
|
$this->cleanParam = $cleanParam; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Check |
|
52
|
|
|
* |
|
53
|
|
|
* @param string $uri |
|
54
|
|
|
* @return bool |
|
55
|
|
|
*/ |
|
56
|
|
|
public function isListed($uri) |
|
57
|
|
|
{ |
|
58
|
|
|
$path = $this->getPath($uri); |
|
59
|
|
|
return ( |
|
60
|
|
|
$this->checkPaths($path, $this->paths) || |
|
61
|
|
|
$this->isHostListed($uri, $this->host->export()) || |
|
|
|
|
|
|
62
|
|
|
!empty($this->cleanParam->detect($path)) |
|
63
|
|
|
); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* Get path and query |
|
68
|
|
|
* |
|
69
|
|
|
* @param string $uri |
|
70
|
|
|
* @return string |
|
71
|
|
|
* @throws ClientException |
|
72
|
|
|
*/ |
|
73
|
|
|
private function getPath($uri) |
|
74
|
|
|
{ |
|
75
|
|
|
$uriParser = new UriParser($uri); |
|
76
|
|
|
// Encode |
|
77
|
|
|
$uri = explode('#', $uriParser->encode(), 2)[0]; |
|
78
|
|
|
if (mb_strpos($uri, '/') === 0) { |
|
79
|
|
|
// URI is already an path |
|
80
|
|
|
return $uri; |
|
81
|
|
|
} |
|
82
|
|
|
if (!$uriParser->validate()) { |
|
83
|
|
|
throw new ClientException('Invalid URI'); |
|
84
|
|
|
} |
|
85
|
|
|
$path = (($path = parse_url($uri, PHP_URL_PATH)) === null) ? '/' : $path; |
|
86
|
|
|
$query = (($query = parse_url($uri, PHP_URL_QUERY)) === null) ? '' : '?' . $query; |
|
87
|
|
|
return $path . $query; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* Is host listed by directive |
|
92
|
|
|
* |
|
93
|
|
|
* @param string $uri |
|
94
|
|
|
* @param string[] $hosts |
|
95
|
|
|
* @return bool |
|
96
|
|
|
*/ |
|
97
|
|
|
private function isHostListed($uri, $hosts) |
|
98
|
|
|
{ |
|
99
|
|
|
$uriParser = new UriParser($uri); |
|
100
|
|
|
$uri = $uriParser->encode(); |
|
101
|
|
|
$parts = [ |
|
102
|
|
|
'scheme' => parse_url($uri, PHP_URL_SCHEME), |
|
103
|
|
|
'host' => parse_url($uri, PHP_URL_HOST), |
|
104
|
|
|
]; |
|
105
|
|
|
$parts['port'] = is_int($port = parse_url($uri, PHP_URL_PORT)) ? $port : getservbyname($parts['scheme'], 'tcp'); |
|
106
|
|
|
$cases = [ |
|
107
|
|
|
$parts['host'], |
|
108
|
|
|
$parts['host'] . ':' . $parts['port'], |
|
109
|
|
|
$parts['scheme'] . '://' . $parts['host'], |
|
110
|
|
|
$parts['scheme'] . '://' . $parts['host'] . ':' . $parts['port'] |
|
111
|
|
|
]; |
|
112
|
|
|
foreach ($hosts as $host) { |
|
113
|
|
|
if (in_array($host, $cases)) { |
|
114
|
|
|
return true; |
|
115
|
|
|
} |
|
116
|
|
|
} |
|
117
|
|
|
return false; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* Export |
|
122
|
|
|
* |
|
123
|
|
|
* @return array |
|
124
|
|
|
*/ |
|
125
|
|
|
public function export() |
|
126
|
|
|
{ |
|
127
|
|
|
return [ |
|
128
|
|
|
'host' => $this->host->export(), |
|
129
|
|
|
'path' => $this->paths, |
|
130
|
|
|
'clean-param' => $this->cleanParam->export(), |
|
131
|
|
|
]; |
|
132
|
|
|
} |
|
133
|
|
|
} |
|
134
|
|
|
|
VIPnytt /
RobotsTxtParser
This check looks at variables that are passed out again to other methods.
If the outgoing method call has stricter type requirements than the method itself, an issue is raised.
An additional type check may prevent trouble.