|
1
|
|
|
<?php |
|
2
|
|
|
namespace vipnytt\RobotsTxtParser\Client\Directives; |
|
3
|
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Exceptions\ClientException; |
|
5
|
|
|
use vipnytt\RobotsTxtParser\Parser\Directives\DirectiveParserCommons; |
|
6
|
|
|
use vipnytt\RobotsTxtParser\Parser\UrlParser; |
|
7
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
|
8
|
|
|
|
|
9
|
|
|
class DisAllowClient implements ClientInterface, RobotsTxtInterface |
|
10
|
|
|
{ |
|
11
|
|
|
use DirectiveParserCommons; |
|
12
|
|
|
use UrlParser; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Paths |
|
16
|
|
|
* @var array |
|
17
|
|
|
*/ |
|
18
|
|
|
private $paths; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Host |
|
22
|
|
|
* @var HostClient |
|
23
|
|
|
*/ |
|
24
|
|
|
private $host; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Clean-param |
|
28
|
|
|
* @var CleanParamClient |
|
29
|
|
|
*/ |
|
30
|
|
|
private $cleanParam; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* DisAllowClient constructor. |
|
34
|
|
|
* |
|
35
|
|
|
* @param array $paths |
|
36
|
|
|
* @param HostClient $host |
|
37
|
|
|
* @param CleanParamClient $cleanParam |
|
38
|
|
|
*/ |
|
39
|
|
|
public function __construct(array $paths, HostClient $host, CleanParamClient $cleanParam) |
|
40
|
|
|
{ |
|
41
|
|
|
$this->paths = $paths; |
|
42
|
|
|
$this->host = $host; |
|
43
|
|
|
$this->cleanParam = $cleanParam; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Check |
|
48
|
|
|
* |
|
49
|
|
|
* @param string $url |
|
50
|
|
|
* @return bool |
|
51
|
|
|
*/ |
|
52
|
|
|
public function isListed($url) |
|
53
|
|
|
{ |
|
54
|
|
|
$path = $this->getPath($url); |
|
55
|
|
|
return ( |
|
56
|
|
|
$this->checkPath($path, $this->paths) || |
|
57
|
|
|
$this->cleanParam->isListed($path) || |
|
58
|
|
|
$this->host->isListed($url) |
|
59
|
|
|
); |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Get path and query |
|
64
|
|
|
* |
|
65
|
|
|
* @param string $url |
|
66
|
|
|
* @return string |
|
67
|
|
|
* @throws ClientException |
|
68
|
|
|
*/ |
|
69
|
|
|
private function getPath($url) |
|
70
|
|
|
{ |
|
71
|
|
|
// Encode |
|
72
|
|
|
$url = mb_split('#', $this->urlEncode($url))[0]; |
|
73
|
|
|
if (mb_stripos($url, '/') === 0) { |
|
74
|
|
|
// URL already an path |
|
75
|
|
|
return $url; |
|
76
|
|
|
} |
|
77
|
|
|
if (!$this->urlValidate($url)) { |
|
78
|
|
|
throw new ClientException('Invalid URL'); |
|
79
|
|
|
} |
|
80
|
|
|
$path = (($path = parse_url($url, PHP_URL_PATH)) === null) ? '/' : $path; |
|
81
|
|
|
$query = (($query = parse_url($url, PHP_URL_QUERY)) === null) ? '' : '?' . $query; |
|
82
|
|
|
return $path . $query; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* Export |
|
87
|
|
|
* |
|
88
|
|
|
* @return array |
|
89
|
|
|
*/ |
|
90
|
|
|
public function export() |
|
91
|
|
|
{ |
|
92
|
|
|
return [ |
|
93
|
|
|
'host' => $this->host->export(), |
|
94
|
|
|
'path' => $this->paths, |
|
95
|
|
|
'clean-param' => $this->cleanParam->export(), |
|
96
|
|
|
]; |
|
97
|
|
|
} |
|
98
|
|
|
} |
|
99
|
|
|
|
VIPnytt /
RobotsTxtParser
