|
1
|
|
|
<?php |
|
2
|
|
|
namespace vipnytt\RobotsTxtParser\Client; |
|
3
|
|
|
|
|
4
|
|
|
use GuzzleHttp; |
|
5
|
|
|
use vipnytt\RobotsTxtParser\Parser\UrlParser; |
|
6
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* Class Download |
|
10
|
|
|
* |
|
11
|
|
|
* @package vipnytt\RobotsTxtParser\Client |
|
12
|
|
|
*/ |
|
13
|
|
|
class Download implements RobotsTxtInterface |
|
14
|
|
|
{ |
|
15
|
|
|
use UrlParser; |
|
16
|
|
|
|
|
17
|
|
|
private $response; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* Download constructor. |
|
21
|
|
|
* |
|
22
|
|
|
* @param string $url |
|
23
|
|
|
*/ |
|
24
|
|
|
public function __construct($url) |
|
25
|
|
|
{ |
|
26
|
|
|
$client = new GuzzleHttp\Client( |
|
27
|
|
|
[ |
|
28
|
|
|
'base_uri' => $this->urlBase($url), |
|
29
|
|
|
'max' => self::MAX_REDIRECTS, |
|
30
|
|
|
'http_errors' => false, |
|
31
|
|
|
] |
|
32
|
|
|
); |
|
33
|
|
|
$this->response = $client->request('GET', '/robots.txt'); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* Status code |
|
38
|
|
|
* |
|
39
|
|
|
* @return int |
|
40
|
|
|
*/ |
|
41
|
|
|
public function getStatusCode() |
|
42
|
|
|
{ |
|
43
|
|
|
return $this->response->getStatusCode(); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Encoding |
|
48
|
|
|
* |
|
49
|
|
|
* @return string |
|
50
|
|
|
*/ |
|
51
|
|
|
public function getEncoding() |
|
52
|
|
|
{ |
|
53
|
|
|
$header = $this->response->getHeader('content-type')[0]; |
|
54
|
|
|
$split = array_map('trim', mb_split(';', $header)); |
|
55
|
|
|
foreach ($split as $string) { |
|
56
|
|
|
if (mb_stripos($string, 'charset=') === 0) { |
|
57
|
|
|
return mb_split('=', $string, 2)[1]; |
|
58
|
|
|
} |
|
59
|
|
|
} |
|
60
|
|
|
return $this->detectEncoding(); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* Manually detect encoding |
|
65
|
|
|
* |
|
66
|
|
|
* @return string |
|
67
|
|
|
*/ |
|
68
|
|
|
protected function detectEncoding() |
|
69
|
|
|
{ |
|
70
|
|
|
if (($encoding = mb_detect_encoding($this->getBody())) !== false) { |
|
71
|
|
|
return $encoding; |
|
72
|
|
|
} |
|
73
|
|
|
return self::ENCODING; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* URL content |
|
78
|
|
|
* |
|
79
|
|
|
* @return string |
|
80
|
|
|
*/ |
|
81
|
|
|
public function getBody() |
|
82
|
|
|
{ |
|
83
|
|
|
return $this->response->getBody()->getContents(); |
|
84
|
|
|
} |
|
85
|
|
|
} |
|
86
|
|
|
|
VIPnytt /
RobotsTxtParser
