|
1
|
|
|
<?php |
|
2
|
|
|
namespace VDB\Spider; |
|
3
|
|
|
|
|
4
|
|
|
use Guzzle\Http\Message\Response; |
|
5
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
6
|
|
|
use VDB\Spider\Uri\DiscoveredUri; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* @author Matthijs van den Bos |
|
10
|
|
|
* @copyright 2013 Matthijs van den Bos |
|
11
|
|
|
*/ |
|
12
|
|
|
class Resource |
|
13
|
|
|
{ |
|
14
|
|
|
/** @var DiscoveredUri */ |
|
15
|
|
|
protected $uri; |
|
16
|
|
|
|
|
17
|
|
|
/** @var Response */ |
|
18
|
|
|
protected $response; |
|
19
|
|
|
|
|
20
|
|
|
/** @var Crawler */ |
|
21
|
|
|
protected $crawler; |
|
22
|
|
|
|
|
23
|
|
|
/** @var string */ |
|
24
|
|
|
protected $body; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @param DiscoveredUri $uri |
|
28
|
|
|
* @param Response $response |
|
29
|
|
|
*/ |
|
30
|
|
|
public function __construct(DiscoveredUri $uri, Response $response) |
|
31
|
|
|
{ |
|
32
|
|
|
$this->uri = $uri; |
|
33
|
|
|
$this->response = $response; |
|
34
|
|
|
|
|
35
|
|
|
// we store the response manually, because otherwise it will not get serialized. It is a php://temp stream |
|
36
|
|
|
$this->body = $response->getBody(true); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* Lazy loads a Crawler object based on the Response; |
|
41
|
|
|
* @return Crawler |
|
42
|
|
|
*/ |
|
43
|
|
|
public function getCrawler() |
|
44
|
|
|
{ |
|
45
|
|
|
if (!$this->crawler instanceof Crawler) { |
|
46
|
|
|
$this->crawler = new Crawler('', $this->getUri()->toString()); |
|
47
|
|
|
$this->crawler->addContent( |
|
48
|
|
|
$this->getResponse()->getBody(true), |
|
49
|
|
|
$this->getResponse()->getHeader('Content-Type', true) |
|
50
|
|
|
); |
|
51
|
|
|
} |
|
52
|
|
|
return $this->crawler; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* @return DiscoveredUri |
|
57
|
|
|
*/ |
|
58
|
|
|
public function getUri() |
|
59
|
|
|
{ |
|
60
|
|
|
return $this->uri; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* @return Response |
|
65
|
|
|
*/ |
|
66
|
|
|
public function getResponse() |
|
67
|
|
|
{ |
|
68
|
|
|
return $this->response; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
public function __sleep() |
|
72
|
|
|
{ |
|
73
|
|
|
/* |
|
74
|
|
|
* Because the Crawler isn't serialized correctly, we exclude it from serialization |
|
75
|
|
|
* It will be available again after wakeup through lazy loading with getCrawler() |
|
76
|
|
|
*/ |
|
77
|
|
|
return array( |
|
78
|
|
|
'uri', |
|
79
|
|
|
'response', |
|
80
|
|
|
'body' |
|
81
|
|
|
); |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* We need to set the body again after deserialization because it was a stream that didn't get serialized |
|
86
|
|
|
*/ |
|
87
|
|
|
public function __wakeup() |
|
88
|
|
|
{ |
|
89
|
|
|
$this->response->setBody($this->body); |
|
90
|
|
|
} |
|
91
|
|
|
} |
|
92
|
|
|
|
mvdbos /
php-spider
