|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace SP\Spiderling; |
|
4
|
|
|
|
|
5
|
|
|
use Psr\Http\Message\UriInterface; |
|
6
|
|
|
use GuzzleHttp\Psr7\Uri; |
|
7
|
|
|
use InvalidArgumentException; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* @author Ivan Kerin <[email protected]> |
|
11
|
|
|
* @copyright 2015, Clippings Ltd. |
|
12
|
|
|
* @license http://spdx.org/licenses/BSD-3-Clause |
|
13
|
|
|
*/ |
|
14
|
|
|
class CrawlerSession |
|
15
|
|
|
{ |
|
16
|
|
|
use TraverseTrait; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* CrawlerInterface |
|
20
|
|
|
*/ |
|
21
|
|
|
private $crawler; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @param CrawlerInterface $crawler |
|
25
|
|
|
*/ |
|
26
|
|
|
public function __construct(CrawlerInterface $crawler) |
|
27
|
|
|
{ |
|
28
|
|
|
$this->crawler = $crawler; |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @param Query\AbstractQuery $query |
|
33
|
|
|
* @return array |
|
34
|
|
|
*/ |
|
35
|
1 |
|
public function queryIds(Query\AbstractQuery $query) |
|
36
|
|
|
{ |
|
37
|
|
|
return $this->crawler->queryIds($query); |
|
38
|
1 |
|
} |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* @return CrawlerInterface |
|
42
|
|
|
*/ |
|
43
|
1 |
|
public function getCrawler() |
|
44
|
|
|
{ |
|
45
|
1 |
|
return $this->crawler; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @param string $uri |
|
50
|
|
|
* @return self |
|
51
|
|
|
*/ |
|
52
|
|
|
public function open($uri) |
|
53
|
|
|
{ |
|
54
|
|
|
$uri = \GuzzleHttp\Psr7\uri_for($uri); |
|
55
|
|
|
|
|
56
|
|
|
$this->crawler->open($uri); |
|
57
|
|
|
|
|
58
|
|
|
return $this; |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* @return UriInterface |
|
63
|
|
|
*/ |
|
64
|
|
|
public function getUri() |
|
65
|
|
|
{ |
|
66
|
|
|
return $this->crawler->getUri(); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* @return string |
|
71
|
|
|
*/ |
|
72
|
|
|
public function getHtml() |
|
73
|
|
|
{ |
|
74
|
|
|
return $this->crawler->getFullHtml(); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
/** |
|
78
|
|
|
* Save the HTML of the session into a file |
|
79
|
|
|
* Optionally resolve all the links with a base uri |
|
80
|
|
|
* |
|
81
|
|
|
* @param string $filename |
|
82
|
|
|
* @throws InvalidArgumentException if directory doesnt exist or is not writable |
|
83
|
|
|
* @param UriInterface|string $base |
|
84
|
|
|
*/ |
|
85
|
1 |
|
public function saveHtml($filename, $base = null) |
|
86
|
|
|
{ |
|
87
|
|
|
$this->ensureWritableDirectory(dirname($filename)); |
|
88
|
|
|
|
|
89
|
|
|
$html = new Html($this->getHtml()); |
|
90
|
|
|
|
|
91
|
1 |
|
if (null !== $base) { |
|
92
|
|
|
$html->resolveLinks(\GuzzleHttp\Psr7\uri_for($base)); |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
file_put_contents($filename, $html->get()); |
|
96
|
|
|
|
|
97
|
1 |
|
return $this; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* @param string $directory |
|
102
|
|
|
* @return boolean |
|
103
|
|
|
*/ |
|
104
|
|
|
public function isWritableDirectory($directory) |
|
105
|
|
|
{ |
|
106
|
|
|
return is_dir($directory) && is_writable($directory); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* @param string $directory |
|
111
|
|
|
* @throws InvalidArgumentException if directory doesnt exist or is not writable |
|
112
|
|
|
*/ |
|
113
|
1 |
|
public function ensureWritableDirectory($directory) |
|
114
|
|
|
{ |
|
115
|
|
|
if (false === $this->isWritableDirectory($directory)) { |
|
116
|
|
|
throw new InvalidArgumentException(sprintf( |
|
117
|
1 |
|
'Make sure directory %s exists and is writable', |
|
118
|
|
|
$directory |
|
119
|
|
|
)); |
|
120
|
|
|
} |
|
121
|
1 |
|
} |
|
122
|
|
|
} |
|
123
|
|
|
|
spiderling-php /
spiderling
