1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/* |
4
|
|
|
* This file is part of the Spider package. |
5
|
|
|
* |
6
|
|
|
* (c) Matthijs van den Bos <[email protected]> |
7
|
|
|
* |
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
9
|
|
|
* file that was distributed with this source code. |
10
|
|
|
*/ |
11
|
|
|
|
12
|
|
|
namespace VDB\Spider\Tests\Downloader; |
13
|
|
|
|
14
|
|
|
use GuzzleHttp\Psr7\Response; |
15
|
|
|
use VDB\Spider\Tests\TestCase; |
16
|
|
|
use VDB\Spider\Downloader\DownloaderInterface; |
17
|
|
|
use VDB\Spider\Downloader\Downloader; |
18
|
|
|
use VDB\Spider\Uri\DiscoveredUri; |
19
|
|
|
use VDB\Spider\Resource; |
20
|
|
|
use VDB\Uri\Uri; |
21
|
|
|
|
22
|
|
|
/** |
23
|
|
|
* |
24
|
|
|
*/ |
25
|
|
|
class DownloaderTest extends TestCase |
26
|
|
|
{ |
27
|
|
|
/** |
28
|
|
|
* @var Downloader |
29
|
|
|
*/ |
30
|
|
|
private $downloader; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @var Resource |
34
|
|
|
*/ |
35
|
|
|
protected $resource; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* @var string |
39
|
|
|
*/ |
40
|
|
|
protected $html; |
41
|
|
|
|
42
|
|
|
public function setUp(): void |
43
|
|
|
{ |
44
|
|
|
$this->html = file_get_contents(__DIR__ . '/../Fixtures/DownloaderTestHTMLResource.html'); |
45
|
|
|
$this->resource = new Resource( |
46
|
|
|
new DiscoveredUri(new Uri('/domains/special', 'http://example.org')), |
47
|
|
|
new Response(200, [], $this->html) |
48
|
|
|
); |
49
|
|
|
|
50
|
|
|
$this->downloader = new Downloader(); |
51
|
|
|
|
52
|
|
|
$requestHandler = $this->getMockBuilder('VDB\Spider\RequestHandler\RequestHandlerInterface')->getMock(); |
53
|
|
|
$requestHandler |
54
|
|
|
->expects($this->any()) |
55
|
|
|
->method('request') |
56
|
|
|
->will($this->returnValue($this->resource)); |
57
|
|
|
|
58
|
|
|
$this->downloader->setRequestHandler($requestHandler); |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @covers VDB\Spider\Downloader\Downloader |
63
|
|
|
*/ |
64
|
|
|
public function testDownload() |
65
|
|
|
{ |
66
|
|
|
$resource = $this->downloader->download(new DiscoveredUri(new Uri('http://foobar.org'))); |
67
|
|
|
$this->assertInstanceOf('VDB\\Spider\\Resource', $resource); |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* @covers VDB\Spider\Downloader\Downloader |
72
|
|
|
*/ |
73
|
|
|
public function testDownloadFailed() |
74
|
|
|
{ |
75
|
|
|
$requestHandler = $this->getMockBuilder('VDB\Spider\RequestHandler\RequestHandlerInterface')->getMock(); |
76
|
|
|
$requestHandler |
77
|
|
|
->expects($this->any()) |
78
|
|
|
->method('request') |
79
|
|
|
->will($this->throwException(new \Exception)); |
80
|
|
|
$this->downloader->setRequestHandler($requestHandler); |
81
|
|
|
|
82
|
|
|
$resource = $this->downloader->download(new DiscoveredUri(new Uri('http://foobar.org'))); |
83
|
|
|
|
84
|
|
|
$this->assertFalse($resource); |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
/** |
88
|
|
|
* @covers VDB\Spider\Downloader\Downloader |
89
|
|
|
*/ |
90
|
|
|
public function testFilterNotMatches() |
91
|
|
|
{ |
92
|
|
|
$filterNeverMatch = $this->getMockBuilder('VDB\Spider\Filter\PostFetchFilterInterface')->getMock(); |
93
|
|
|
$filterNeverMatch |
94
|
|
|
->expects($this->any()) |
95
|
|
|
->method('match') |
96
|
|
|
->will($this->returnValue(false)); |
97
|
|
|
$this->downloader->addPostFetchFilter($filterNeverMatch); |
98
|
|
|
|
99
|
|
|
$resource = $this->downloader->download(new DiscoveredUri(new Uri('http://foobar.org'))); |
100
|
|
|
|
101
|
|
|
$this->assertInstanceOf('VDB\\Spider\\Resource', $resource); |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* @covers VDB\Spider\Downloader\Downloader |
106
|
|
|
*/ |
107
|
|
|
public function testFilterMatches() |
108
|
|
|
{ |
109
|
|
|
$filterAlwaysMatch = $this->getMockBuilder('VDB\Spider\Filter\PostFetchFilterInterface')->getMock(); |
110
|
|
|
$filterAlwaysMatch |
111
|
|
|
->expects($this->any()) |
112
|
|
|
->method('match') |
113
|
|
|
->will($this->returnValue(true)); |
114
|
|
|
$this->downloader->addPostFetchFilter($filterAlwaysMatch); |
115
|
|
|
|
116
|
|
|
$resource = $this->downloader->download(new DiscoveredUri(new Uri('http://foobar.org'))); |
117
|
|
|
|
118
|
|
|
$this->assertFalse($resource); |
119
|
|
|
} |
120
|
|
|
} |
121
|
|
|
|