|
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 VDB\Spider\Tests\TestCase; |
|
15
|
|
|
use VDB\Spider\PersistenceHandler\FileSerializedResourcePersistenceHandler; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* |
|
19
|
|
|
*/ |
|
20
|
|
|
class FileSerializedResourcePersistenceHandlerTest extends TestCase |
|
21
|
|
|
{ |
|
22
|
|
|
/** |
|
23
|
|
|
* @var FileSerializedResourcePersistenceHandler |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $handler; |
|
26
|
|
|
|
|
27
|
|
|
protected $persistenceRootPath; |
|
28
|
|
|
|
|
29
|
|
|
|
|
30
|
|
|
public function setUp(): void |
|
31
|
|
|
{ |
|
32
|
|
|
$this->persistenceRootPath = sys_get_temp_dir() . DIRECTORY_SEPARATOR . 'spider-UT' . DIRECTORY_SEPARATOR; |
|
33
|
|
|
exec('rm -rf ' . $this->persistenceRootPath); |
|
34
|
|
|
|
|
35
|
|
|
$this->handler = new FileSerializedResourcePersistenceHandler(sys_get_temp_dir()); |
|
36
|
|
|
$this->handler->setSpiderId('spider-UT'); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @covers VDB\Spider\PersistenceHandler\FileSerializedResourcePersistenceHandler |
|
41
|
|
|
* @covers VDB\Spider\PersistenceHandler\FilePersistenceHandler |
|
42
|
|
|
* |
|
43
|
|
|
* @dataProvider persistenceProvider |
|
44
|
|
|
*/ |
|
45
|
|
|
public function testPersist($resource, $expectedFilePath, $expectedFileContents) |
|
46
|
|
|
{ |
|
47
|
|
|
$this->handler->persist($resource); |
|
48
|
|
|
|
|
49
|
|
|
$this->assertFileExists($expectedFilePath); |
|
50
|
|
|
|
|
51
|
|
|
$savedResource = unserialize(file_get_contents($expectedFilePath)); |
|
52
|
|
|
$this->assertEquals( |
|
53
|
|
|
$expectedFileContents, |
|
54
|
|
|
$savedResource->getResponse()->getBody() |
|
55
|
|
|
); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @covers VDB\Spider\PersistenceHandler\FileSerializedResourcePersistenceHandler |
|
60
|
|
|
* @covers VDB\Spider\PersistenceHandler\FilePersistenceHandler |
|
61
|
|
|
* |
|
62
|
|
|
* @dataProvider persistenceWithoutFilenameProvider |
|
63
|
|
|
*/ |
|
64
|
|
|
public function testPersistResourcesWithoutFilename($resource, $expectedFilePath, $expectedFileContents) |
|
65
|
|
|
{ |
|
66
|
|
|
$this->handler->persist($resource); |
|
67
|
|
|
|
|
68
|
|
|
$this->assertFileExists($expectedFilePath); |
|
69
|
|
|
|
|
70
|
|
|
$savedResource = unserialize(file_get_contents($expectedFilePath)); |
|
71
|
|
|
$this->assertEquals( |
|
72
|
|
|
$expectedFileContents, |
|
73
|
|
|
$savedResource->getResponse()->getBody() |
|
74
|
|
|
); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
public function persistenceWithoutFilenameProvider() |
|
78
|
|
|
{ |
|
79
|
|
|
// This must be set here instead of in setup methods, because providers |
|
80
|
|
|
// get executed first |
|
81
|
|
|
if (is_null($this->persistenceRootPath)) { |
|
82
|
|
|
$this->persistenceRootPath = sys_get_temp_dir() . DIRECTORY_SEPARATOR . 'spider-UT' . DIRECTORY_SEPARATOR; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
$data = []; |
|
86
|
|
|
|
|
87
|
|
|
$data[] = $this->buildPersistenceProviderRecord( |
|
88
|
|
|
__DIR__ . '/../Fixtures/DownloaderTestHTMLResource.html', |
|
89
|
|
|
'http://example.org/domains/Internet/' |
|
90
|
|
|
); |
|
91
|
|
|
|
|
92
|
|
|
$data[] = $this->buildPersistenceProviderRecord( |
|
93
|
|
|
__DIR__ . '/../Fixtures/DownloaderTestHTMLResource.html', |
|
94
|
|
|
'http://example.org/domains/Internet/Abuse/' |
|
95
|
|
|
); |
|
96
|
|
|
|
|
97
|
|
|
return $data; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
public function persistenceProvider() |
|
101
|
|
|
{ |
|
102
|
|
|
// This must be set here instead of in setup methods, because providers |
|
103
|
|
|
// get executed first |
|
104
|
|
|
if (is_null($this->persistenceRootPath)) { |
|
105
|
|
|
$this->persistenceRootPath = sys_get_temp_dir() . DIRECTORY_SEPARATOR . 'spider-UT' . DIRECTORY_SEPARATOR; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
$data = []; |
|
109
|
|
|
|
|
110
|
|
|
$data[] = $this->buildPersistenceProviderRecord( |
|
111
|
|
|
__DIR__ . '/../Fixtures/DownloaderTestHTMLResource.html', |
|
112
|
|
|
'http://example.org/domains/special/test1.html' |
|
113
|
|
|
); |
|
114
|
|
|
|
|
115
|
|
|
$data[] = $this->buildPersistenceProviderRecord( |
|
116
|
|
|
__DIR__ . '/../Fixtures/DownloaderTestHTMLResource.html', |
|
117
|
|
|
'http://example.org/domains/special/test2.html' |
|
118
|
|
|
); |
|
119
|
|
|
|
|
120
|
|
|
$data[] = $this->buildPersistenceProviderRecord( |
|
121
|
|
|
__DIR__ . '/../Fixtures/DownloaderTestHTMLResource.html', |
|
122
|
|
|
'http://example.org/domains/special/subdir/test3.html' |
|
123
|
|
|
); |
|
124
|
|
|
|
|
125
|
|
|
return $data; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
protected function buildPersistenceProviderRecord($fixturePath, $uriString) |
|
129
|
|
|
{ |
|
130
|
|
|
$resource = $this->buildResourceFromFixture( |
|
131
|
|
|
$fixturePath, |
|
132
|
|
|
$uriString |
|
133
|
|
|
); |
|
134
|
|
|
$expectedFileContents = $this->getFixtureContent(__DIR__ . '/../Fixtures/DownloaderTestHTMLResource.html'); |
|
135
|
|
|
$expectedFilePath = $this->buildExpectedFilePath($uriString); |
|
136
|
|
|
|
|
137
|
|
|
return [$resource, $expectedFilePath, $expectedFileContents]; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
protected function buildExpectedFilePath($uriString) |
|
141
|
|
|
{ |
|
142
|
|
|
$expectedFilePath = $this->persistenceRootPath . parse_url($uriString)['host'] . parse_url($uriString)['path']; |
|
143
|
|
|
if (substr($expectedFilePath, -1, 1) === '/') { |
|
144
|
|
|
$expectedFilePath .= 'index.html'; |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
return $expectedFilePath; |
|
148
|
|
|
} |
|
149
|
|
|
} |
|
150
|
|
|
|
mvdbos /
php-spider
