|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace App\Domain\Publisher\Tests; |
|
6
|
|
|
|
|
7
|
|
|
use App\Application\PublisherAction; |
|
8
|
|
|
use App\Domain\Publisher\WebMapper; |
|
9
|
|
|
use App\Infrastructure\Logger; |
|
10
|
|
|
use Exception; |
|
11
|
|
|
use PHPUnit\Framework\TestCase; |
|
12
|
|
|
|
|
13
|
|
|
class WebMapperTest extends TestCase |
|
14
|
|
|
{ |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* @dataProvider provideMappingFromFile |
|
18
|
|
|
* |
|
19
|
|
|
* @param $filename |
|
20
|
|
|
* @param array $expected |
|
21
|
|
|
* |
|
22
|
|
|
* @return void |
|
23
|
|
|
* @throws Exception |
|
24
|
|
|
*/ |
|
25
|
|
|
public function testMappingProcess($filename, array $expected): void |
|
26
|
|
|
{ |
|
27
|
|
|
$html = file_get_contents($filename); |
|
28
|
|
|
|
|
29
|
|
|
$publiAction = new PublisherAction('bla'); |
|
30
|
|
|
$htmlData = $publiAction->extractWebData($html); |
|
31
|
|
|
$mapper = new WebMapper(new Logger()); |
|
32
|
|
|
$data = $mapper->process($htmlData); |
|
33
|
|
|
if (isset($data['consulté le'])) { |
|
34
|
|
|
$data['consulté le'] = '11-04-2020';// unit testing date... |
|
35
|
|
|
} |
|
36
|
|
|
$this::assertSame($expected, $data); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
public function provideMappingFromFile() |
|
40
|
|
|
{ |
|
41
|
|
|
return [ |
|
42
|
|
|
[ |
|
43
|
|
|
__DIR__.'/fixture_lemonde.html', |
|
44
|
|
|
[ |
|
45
|
|
|
'DATA-TYPE' => 'JSON-LD', |
|
46
|
|
|
'DATA-ARTICLE' => true, |
|
47
|
|
|
'périodique' => 'Le Monde', |
|
48
|
|
|
'titre' => 'Coronavirus : la Californie placée à son tour en confinement', |
|
49
|
|
|
'url' => 'https://www.lemonde.fr/planete/article/2020/03/20/coronavirus-la-californie-placee-en-confinement_6033754_3244.html', |
|
50
|
|
|
'date' => '20-03-2020', |
|
51
|
|
|
], |
|
52
|
|
|
], |
|
53
|
|
|
[ |
|
54
|
|
|
__DIR__.'/fixture_journalsOpenEdition.html', |
|
55
|
|
|
[ |
|
56
|
|
|
|
|
57
|
|
|
'DATA-TYPE' => 'Open Graph/Dublin Core', |
|
58
|
|
|
'DATA-ARTICLE' => true, |
|
59
|
|
|
'titre' => 'Alger', |
|
60
|
|
|
'url' => 'http://journals.openedition.org/encyclopedieberbere/2434', |
|
61
|
|
|
'consulté le' => '11-04-2020', |
|
62
|
|
|
'date' => '01-07-1986', |
|
63
|
|
|
'url-access' => 'limité', |
|
64
|
|
|
'périodique' => 'Encyclopédie berbère', |
|
65
|
|
|
'et al.' => 'oui', |
|
66
|
|
|
'auteur1' => 'Camps, G.; Leglay, M.', |
|
67
|
|
|
'numéro' => '4', |
|
68
|
|
|
'page' => '447–472', |
|
69
|
|
|
'éditeur' => 'Éditions Peeters', |
|
70
|
|
|
'issn' => '1015-7344', |
|
71
|
|
|
'isbn' => '2-85744-282-3', |
|
72
|
|
|
], |
|
73
|
|
|
], |
|
74
|
|
|
[ |
|
75
|
|
|
__DIR__.'/fixture_pubmed.html', |
|
76
|
|
|
[ |
|
77
|
|
|
'DATA-TYPE' => 'Open Graph/Dublin Core', |
|
78
|
|
|
'DATA-ARTICLE' => true, |
|
79
|
|
|
'site' => 'PubMed Central (PMC)', |
|
80
|
|
|
'titre' => 'The Diesel Exhaust in Miners Study: A Nested Case–Control Study of Lung Cancer and Diesel Exhaust', |
|
81
|
|
|
'url' => 'https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3369553/', |
|
82
|
|
|
'consulté le' => '11-04-2020', |
|
83
|
|
|
'date' => '06-06-2012', |
|
84
|
|
|
'périodique' => 'JNCI Journal of the National Cancer Institute', |
|
85
|
|
|
'et al.' => 'oui', |
|
86
|
|
|
'auteur1' => 'Debra T. Silverman, Claudine M. Samanic', |
|
87
|
|
|
'volume' => '104', |
|
88
|
|
|
'numéro' => '11', |
|
89
|
|
|
'page' => '855', |
|
90
|
|
|
'doi' => '10.1093/jnci/djs034', |
|
91
|
|
|
'pmid' => '22393209', |
|
92
|
|
|
], |
|
93
|
|
|
], |
|
94
|
|
|
[ |
|
95
|
|
|
__DIR__.'/fixture_figaro.html', |
|
96
|
|
|
[ |
|
97
|
|
|
'DATA-TYPE' => 'JSON-LD', |
|
98
|
|
|
'DATA-ARTICLE' => true, |
|
99
|
|
|
'périodique' => 'Le Figaro', |
|
100
|
|
|
'titre' => 'Face au Covid-19, les cliniques privées mobilisées… mais en manque de masques', |
|
101
|
|
|
'url' => 'http://www.lefigaro.fr/sciences/face-au-covid-19-les-cliniques-privees-mobilisees-mais-en-manque-de-masques-20200318', |
|
102
|
|
|
'date' => '18-03-2020', |
|
103
|
|
|
'auteur1' => 'Marie-Cécile Renault', |
|
104
|
|
|
'url-access' => 'limité', |
|
105
|
|
|
], |
|
106
|
|
|
], |
|
107
|
|
|
[ |
|
108
|
|
|
__DIR__.'/fixture_liberation.html', |
|
109
|
|
|
[ |
|
110
|
|
|
'DATA-TYPE' => 'JSON-LD', |
|
111
|
|
|
'DATA-ARTICLE' => true, |
|
112
|
|
|
'périodique' => 'Libération', |
|
113
|
|
|
'titre' => 'En Bretagne, le Parisiens-bashing guette', |
|
114
|
|
|
'url' => 'https://www.liberation.fr/france/2020/03/20/en-bretagne-le-parisiens-bashing-guette_1782471', |
|
115
|
|
|
'date' => '20-03-2020', |
|
116
|
|
|
'auteur1' => 'Pierre-Henri Allain', |
|
117
|
|
|
'url-access' => 'ouvert', |
|
118
|
|
|
], |
|
119
|
|
|
], |
|
120
|
|
|
|
|
121
|
|
|
]; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
} |
|
125
|
|
|
|
Dispositif /
Wikibot
