|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of dispositif/wikibot application |
|
4
|
|
|
* 2019 © Philippe M. <[email protected]> |
|
5
|
|
|
* For the full copyright and MIT license information, please view the LICENSE file. |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace App\Domain\Publisher; |
|
11
|
|
|
|
|
12
|
|
|
use DateTime; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Generic mapper for press/revue article on web. |
|
16
|
|
|
* Using JSON-LD and meta tags to obtain {article} data. |
|
17
|
|
|
* Class WebMapper |
|
18
|
|
|
* |
|
19
|
|
|
* @package App\Domain\Publisher |
|
20
|
|
|
*/ |
|
21
|
|
|
class WebMapper implements MapperInterface |
|
22
|
|
|
{ |
|
23
|
|
|
public function process($data): array |
|
24
|
|
|
{ |
|
25
|
|
|
if (!isset($data['JSON-LD'])) { |
|
26
|
|
|
return []; |
|
27
|
|
|
} |
|
28
|
|
|
$data = $data['JSON-LD']; |
|
29
|
|
|
// |
|
30
|
|
|
// foreach ($data as $dat) { |
|
31
|
|
|
// if ('NewsArticle' === $dat['@type']) { |
|
32
|
|
|
// $data = $dat; |
|
33
|
|
|
// goto mapping; |
|
34
|
|
|
// } |
|
35
|
|
|
// // todo ('WebPage' == @type) ==> {{lien web}} ? |
|
36
|
|
|
// } |
|
37
|
|
|
// |
|
38
|
|
|
// return []; // exception ? |
|
39
|
|
|
// |
|
40
|
|
|
// mapping: |
|
41
|
|
|
|
|
42
|
|
|
return [ |
|
43
|
|
|
// 'langue' => 'fr', |
|
44
|
|
|
'périodique' => $data['publisher']['name'] ?? null, |
|
45
|
|
|
// 'acces' => $data['isAccessibleForFree'], |
|
46
|
|
|
'titre' => html_entity_decode($data['headline']), |
|
47
|
|
|
'lire en ligne' => $data['mainEntityOfPage']['@id'], |
|
48
|
|
|
'date' => $this->convertDate($data['datePublished'] ?? null), // 2020-03-19T19:13:01.000Z |
|
49
|
|
|
'auteur1' => $this->convertAuteur($data, 0), |
|
50
|
|
|
'auteur2' => $this->convertAuteur($data, 1), |
|
51
|
|
|
'auteur3' => $this->convertAuteur($data, 2), |
|
52
|
|
|
'auteur institutionnel' => $this->convertInstitutionnel($data), |
|
53
|
|
|
]; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
protected function convertAuteur($data, $indice) |
|
57
|
|
|
{ |
|
58
|
|
|
// author ['name'=>'Bob','@type'=>'Person'] |
|
59
|
|
|
if (0 === $indice |
|
60
|
|
|
&& isset($data['author']) |
|
61
|
|
|
&& isset($data['author']['name']) |
|
62
|
|
|
&& (!isset($data['author']['@type']) |
|
63
|
|
|
|| 'Person' === $data['author']['@type']) |
|
64
|
|
|
) { |
|
65
|
|
|
return html_entity_decode($data['author']['name']); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
// author [ 0 => ['name'=>'Bob'], 1=> ...] |
|
69
|
|
|
if (isset($data['author']) && isset($data['author'][$indice]) |
|
70
|
|
|
&& (!isset($data['author'][$indice]['@type']) |
|
71
|
|
|
|| 'Person' === $data['author'][$indice]['@type']) |
|
72
|
|
|
) { |
|
73
|
|
|
return html_entity_decode($data['author'][$indice]['name']); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return null; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
protected function convertInstitutionnel($data) |
|
80
|
|
|
{ |
|
81
|
|
|
if (isset($data['author']) && isset($data['author'][0]) && isset($data['author'][0]['@type']) |
|
82
|
|
|
&& 'Person' !== $data['author'][0]['@type'] |
|
83
|
|
|
) { |
|
84
|
|
|
return html_entity_decode($data['author'][0]['name']); |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
return null; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* @param string $str |
|
92
|
|
|
* |
|
93
|
|
|
* @return string |
|
94
|
|
|
*/ |
|
95
|
|
|
protected function convertDate(?string $str): ?string |
|
96
|
|
|
{ |
|
97
|
|
|
if (empty($str)) { |
|
98
|
|
|
return null; |
|
99
|
|
|
} |
|
100
|
|
|
try { |
|
101
|
|
|
$date = new DateTime($str); |
|
102
|
|
|
} catch (\Exception $e) { |
|
103
|
|
|
return null; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
return $date->format('d-m-Y'); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* Wikification des noms/acronymes d'agences de presse. |
|
111
|
|
|
* |
|
112
|
|
|
* @param string $str |
|
113
|
|
|
* |
|
114
|
|
|
* @return string |
|
115
|
|
|
*/ |
|
116
|
|
|
protected function wikifyPressAgency(string $str): string |
|
117
|
|
|
{ |
|
118
|
|
|
// skip potential wikilinks |
|
119
|
|
|
if (strpos($str, '[') !== false) { |
|
120
|
|
|
return $str; |
|
121
|
|
|
} |
|
122
|
|
|
$str = preg_replace('#\b(AFP)\b#', '[[Agence France-Presse|AFP]]', $str); |
|
123
|
|
|
$str = str_replace('Reuters', '[[Reuters]]', $str); |
|
124
|
|
|
$str = str_replace('Associated Press', '[[Associated Press]]', $str); |
|
125
|
|
|
$str = preg_replace('#\b(PA)\b#', '[[Press Association|PA]]', $str); |
|
126
|
|
|
$str = preg_replace('#\b(AP)\b#', '[[Associated Press|AP]]', $str); |
|
127
|
|
|
$str = str_replace('Xinhua', '[[Xinhua]]', $str); |
|
128
|
|
|
$str = preg_replace('#\b(ATS)\b#', '[[Agence télégraphique suisse|ATS]]', $str); |
|
129
|
|
|
$str = preg_replace('#\b(PC|CP)\b#', '[[La Presse canadienne|PC]]', $str); |
|
130
|
|
|
|
|
131
|
|
|
return $str; |
|
132
|
|
|
} |
|
133
|
|
|
} |
|
134
|
|
|
|
Dispositif /
Wikibot
