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\Infrastructure; |
11
|
|
|
|
12
|
|
|
use Exception; |
13
|
|
|
use GuzzleHttp\Client; |
14
|
|
|
use Normalizer; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* dirty scratch WikiData read requests |
18
|
|
|
* Class WikidataAdapter |
19
|
|
|
* |
20
|
|
|
* @package App\Infrastructure |
21
|
|
|
*/ |
22
|
|
|
class WikidataAdapter |
23
|
|
|
{ |
24
|
|
|
private $client; |
25
|
|
|
|
26
|
|
|
public function __construct(?Client $client = null) |
27
|
|
|
{ |
28
|
|
|
if (null === $client) { |
29
|
|
|
// lazy dependency factory :) |
30
|
|
|
$this->client = new Client(['timeout' => 5]); |
31
|
|
|
} else { |
32
|
|
|
$this->client = $client; |
33
|
|
|
} |
34
|
|
|
} |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Get WD item, sitelink, VIAF from search by ISNI (author) |
38
|
|
|
* |
39
|
|
|
* @param string $isni |
40
|
|
|
* |
41
|
|
|
* @return null |
42
|
|
|
* @throws Exception |
43
|
|
|
*/ |
44
|
|
|
public function searchByISNI(string $isni): ?array |
45
|
|
|
{ |
46
|
|
|
if (!$this->ISNIvalide($isni)) { |
47
|
|
|
new Exception('Invalid format for ISNI'); |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
$sparql = sprintf( |
51
|
|
|
'SELECT distinct ?item ?itemLabel ?article ?isni ?viaf WHERE { |
52
|
|
|
?item wdt:P213 "%s" . |
53
|
|
|
?item wdt:P213 ?isni. |
54
|
|
|
?item wdt:P214 ?viaf. |
55
|
|
|
?article schema:about ?item ; |
56
|
|
|
schema:isPartOf <https://fr.wikipedia.org/> |
57
|
|
|
SERVICE wikibase:label { |
58
|
|
|
bd:serviceParam wikibase:language "fr" . |
59
|
|
|
} |
60
|
|
|
}', |
61
|
|
|
$isni |
62
|
|
|
); |
63
|
|
|
|
64
|
|
|
$result = $this->sparqlRequest($sparql); |
65
|
|
|
if ($result && isset($result['results']) && isset($result['results']) |
66
|
|
|
&& isset($result['results']['bindings']) |
67
|
|
|
&& count($result['results']['bindings']) === 1 |
68
|
|
|
) { |
69
|
|
|
return $result['results']['bindings'][0]; |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
return null; |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* @param string $sparql |
77
|
|
|
* |
78
|
|
|
* @return array|null |
79
|
|
|
* @throws Exception |
80
|
|
|
*/ |
81
|
|
|
private function sparqlRequest(string $sparql): ?array |
82
|
|
|
{ |
83
|
|
|
$url = 'https://query.wikidata.org/bigdata/namespace/wdq/sparql?'.http_build_query( |
84
|
|
|
[ |
85
|
|
|
'format' => 'json', |
86
|
|
|
'query' => urlencode($sparql), |
87
|
|
|
] |
88
|
|
|
); |
89
|
|
|
|
90
|
|
|
$response = $this->client->get($url); |
91
|
|
|
// todo : catch + return null ? |
92
|
|
|
if (200 !== $response->getStatusCode()) { |
93
|
|
|
throw new Exception('response error '.$response->getStatusCode()); |
94
|
|
|
} |
95
|
|
|
$json = $response->getBody()->getContents(); |
96
|
|
|
|
97
|
|
|
if (empty($json)) { |
98
|
|
|
return null; |
99
|
|
|
} |
100
|
|
|
$json = Normalizer::normalize($json); |
101
|
|
|
|
102
|
|
|
return json_decode($json, true) ?? null; |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
/** |
106
|
|
|
* todo move |
107
|
|
|
* |
108
|
|
|
* @param string $isni |
109
|
|
|
* |
110
|
|
|
* @return bool |
111
|
|
|
*/ |
112
|
|
|
private function ISNIvalide(string $isni): bool |
113
|
|
|
{ |
114
|
|
|
return (!preg_match('#^0000(000[0-4])([0-9]{4})([0-9]{3}[0-9X])$#', $isni)) ? false : true; |
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
|