|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of dispositif/wikibot application (@github) |
|
4
|
|
|
* 2019/2020 © 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 DomainException; |
|
13
|
|
|
use Exception; |
|
14
|
|
|
use GuzzleHttp\Client; |
|
15
|
|
|
use Normalizer; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* dirty scratch WikiData read requests |
|
19
|
|
|
* Class WikidataAdapter |
|
20
|
|
|
* |
|
21
|
|
|
* @package App\Infrastructure |
|
22
|
|
|
*/ |
|
23
|
|
|
class WikidataAdapter |
|
24
|
|
|
{ |
|
25
|
|
|
private $client; |
|
26
|
|
|
|
|
27
|
2 |
|
public function __construct(?Client $client = null) |
|
28
|
|
|
{ |
|
29
|
2 |
|
if ($client === null) { |
|
30
|
|
|
// lazy dependency factory :) |
|
31
|
|
|
$this->client = new Client(['timeout' => 60, 'headers' => ['User-Agent' => getenv('USER_AGENT')]]); |
|
32
|
|
|
} else { |
|
33
|
2 |
|
$this->client = $client; |
|
34
|
|
|
} |
|
35
|
2 |
|
} |
|
36
|
|
|
|
|
37
|
|
|
public function getDataByInfos(?array $infos) |
|
38
|
|
|
{ |
|
39
|
|
|
$res = []; |
|
40
|
|
|
if (isset($infos['ISNIAuteur1'])) { |
|
41
|
|
|
$res = $this->searchByISNI($infos['ISNIAuteur1']); |
|
42
|
|
|
} |
|
43
|
|
|
if (isset($infos['isbn'])) { |
|
44
|
|
|
if(!empty($res)) { |
|
45
|
|
|
sleep(2); |
|
46
|
|
|
} |
|
47
|
|
|
$res = array_merge($res ?? [], $this->findArticleByISBN13($infos['isbn'])); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
return $res ?? []; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
1 |
|
public function findArticleByISBN13(string $isbn): ?array |
|
54
|
|
|
{ |
|
55
|
|
|
// strip ISBN formating |
|
56
|
1 |
|
$isbn = preg_replace('#[^0-9X]#', '', $isbn); |
|
57
|
1 |
|
if (strlen($isbn) !== 13) { |
|
58
|
|
|
throw new DomainException('ISBN-13 format error'); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
1 |
|
$sparql = sprintf( |
|
62
|
1 |
|
'select ?work ?workLabel ?articleBook ?edition ?isbn |
|
63
|
|
|
WHERE { |
|
64
|
|
|
?work wdt:P31 wd:Q47461344 ; # instance of written work |
|
65
|
|
|
wdt:P747 ?edition . # has edition (P747) |
|
66
|
|
|
?edition wdt:P212 $isbn . # ISBN-13 (P212) |
|
67
|
|
|
FILTER(REGEX(REPLACE(?isbn,"-",""), "%s", "i")). # strip ISBN formating |
|
68
|
|
|
?articleBook schema:about ?work ; |
|
69
|
|
|
schema:isPartOf <https://fr.wikipedia.org/> # frwiki sitelink |
|
70
|
|
|
SERVICE wikibase:label { |
|
71
|
|
|
bd:serviceParam wikibase:language "fr" . |
|
72
|
|
|
} |
|
73
|
|
|
}', |
|
74
|
1 |
|
$isbn |
|
75
|
|
|
); |
|
76
|
|
|
|
|
77
|
1 |
|
return $this->sparqlRequest($sparql); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* Get WD item, sitelink, VIAF from search by ISNI (author) |
|
82
|
|
|
* |
|
83
|
|
|
* @param string $isni |
|
84
|
|
|
* |
|
85
|
|
|
* @return array|null |
|
86
|
|
|
* @throws Exception |
|
87
|
|
|
*/ |
|
88
|
1 |
|
public function searchByISNI(string $isni): ?array |
|
89
|
|
|
{ |
|
90
|
1 |
|
if (!$this->ISNIvalide($isni)) { |
|
91
|
1 |
|
new Exception('Invalid format for ISNI'); |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
1 |
|
$sparql = sprintf( |
|
95
|
1 |
|
'SELECT distinct ?item ?itemLabel ?articleAuthor ?isni ?viaf WHERE { |
|
96
|
|
|
?item wdt:P213 "%s" . |
|
97
|
|
|
?item wdt:P213 ?isni. |
|
98
|
|
|
?item wdt:P214 ?viaf. |
|
99
|
|
|
?articleAuthor schema:about ?item ; |
|
100
|
|
|
schema:isPartOf <https://fr.wikipedia.org/> |
|
101
|
|
|
SERVICE wikibase:label { |
|
102
|
|
|
bd:serviceParam wikibase:language "fr" . |
|
103
|
|
|
} |
|
104
|
|
|
}', |
|
105
|
1 |
|
$isni |
|
106
|
|
|
); |
|
107
|
|
|
|
|
108
|
1 |
|
return $this->sparqlRequest($sparql); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* @param string $sparql |
|
113
|
|
|
* |
|
114
|
|
|
* @return array|null |
|
115
|
|
|
* @throws Exception |
|
116
|
|
|
*/ |
|
117
|
2 |
|
private function sparqlRequest(string $sparql): ?array |
|
118
|
|
|
{ |
|
119
|
2 |
|
$url = 'https://query.wikidata.org/bigdata/namespace/wdq/sparql?'.http_build_query( |
|
120
|
|
|
[ |
|
121
|
2 |
|
'format' => 'json', |
|
122
|
2 |
|
'query' => $sparql, // rawurlencode() |
|
123
|
|
|
] |
|
124
|
|
|
); |
|
125
|
|
|
|
|
126
|
|
|
|
|
127
|
|
|
// todo : catch + return null ? |
|
128
|
2 |
|
$response = $this->client->get($url); |
|
129
|
|
|
|
|
130
|
2 |
|
if (200 !== $response->getStatusCode()) { |
|
131
|
|
|
throw new Exception('response error '.$response->getStatusCode().' '.$response->getReasonPhrase()); |
|
132
|
|
|
} |
|
133
|
2 |
|
$json = $response->getBody()->getContents(); |
|
134
|
|
|
|
|
135
|
2 |
|
if (empty($json)) { |
|
136
|
|
|
return null; |
|
137
|
|
|
} |
|
138
|
2 |
|
$json = Normalizer::normalize($json); |
|
139
|
|
|
|
|
140
|
2 |
|
$array = json_decode($json, true) ?? null; |
|
141
|
|
|
|
|
142
|
|
|
// return first result only |
|
143
|
2 |
|
if ($array && isset($array['results']) && isset($array['results']) |
|
144
|
2 |
|
&& isset($array['results']['bindings']) |
|
145
|
2 |
|
&& count($array['results']['bindings']) === 1 |
|
146
|
|
|
) { |
|
147
|
2 |
|
return $array['results']['bindings'][0]; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
return null; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
/** |
|
154
|
|
|
* todo move |
|
155
|
|
|
* |
|
156
|
|
|
* @param string $isni |
|
157
|
|
|
* |
|
158
|
|
|
* @return bool |
|
159
|
|
|
*/ |
|
160
|
1 |
|
private function ISNIvalide(string $isni): bool |
|
161
|
|
|
{ |
|
162
|
1 |
|
return (!preg_match('#^0000(000[0-4])([0-9]{4})([0-9]{3}[0-9X])$#', $isni)) ? false : true; |
|
163
|
|
|
} |
|
164
|
|
|
} |
|
165
|
|
|
|
Dispositif /
Wikibot
