|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PPP\Wikidata\Wikipedia; |
|
4
|
|
|
|
|
5
|
|
|
use Wikibase\DataModel\SiteLink; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* @licence AGPLv3+ |
|
9
|
|
|
* @author Thomas Pellissier Tanon |
|
10
|
|
|
*/ |
|
11
|
|
|
class MediawikiArticleProvider extends PerSiteLinkProvider { |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* @param SiteLink $siteLink |
|
15
|
|
|
* @return MediawikiArticle |
|
16
|
|
|
*/ |
|
17
|
5 |
|
public function getArticleForSiteLink(SiteLink $siteLink) { |
|
18
|
5 |
|
return $this->getForSiteLink($siteLink); |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
4 |
|
protected function buildRequest($titles) { |
|
22
|
|
|
return array( |
|
23
|
4 |
|
'action' => 'query', |
|
24
|
4 |
|
'titles' => implode('|', $titles), |
|
25
|
4 |
|
'prop' => 'extracts|info|pageimages', |
|
26
|
4 |
|
'inprop' => 'url', |
|
27
|
4 |
|
'redirects' => true, |
|
28
|
4 |
|
'exintro' => true, |
|
29
|
4 |
|
'exsectionformat' => 'plain', |
|
30
|
4 |
|
'explaintext' => true, |
|
31
|
4 |
|
'exsentences' => 3, |
|
32
|
4 |
|
'exlimit' => 20, |
|
33
|
4 |
|
'piprop' => 'thumbnail|name', |
|
34
|
4 |
|
'pithumbsize' => 300, |
|
35
|
|
|
'pilimit' => 20 |
|
36
|
4 |
|
); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
4 |
|
protected function parseResult($wikiId, $titles, $result) { |
|
40
|
4 |
|
$articleHeaders = array(); |
|
41
|
|
|
|
|
42
|
4 |
|
foreach($result['query']['pages'] as $pageResult) { |
|
43
|
3 |
|
if(array_key_exists('extract', $pageResult)) { |
|
44
|
3 |
|
$image = null; |
|
45
|
|
|
|
|
46
|
3 |
|
if(array_key_exists('thumbnail', $pageResult)) { |
|
47
|
1 |
|
$image = new MediawikiArticleImage( |
|
48
|
1 |
|
$pageResult['thumbnail']['source'], |
|
49
|
1 |
|
$pageResult['thumbnail']['width'], |
|
50
|
1 |
|
$pageResult['thumbnail']['height'], |
|
51
|
1 |
|
str_replace('_', ' ', $pageResult['pageimage']) |
|
52
|
1 |
|
); |
|
53
|
1 |
|
} |
|
54
|
|
|
|
|
55
|
3 |
|
$articleHeaders[] = new MediawikiArticle( |
|
56
|
3 |
|
new SiteLink($wikiId, $pageResult['title']), |
|
57
|
3 |
|
$pageResult['extract'], |
|
58
|
3 |
|
$pageResult['pagelanguage'], |
|
59
|
3 |
|
$pageResult['canonicalurl'], |
|
60
|
|
|
$image |
|
61
|
3 |
|
); |
|
62
|
|
|
|
|
63
|
3 |
|
} |
|
64
|
4 |
|
} |
|
65
|
|
|
|
|
66
|
4 |
|
return $articleHeaders; |
|
67
|
|
|
} |
|
68
|
|
|
} |
|
69
|
|
|
|