|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PPP\Wikidata\Wikipedia; |
|
4
|
|
|
|
|
5
|
|
|
use InvalidArgumentException; |
|
6
|
|
|
use Mediawiki\Api\MediawikiApi; |
|
7
|
|
|
use Mediawiki\Api\SimpleRequest; |
|
8
|
|
|
use OutOfBoundsException; |
|
9
|
|
|
use PPP\Wikidata\Cache\PerSiteLinkCache; |
|
10
|
|
|
use Wikibase\DataModel\SiteLink; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* @licence AGPLv3+ |
|
14
|
|
|
* @author Thomas Pellissier Tanon |
|
15
|
|
|
*/ |
|
16
|
|
|
abstract class PerSiteLinkProvider { |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* @var PerSiteLinkCache |
|
20
|
|
|
*/ |
|
21
|
|
|
private $cache; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @var MediawikiApi[] |
|
25
|
|
|
*/ |
|
26
|
|
|
private $apis; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @param MediawikiApi[] $apis |
|
30
|
|
|
* @param PerSiteLinkCache $cache |
|
31
|
|
|
*/ |
|
32
|
|
|
public function __construct(array $apis, PerSiteLinkCache $cache) { |
|
33
|
|
|
$this->apis = $apis; |
|
34
|
|
|
$this->cache = $cache; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @param SiteLink $siteLink |
|
39
|
|
|
* @return SiteLinkProvider |
|
40
|
|
|
*/ |
|
41
|
|
|
protected function getForSiteLink(SiteLink $siteLink) { |
|
42
|
|
|
try { |
|
43
|
|
|
return $this->cache->fetch($siteLink); |
|
44
|
|
|
} catch(OutOfBoundsException $e) { |
|
45
|
|
|
$headers = $this->getFromWiki($siteLink->getSiteId(), array($siteLink->getPageName())); |
|
46
|
|
|
if(empty($headers)) { |
|
47
|
|
|
throw new OutOfBoundsException('The page ' . $siteLink->getPageName() . ' of ' . $siteLink->getSiteId() . ' does not exists'); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
return reset($headers); |
|
51
|
|
|
} |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* Makes sure that data are loaded into the cache |
|
56
|
|
|
* |
|
57
|
|
|
* @param SiteLink[] $titles |
|
58
|
|
|
*/ |
|
59
|
|
|
public function loadFromSiteLinks(array $titles) { |
|
60
|
|
|
$titlesPerWiki = array(); |
|
61
|
|
|
|
|
62
|
|
|
foreach($titles as $title) { |
|
63
|
|
|
if(!$this->cache->contains($title) && $this->isWikiIdSupported($title->getSiteId())) { |
|
64
|
|
|
$titlesPerWiki[$title->getSiteId()][] = $title->getPageName(); |
|
65
|
|
|
} |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
foreach($titlesPerWiki as $wikiId => $titles) { |
|
69
|
|
|
foreach($this->getFromWiki($wikiId, $titles) as $articleHeader) { |
|
70
|
|
|
$this->cache->save($articleHeader); |
|
71
|
|
|
} |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
private function getFromWiki($wikiId, $titles) { |
|
76
|
|
|
if(empty($titles)) { |
|
77
|
|
|
return array(); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
$request = $this->buildRequest($titles); |
|
81
|
|
|
$api = $this->getApiFromWikiId($wikiId); |
|
82
|
|
|
$finalResults = array(); |
|
83
|
|
|
$result = array('continue' => ''); |
|
84
|
|
|
do { |
|
85
|
|
|
$request['continue'] = $result['continue']; |
|
86
|
|
|
$result = $api->getRequest(new SimpleRequest($request['action'], $request)); |
|
87
|
|
|
$finalResults = array_merge($finalResults, $this->parseResult($wikiId, $titles, $result)); |
|
88
|
|
|
|
|
89
|
|
|
} while(array_key_exists('continue', $result)); |
|
90
|
|
|
|
|
91
|
|
|
return $finalResults; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* @param string[] $titles |
|
96
|
|
|
* @return array |
|
97
|
|
|
*/ |
|
98
|
|
|
protected abstract function buildRequest($titles); |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* @param string $wikiId |
|
102
|
|
|
* @param string[] $titles |
|
103
|
|
|
* @param array $result |
|
104
|
|
|
* @return SiteLinkProvider[] |
|
105
|
|
|
*/ |
|
106
|
|
|
protected abstract function parseResult($wikiId, $titles, $result); |
|
107
|
|
|
|
|
108
|
|
|
/** |
|
109
|
|
|
* @param string $wikiId |
|
110
|
|
|
* @return MediawikiApi |
|
111
|
|
|
*/ |
|
112
|
|
|
protected function getApiFromWikiId($wikiId) { |
|
113
|
|
|
if(!array_key_exists($wikiId, $this->apis)) { |
|
114
|
|
|
throw new InvalidArgumentException('Unknown wiki id: ' . $wikiId); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
return $this->apis[$wikiId]; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* @param string $wikiId |
|
122
|
|
|
* @return bool |
|
123
|
|
|
*/ |
|
124
|
|
|
public function isWikiIdSupported($wikiId) { |
|
125
|
|
|
return array_key_exists($wikiId, $this->apis); |
|
126
|
|
|
} |
|
127
|
|
|
} |
|
128
|
|
|
|
ProjetPP /
PPP-Wikidata
