|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace App\Repository; |
|
6
|
|
|
|
|
7
|
|
|
use GuzzleHttp\Client as GuzzleClient; |
|
8
|
|
|
use Psr\Cache\CacheItemPoolInterface; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* A WikiDomainRepository is responsible for fetching a list of all wikis from the Sitematrix API. |
|
12
|
|
|
*/ |
|
13
|
|
|
class WikiDomainRepository |
|
14
|
|
|
{ |
|
15
|
|
|
/** @var GuzzleClient */ |
|
16
|
|
|
private $guzzle; |
|
17
|
|
|
|
|
18
|
|
|
/** @var CacheItemPoolInterface */ |
|
19
|
|
|
private $cache; |
|
20
|
|
|
|
|
21
|
|
|
private const CACHE_KEY = 'global-search-wikidomainlookup'; |
|
22
|
|
|
|
|
23
|
|
|
/** @var string Duration of cache for lookup table, as accepted by DateInterval::createFromDateString() */ |
|
24
|
|
|
private const CACHE_TIME = '10 hours'; |
|
25
|
|
|
|
|
26
|
|
|
public function __construct(GuzzleClient $guzzle, CacheItemPoolInterface $cache) |
|
27
|
|
|
{ |
|
28
|
|
|
$this->guzzle = $guzzle; |
|
29
|
|
|
$this->cache = $cache; |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* Load the site matrix, either from cache or using the sitematrix API. |
|
34
|
|
|
* @return string[] |
|
35
|
|
|
*/ |
|
36
|
|
|
public function load(): array |
|
37
|
|
|
{ |
|
38
|
|
|
$cacheItem = $this->cache->getItem(self::CACHE_KEY); |
|
39
|
|
|
$lookup = $cacheItem->get(); |
|
40
|
|
|
if (null === $lookup) { |
|
41
|
|
|
$lookup = $this->loadUncached(); |
|
42
|
|
|
$cacheItem->set($lookup) |
|
43
|
|
|
->expiresAfter(\DateInterval::createFromDateString(self::CACHE_TIME)); |
|
44
|
|
|
$this->cache->save($cacheItem); |
|
45
|
|
|
} |
|
46
|
|
|
return $lookup; |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* Fetch the site matrix from the API. |
|
51
|
|
|
* @return string[] |
|
52
|
|
|
*/ |
|
53
|
|
|
public function loadUncached(): array |
|
54
|
|
|
{ |
|
55
|
|
|
$res = $this->guzzle->request('GET', 'https://meta.wikimedia.org/w/api.php', [ |
|
56
|
|
|
'query' => [ |
|
57
|
|
|
'format' => 'json', |
|
58
|
|
|
'formatversion' => 2, |
|
59
|
|
|
'action' => 'sitematrix', |
|
60
|
|
|
'smlangprop' => 'site', |
|
61
|
|
|
'smsiteprop' => 'url|dbname', |
|
62
|
|
|
], |
|
63
|
|
|
]); |
|
64
|
|
|
$decoded = json_decode($res->getBody()->getContents(), true)['sitematrix']; |
|
65
|
|
|
$lookup = []; |
|
66
|
|
|
foreach ($decoded as $k => $v) { |
|
67
|
|
|
if ('count' === $k) { |
|
68
|
|
|
continue; |
|
69
|
|
|
} |
|
70
|
|
|
$sites = 'specials' === $k ? $v : $v['site']; |
|
71
|
|
|
foreach ($sites as $site) { |
|
72
|
|
|
$lookup[$site['dbname']] = parse_url($site['url'], PHP_URL_HOST); |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
return $lookup; |
|
76
|
|
|
} |
|
77
|
|
|
} |
|
78
|
|
|
|
MusikAnimal /
global-search
