|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Wikibase\Lib\Maintenance; |
|
4
|
|
|
|
|
5
|
|
|
use Maintenance; |
|
6
|
|
|
use MediaWiki\MediaWikiServices; |
|
7
|
|
|
use MWException; |
|
8
|
|
|
use Wikibase\Lib\Sites\SiteMatrixParser; |
|
9
|
|
|
use Wikibase\Lib\Sites\SitesBuilder; |
|
10
|
|
|
|
|
11
|
|
|
$basePath = getenv( 'MW_INSTALL_PATH' ) !== false |
|
12
|
|
|
? getenv( 'MW_INSTALL_PATH' ) |
|
13
|
|
|
: __DIR__ . '/../../../..'; |
|
14
|
|
|
|
|
15
|
|
|
require_once $basePath . '/maintenance/Maintenance.php'; |
|
16
|
|
|
|
|
17
|
|
|
if ( !class_exists( SitesBuilder::class ) ) { |
|
18
|
|
|
require_once __DIR__ . '/../includes/Sites/SitesBuilder.php'; |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
|
|
if ( !class_exists( SiteMatrixParser::class ) ) { |
|
22
|
|
|
require_once __DIR__ . '/../includes/Sites/SiteMatrixParser.php'; |
|
23
|
|
|
} |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Maintenance script for populating the Sites table from another wiki that runs the |
|
27
|
|
|
* SiteMatrix extension. |
|
28
|
|
|
* |
|
29
|
|
|
* @note: this should move out of Wikibase |
|
30
|
|
|
* |
|
31
|
|
|
* @license GPL-2.0-or-later |
|
32
|
|
|
* @author Daniel Kinzler |
|
33
|
|
|
* @author Katie Filbert < [email protected] > |
|
34
|
|
|
*/ |
|
35
|
|
|
class PopulateSitesTable extends Maintenance { |
|
36
|
|
|
|
|
37
|
|
|
public function __construct() { |
|
38
|
|
|
parent::__construct(); |
|
39
|
|
|
|
|
40
|
|
|
$this->addDescription( 'Populate the sites table from another wiki that runs the SiteMatrix extension' ); |
|
41
|
|
|
|
|
42
|
|
|
$this->addOption( 'strip-protocols', "Strip http/https from URLs to make them protocol relative." ); |
|
43
|
|
|
$this->addOption( 'force-protocol', "Force a specific protocol for all URLs (like http/https).", false, true ); |
|
44
|
|
|
$this->addOption( 'load-from', "Full URL to the API of the wiki to fetch the site info from. " |
|
45
|
|
|
. "Default is https://meta.wikimedia.org/w/api.php", false, true ); |
|
46
|
|
|
$this->addOption( 'script-path', 'Script path to use for wikis in the site matrix. ' |
|
47
|
|
|
. ' (e.g. "/w/$1")', false, true ); |
|
48
|
|
|
$this->addOption( 'article-path', 'Article path for wikis in the site matrix. ' |
|
49
|
|
|
. ' (e.g. "/wiki/$1")', false, true ); |
|
50
|
|
|
$this->addOption( 'site-group', 'Site group that this wiki is a member of. Used to populate ' |
|
51
|
|
|
. ' local interwiki identifiers in the site identifiers table. If not set and --wiki' |
|
52
|
|
|
. ' is set, the script will try to determine which site group the wiki is part of' |
|
53
|
|
|
. ' and populate interwiki ids for sites in that group.', false, true ); |
|
54
|
|
|
$this->addOption( 'no-expand-group', 'Do not expand site group codes in site matrix. ' |
|
55
|
|
|
. ' By default, "wiki" is expanded to "wikipedia".' ); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
public function execute() { |
|
59
|
|
|
$stripProtocols = (bool)$this->getOption( 'strip-protocols', false ); |
|
60
|
|
|
$forceProtocol = $this->getOption( 'force-protocol', null ); |
|
61
|
|
|
$url = $this->getOption( 'load-from', 'https://meta.wikimedia.org/w/api.php' ); |
|
62
|
|
|
$scriptPath = $this->getOption( 'script-path', '/w/$1' ); |
|
63
|
|
|
$articlePath = $this->getOption( 'article-path', '/wiki/$1' ); |
|
64
|
|
|
$expandGroup = !$this->getOption( 'no-expand-group', false ); |
|
65
|
|
|
$siteGroup = $this->getOption( 'site-group' ); |
|
66
|
|
|
$wikiId = $this->getOption( 'wiki' ); |
|
67
|
|
|
|
|
68
|
|
|
if ( $stripProtocols && is_string( $forceProtocol ) ) { |
|
69
|
|
|
$this->fatalError( "You can't use both strip-protocols and force-protocol" ); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
$protocol = true; |
|
73
|
|
|
if ( $stripProtocols ) { |
|
74
|
|
|
$protocol = false; |
|
75
|
|
|
} elseif ( is_string( $forceProtocol ) ) { |
|
76
|
|
|
$protocol = $forceProtocol; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
// @todo make it configurable, such as from a config file. |
|
80
|
|
|
$validGroups = [ 'wikipedia', 'wikivoyage', 'wikiquote', 'wiktionary', |
|
81
|
|
|
'wikibooks', 'wikisource', 'wikiversity', 'wikinews' ]; |
|
82
|
|
|
|
|
83
|
|
|
try { |
|
84
|
|
|
$json = $this->getSiteMatrixData( $url ); |
|
85
|
|
|
|
|
86
|
|
|
$siteMatrixParser = new SiteMatrixParser( $scriptPath, $articlePath, |
|
87
|
|
|
$protocol, $expandGroup ); |
|
88
|
|
|
|
|
89
|
|
|
$sites = $siteMatrixParser->sitesFromJson( $json ); |
|
90
|
|
|
|
|
91
|
|
|
$store = MediaWikiServices::getInstance()->getSiteStore(); |
|
92
|
|
|
$sitesBuilder = new SitesBuilder( $store, $validGroups ); |
|
93
|
|
|
$sitesBuilder->buildStore( $sites, $siteGroup, $wikiId ); |
|
94
|
|
|
|
|
95
|
|
|
} catch ( MWException $e ) { |
|
|
|
|
|
|
96
|
|
|
$this->output( $e->getMessage() ); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
$this->output( "done.\n" ); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* @param string $url |
|
104
|
|
|
* |
|
105
|
|
|
* @throws MWException |
|
106
|
|
|
* @return string |
|
107
|
|
|
*/ |
|
108
|
|
|
protected function getSiteMatrixData( $url ) { |
|
109
|
|
|
$url .= '?action=sitematrix&format=json'; |
|
110
|
|
|
|
|
111
|
|
|
$json = MediaWikiServices::getInstance()->getHttpRequestFactory()->get( $url, [], __METHOD__ ); |
|
112
|
|
|
|
|
113
|
|
|
if ( !$json ) { |
|
114
|
|
|
throw new MWException( "Got no data from $url\n" ); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
return $json; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
$maintClass = PopulateSitesTable::class; |
|
123
|
|
|
require_once RUN_MAINTENANCE_IF_MAIN; |
|
124
|
|
|
|
wikimedia /
mediawiki-extensions-Wikibase
Loading history...