Code Coverage
 
Lines
Functions and Methods
Classes and Traits
Total
0.00% covered (danger)
0.00%
0 / 79
0.00% covered (danger)
0.00%
0 / 4
CRAP
0.00% covered (danger)
0.00%
0 / 1
MigrateExternallinks
0.00% covered (danger)
0.00%
0 / 79
0.00% covered (danger)
0.00%
0 / 4
156
0.00% covered (danger)
0.00%
0 / 1
 __construct
0.00% covered (danger)
0.00%
0 / 11
0.00% covered (danger)
0.00%
0 / 1
2
 getUpdateKey
0.00% covered (danger)
0.00%
0 / 1
0.00% covered (danger)
0.00%
0 / 1
2
 doDBUpdates
0.00% covered (danger)
0.00%
0 / 27
0.00% covered (danger)
0.00%
0 / 1
30
 handleBatch
0.00% covered (danger)
0.00%
0 / 40
0.00% covered (danger)
0.00%
0 / 1
30
1<?php
2
3use MediaWiki\Deferred\LinksUpdate\ExternalLinksTable;
4use MediaWiki\ExternalLinks\LinkFilter;
5use MediaWiki\Maintenance\LoggedUpdateMaintenance;
6
7// @codeCoverageIgnoreStart
8require_once __DIR__ . '/Maintenance.php';
9// @codeCoverageIgnoreEnd
10
11/**
12 * Maintenance script that migrates externallinks data
13 *
14 * @ingroup Maintenance
15 * @since 1.40
16 */
17class MigrateExternallinks extends LoggedUpdateMaintenance {
18    public function __construct() {
19        parent::__construct();
20        $this->addDescription(
21            'Migrate externallinks data'
22        );
23        $this->addOption(
24            'sleep',
25            'Sleep time (in seconds) between every batch. Default: 0',
26            false,
27            true
28        );
29        $this->setBatchSize( 1000 );
30    }
31
32    /** @inheritDoc */
33    protected function getUpdateKey() {
34        return __CLASS__;
35    }
36
37    /** @inheritDoc */
38    protected function doDBUpdates() {
39        /** @var \Wikimedia\Rdbms\Database $dbw */
40        $dbw = $this->getServiceContainer()->getConnectionProvider()->getPrimaryDatabase(
41            ExternalLinksTable::VIRTUAL_DOMAIN
42        );
43        '@phan-var \Wikimedia\Rdbms\Database $dbw';
44        $table = 'externallinks';
45        if ( !$dbw->fieldExists( $table, 'el_to', __METHOD__ ) ) {
46            $this->output( "Old fields don't exist. There is no need to run this script\n" );
47            return true;
48        }
49        if ( !$dbw->fieldExists( $table, 'el_to_path', __METHOD__ ) ) {
50            $this->output( "Run update.php to create the el_to_path column.\n" );
51            return false;
52        }
53
54        $this->output( "Populating el_to_domain_index and el_to_path columns\n" );
55        $updated = 0;
56
57        $highestId = $dbw->newSelectQueryBuilder()
58            ->select( 'el_id' )
59            ->from( $table )
60            ->caller( __METHOD__ )
61            ->orderBy( 'el_id', 'DESC' )
62            ->fetchField();
63        if ( !$highestId ) {
64            $this->output( "Page table is empty.\n" );
65            return true;
66        }
67        $id = 0;
68        while ( $id <= $highestId ) {
69            $updated += $this->handleBatch( $id );
70            $id += $this->getBatchSize();
71        }
72
73        $this->output( "Completed normalization of $table$updated rows updated.\n" );
74
75        return true;
76    }
77
78    private function handleBatch( int $lowId ): int {
79        $batchSize = $this->getBatchSize();
80        // range is inclusive, let's subtract one.
81        $highId = $lowId + $batchSize - 1;
82        $dbw = $this->getServiceContainer()->getConnectionProvider()->getPrimaryDatabase(
83            ExternalLinksTable::VIRTUAL_DOMAIN
84        );
85        $updated = 0;
86        $res = $dbw->newSelectQueryBuilder()
87            ->select( [ 'el_id', 'el_to' ] )
88            ->from( 'externallinks' )
89            ->where( [
90                'el_to_domain_index' => '',
91                $dbw->expr( 'el_id', '>=', $lowId ),
92                $dbw->expr( 'el_id', '<=', $highId ),
93            ] )
94            ->limit( $batchSize )
95            ->caller( __METHOD__ )
96            ->fetchResultSet();
97        if ( !$res->numRows() ) {
98            return $updated;
99        }
100        foreach ( $res as $row ) {
101            $url = $row->el_to;
102            $paths = LinkFilter::makeIndexes( $url );
103            if ( !$paths ) {
104                continue;
105            }
106            $dbw->newUpdateQueryBuilder()
107                ->update( 'externallinks' )
108                // just take the first one, we are not sending proto-relative to LinkFilter
109                ->set( [
110                    'el_to_domain_index' => substr( $paths[0][0], 0, 255 ),
111                    'el_to_path' => $paths[0][1]
112                ] )
113                ->where( [ 'el_id' => $row->el_id ] )
114                ->caller( __METHOD__ )->execute();
115
116            $updated += $dbw->affectedRows();
117        }
118        $this->output( "Updated $updated rows\n" );
119        // Sleep between batches for replication to catch up
120        $this->waitForReplication();
121        $sleep = (int)$this->getOption( 'sleep', 0 );
122        if ( $sleep > 0 ) {
123            sleep( $sleep );
124        }
125        return $updated;
126    }
127
128}
129
130// @codeCoverageIgnoreStart
131$maintClass = MigrateExternallinks::class;
132require_once RUN_MAINTENANCE_IF_MAIN;
133// @codeCoverageIgnoreEnd