MediaWiki  master
refreshExternallinksIndex.php
Go to the documentation of this file.
1 <?php
24 require_once __DIR__ . '/Maintenance.php';
25 
34  public function __construct() {
35  parent::__construct();
36  $this->addDescription(
37  'Refresh the externallinks table el_index and el_index_60 from el_to' );
38  $this->setBatchSize( 10000 );
39  }
40 
41  protected function getUpdateKey() {
42  return static::class
43  . ' v' . LinkFilter::VERSION
44  . ( LinkFilter::supportsIDN() ? '+' : '-' ) . 'IDN';
45  }
46 
47  protected function updateSkippedMessage() {
48  return 'externallinks table indexes up to date';
49  }
50 
51  protected function doDBUpdates() {
52  $dbw = $this->getDB( DB_MASTER );
53  if ( !$dbw->tableExists( 'externallinks' ) ) {
54  $this->error( "externallinks table does not exist" );
55  return false;
56  }
57  $this->output( "Updating externallinks table index fields\n" );
58 
59  $minmax = $dbw->selectRow(
60  'externallinks',
61  [ 'min' => 'MIN(el_id)', 'max' => 'MAX(el_id)' ],
62  '',
63  __METHOD__
64  );
65 
66  $updated = 0;
67  $deleted = 0;
68  $start = $minmax->min - 1;
69  $last = $minmax->max;
70  while ( $start < $last ) {
71  $end = min( $start + $this->mBatchSize, $last );
72  $this->output( "el_id $start - $end of $last\n" );
73  $res = $dbw->select( 'externallinks', [ 'el_id', 'el_to', 'el_index' ],
74  [
75  "el_id > $start",
76  "el_id <= $end",
77  ],
78  __METHOD__,
79  [ 'ORDER BY' => 'el_id' ]
80  );
81  foreach ( $res as $row ) {
82  $newIndexes = LinkFilter::makeIndexes( $row->el_to );
83  if ( !$newIndexes ) {
84  $dbw->delete( 'externallinks', [ 'el_id' => $row->el_id ], __METHOD__ );
85  $deleted++;
86  continue;
87  }
88  if ( in_array( $row->el_index, $newIndexes, true ) ) {
89  continue;
90  }
91 
92  if ( count( $newIndexes ) === 1 ) {
93  $newIndex = $newIndexes[0];
94  } else {
95  // Assume the scheme is the only difference between the different $newIndexes.
96  // Keep this row's scheme, assuming there's another row with the other scheme.
97  $newIndex = substr( $row->el_index, 0, strpos( $row->el_index, ':' ) ) .
98  substr( $newIndexes[0], strpos( $newIndexes[0], ':' ) );
99  }
100  $dbw->update( 'externallinks',
101  [
102  'el_index' => $newIndex,
103  'el_index_60' => substr( $newIndex, 0, 60 ),
104  ],
105  [ 'el_id' => $row->el_id ],
106  __METHOD__
107  );
108  $updated++;
109  }
110  wfWaitForSlaves();
111  $start = $end;
112  }
113  $this->output( "Done, $updated rows updated, $deleted deleted.\n" );
114 
115  return true;
116  }
117 }
118 
119 $maintClass = "RefreshExternallinksIndex";
120 require_once RUN_MAINTENANCE_IF_MAIN;
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:39
error( $err, $die=0)
Throw an error to the user.
setBatchSize( $s=0)
Set the batch size.
const DB_MASTER
Definition: defines.php:26
$last
Maintenance script that refreshes the externallinks table el_index and el_index_60 from el_to...
wfWaitForSlaves( $ifWritesSince=null, $wiki=false, $cluster=false, $timeout=null)
Waits for the replica DBs to catch up to the master position.
addDescription( $text)
Set the description text.
output( $out, $channel=null)
Throw some output to the user.
static supportsIDN()
Indicate whether LinkFilter IDN support is available.
Definition: LinkFilter.php:88
Class for scripts that perform database maintenance and want to log the update in updatelog so we can...
static makeIndexes( $url)
Converts a URL into a format for el_index.
Definition: LinkFilter.php:171
const VERSION
Increment this when makeIndexes output changes.
Definition: LinkFilter.php:39
getDB( $db, $groups=[], $dbDomain=false)
Returns a database to be used by current maintenance script.