maintenance/removeInvalidEmails.php

   1 <?php
   2
   3 use MediaWiki\Maintenance\Maintenance;
   4 use MediaWiki\Parser\Sanitizer;
   5 use MediaWiki\User\User;
   6
   7 // @codeCoverageIgnoreStart
   8 require_once __DIR__ . '/Maintenance.php';
   9 // @codeCoverageIgnoreEnd
  10
  11 /**
  12  * A script to remove emails that are invalid from
  13  * the user_email column of the user table. Emails
  14  * are validated before users can add them, but
  15  * this was not always the case so older users may
  16  * have invalid ones.
  17  *
  18  * By default it does a dry-run, pass --commit
  19  * to actually update the database.
  20  */
  21 class RemoveInvalidEmails extends Maintenance {
  22
  23         /** @var bool */
  24         private $commit = false;
  25
  26         public function __construct() {
  27                 parent::__construct();
  28                 $this->addOption( 'commit', 'Whether to actually update the database', false, false );
  29                 $this->setBatchSize( 500 );
  30         }
  31
  32         public function execute() {
  33                 $this->commit = $this->hasOption( 'commit' );
  34                 $dbr = $this->getReplicaDB();
  35                 $dbw = $this->getPrimaryDB();
  36                 $lastId = 0;
  37                 do {
  38                         $rows = $dbr->newSelectQueryBuilder()
  39                                 ->select( [ 'user_id', 'user_email' ] )
  40                                 ->from( 'user' )
  41                                 ->where( [
  42                                         $dbr->expr( 'user_id', '>', $lastId ),
  43                                         $dbr->expr( 'user_email', '!=', '' ),
  44                                         'user_email_authenticated' => null,
  45                                 ] )
  46                                 ->limit( $this->getBatchSize() )
  47                                 ->caller( __METHOD__ )->fetchResultSet();
  48                         $count = $rows->numRows();
  49                         $badIds = [];
  50                         foreach ( $rows as $row ) {
  51                                 if ( !Sanitizer::validateEmail( trim( $row->user_email ) ) ) {
  52                                         $this->output( "Found bad email: {$row->user_email} for user #{$row->user_id}\n" );
  53                                         $badIds[] = $row->user_id;
  54                                 }
  55                                 if ( $row->user_id > $lastId ) {
  56                                         $lastId = $row->user_id;
  57                                 }
  58                         }
  59
  60                         if ( $badIds ) {
  61                                 $badCount = count( $badIds );
  62                                 if ( $this->commit ) {
  63                                         $this->output( "Removing $badCount emails from the database.\n" );
  64                                         $dbw->newUpdateQueryBuilder()
  65                                                 ->update( 'user' )
  66                                                 ->set( [ 'user_email' => '' ] )
  67                                                 ->where( [ 'user_id' => $badIds ] )
  68                                                 ->caller( __METHOD__ )
  69                                                 ->execute();
  70                                         foreach ( $badIds as $badId ) {
  71                                                 User::newFromId( $badId )->invalidateCache();
  72                                         }
  73                                         $this->waitForReplication();
  74                                 } else {
  75                                         $this->output( "Would have removed $badCount emails from the database.\n" );
  76
  77                                 }
  78                         }
  79                 } while ( $count !== 0 );
  80                 $this->output( "Done.\n" );
  81         }
  82 }
  83
  84 // @codeCoverageIgnoreStart
  85 $maintClass = RemoveInvalidEmails::class;
  86 require_once RUN_MAINTENANCE_IF_MAIN;
  87 // @codeCoverageIgnoreEnd