maintenance/removeInvalidEmails.php

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