maintenance/removeInvalidEmails.php

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