userDupes.inc

Go to the documentation of this file.

<?php
use Wikimedia\Rdbms\IMaintainableDatabase;
 
class UserDupes {
    private $db;
    private $reassigned;
    private $trimmed;
    private $failed;
    private $outputCallback;
 
    public function __construct( &$database, $outputCallback ) {
        $this->db = $database;
        $this->outputCallback = $outputCallback;
    }
 
    private function out( $str ) {
        call_user_func( $this->outputCallback, $str );
    }
 
    public function hasUniqueIndex() {
        $info = $this->db->indexInfo( 'user', 'user_name', __METHOD__ );
        if ( !$info ) {
            $this->out( "WARNING: doesn't seem to have user_name index at all!\n" );
 
            return false;
        }
 
        # Confusingly, 'Non_unique' is 0 for *unique* indexes,
        # and 1 for *non-unique* indexes. Pass the crack, MySQL,
        # it's obviously some good stuff!
        return ( $info[0]->Non_unique == 0 );
    }
 
    public function clearDupes() {
        return $this->checkDupes( true );
    }
 
    private function checkDupes( $doDelete = false ) {
        $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
        if ( $this->hasUniqueIndex() ) {
            echo "$dbDomain already has a unique index on its user table.\n";
 
            return true;
        }
 
        $this->lock();
 
        $this->out( "Checking for duplicate accounts...\n" );
        $dupes = $this->getDupes();
        $count = count( $dupes );
 
        $this->out( "Found $count accounts with duplicate records on $dbDomain.\n" );
        $this->trimmed = 0;
        $this->reassigned = 0;
        $this->failed = 0;
        foreach ( $dupes as $name ) {
            $this->examine( $name, $doDelete );
        }
 
        $this->unlock();
 
        $this->out( "\n" );
 
        // @phan-suppress-next-line PhanSuspiciousValueComparison
        if ( $this->reassigned > 0 ) {
            if ( $doDelete ) {
                $this->out( "$this->reassigned duplicate accounts had edits "
                    . "reassigned to a canonical record id.\n" );
            } else {
                $this->out( "$this->reassigned duplicate accounts need to have edits reassigned.\n" );
            }
        }
 
        // @phan-suppress-next-line PhanSuspiciousValueComparison
        if ( $this->trimmed > 0 ) {
            if ( $doDelete ) {
                $this->out(
                    "$this->trimmed duplicate user records were deleted from $dbDomain.\n" );
            } else {
                $this->out(
                    "$this->trimmed duplicate user accounts were found on $dbDomain " .
                    "which can be removed safely.\n"
                );
            }
        }
 
        // @phan-suppress-next-line PhanSuspiciousValueComparison
        if ( $this->failed > 0 ) {
            $this->out( "Something terribly awry; $this->failed duplicate accounts were not removed.\n" );
 
            return false;
        }
 
        // @phan-suppress-next-line PhanSuspiciousValueComparison
        if ( $this->trimmed == 0 || $doDelete ) {
            $this->out( "It is now safe to apply the unique index on user_name.\n" );
 
            return true;
        } else {
            $this->out( "Run this script again with the --fix option to automatically delete them.\n" );
 
            return false;
        }
    }
 
    private function lock() {
        $set = [ 'user', 'revision' ];
        $names = array_map( [ $this, 'lockTable' ], $set );
        $tables = implode( ',', $names );
 
        $this->db->query( "LOCK TABLES $tables", __METHOD__ );
    }
 
    private function lockTable( $table ) {
        return $this->db->tableName( $table ) . ' WRITE';
    }
 
    private function unlock() {
        $this->db->query( "UNLOCK TABLES", __METHOD__ );
    }
 
    private function getDupes() {
        $user = $this->db->tableName( 'user' );
        $result = $this->db->query(
            "SELECT user_name,COUNT(*) AS n
                FROM $user
            GROUP BY user_name
                HAVING n > 1", __METHOD__ );
 
        $list = [];
        foreach ( $result as $row ) {
            $list[] = $row->user_name;
        }
 
        return $list;
    }
 
    private function examine( $name, $doDelete ) {
        $result = $this->db->select( 'user',
            [ 'user_id' ],
            [ 'user_name' => $name ],
            __METHOD__ );
 
        $firstRow = $this->db->fetchObject( $result );
        $firstId = $firstRow->user_id;
        $this->out( "Record that will be used for '$name' is user_id=$firstId\n" );
 
        foreach ( $result as $row ) {
            $dupeId = $row->user_id;
            $this->out( "... dupe id $dupeId: " );
            $edits = $this->editCount( $dupeId );
            if ( $edits > 0 ) {
                $this->reassigned++;
                $this->out( "has $edits edits! " );
                if ( $doDelete ) {
                    $this->reassignEdits( $dupeId, $firstId );
                    $newEdits = $this->editCount( $dupeId );
                    if ( $newEdits == 0 ) {
                        $this->out( "confirmed cleaned. " );
                    } else {
                        $this->failed++;
                        $this->out( "WARNING! $newEdits remaining edits for $dupeId; NOT deleting user.\n" );
                        continue;
                    }
                } else {
                    $this->out( "(will need to reassign edits on fix)" );
                }
            } else {
                $this->out( "ok, no edits. " );
            }
            $this->trimmed++;
            if ( $doDelete ) {
                $this->trimAccount( $dupeId );
            }
            $this->out( "\n" );
        }
    }
 
    private function editCount( $userid ) {
        return intval( $this->db->selectField(
            'revision',
            'COUNT(*)',
            [ 'rev_user' => $userid ],
            __METHOD__ ) );
    }
 
    private function reassignEdits( $from, $to ) {
        $this->out( 'reassigning... ' );
        $this->db->update( 'revision',
            [ 'rev_user' => $to ],
            [ 'rev_user' => $from ],
            __METHOD__ );
        $this->out( "ok. " );
    }
 
    private function trimAccount( $userid ) {
        $this->out( "deleting..." );
        $this->db->delete( 'user', [ 'user_id' => $userid ], __METHOD__ );
        $this->out( " ok" );
    }
}