maintenance/cleanupTable.inc

   1 <?php
   2 /**
   3  * Generic class to cleanup a database table.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Maintenance
  22  */
  23
  24 require_once __DIR__ . '/Maintenance.php';
  25
  26 /**
  27  * Generic class to cleanup a database table. Already subclasses Maintenance.
  28  *
  29  * @ingroup Maintenance
  30  */
  31 class TableCleanup extends Maintenance {
  32         protected $defaultParams = [
  33                 'table' => 'page',
  34                 'conds' => [],
  35                 'index' => 'page_id',
  36                 'callback' => 'processRow',
  37         ];
  38
  39         protected $dryrun = false;
  40         public $batchSize = 100;
  41         public $reportInterval = 100;
  42
  43         protected $processed, $updated, $count, $startTime, $table;
  44
  45         public function __construct() {
  46                 parent::__construct();
  47                 $this->addOption( 'dry-run', 'Perform a dry run' );
  48         }
  49
  50         public function execute() {
  51                 global $wgUser;
  52                 $this->dryrun = $this->hasOption( 'dry-run' );
  53                 if ( $this->dryrun ) {
  54                         $wgUser = User::newFromName( 'Conversion script' );
  55                         $this->output( "Checking for bad titles...\n" );
  56                 } else {
  57                         $wgUser = User::newSystemUser( 'Conversion script', [ 'steal' => true ] );
  58                         $this->output( "Checking and fixing bad titles...\n" );
  59                 }
  60                 $this->runTable( $this->defaultParams );
  61         }
  62
  63         protected function init( $count, $table ) {
  64                 $this->processed = 0;
  65                 $this->updated = 0;
  66                 $this->count = $count;
  67                 $this->startTime = microtime( true );
  68                 $this->table = $table;
  69         }
  70
  71         /**
  72          * @param int $updated
  73          */
  74         protected function progress( $updated ) {
  75                 $this->updated += $updated;
  76                 $this->processed++;
  77                 if ( $this->processed % $this->reportInterval != 0 ) {
  78                         return;
  79                 }
  80                 $portion = $this->processed / $this->count;
  81                 $updateRate = $this->updated / $this->processed;
  82
  83                 $now = microtime( true );
  84                 $delta = $now - $this->startTime;
  85                 $estimatedTotalTime = $delta / $portion;
  86                 $eta = $this->startTime + $estimatedTotalTime;
  87
  88                 $this->output(
  89                         sprintf( "%s %s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec <%.2f%% updated>\n",
  90                                 wfWikiID(),
  91                                 wfTimestamp( TS_DB, intval( $now ) ),
  92                                 $portion * 100.0,
  93                                 $this->table,
  94                                 wfTimestamp( TS_DB, intval( $eta ) ),
  95                                 $this->processed,
  96                                 $this->count,
  97                                 $this->processed / $delta,
  98                                 $updateRate * 100.0
  99                         )
 100                 );
 101                 flush();
 102         }
 103
 104         /**
 105          * @param array $params
 106          * @throws MWException
 107          */
 108         public function runTable( $params ) {
 109                 $dbr = $this->getDB( DB_REPLICA );
 110
 111                 if ( array_diff( array_keys( $params ),
 112                         [ 'table', 'conds', 'index', 'callback' ] )
 113                 ) {
 114                         throw new MWException( __METHOD__ . ': Missing parameter ' . implode( ', ', $params ) );
 115                 }
 116
 117                 $table = $params['table'];
 118                 // count(*) would melt the DB for huge tables, we can estimate here
 119                 $count = $dbr->estimateRowCount( $table, '*', '', __METHOD__ );
 120                 $this->init( $count, $table );
 121                 $this->output( "Processing $table...\n" );
 122
 123                 $index = (array)$params['index'];
 124                 $indexConds = [];
 125                 $options = [
 126                         'ORDER BY' => implode( ',', $index ),
 127                         'LIMIT' => $this->batchSize
 128                 ];
 129                 $callback = [ $this, $params['callback'] ];
 130
 131                 while ( true ) {
 132                         $conds = array_merge( $params['conds'], $indexConds );
 133                         $res = $dbr->select( $table, '*', $conds, __METHOD__, $options );
 134                         if ( !$res->numRows() ) {
 135                                 // Done
 136                                 break;
 137                         }
 138
 139                         foreach ( $res as $row ) {
 140                                 call_user_func( $callback, $row );
 141                         }
 142
 143                         if ( $res->numRows() < $this->batchSize ) {
 144                                 // Done
 145                                 break;
 146                         }
 147
 148                         // Update the conditions to select the next batch.
 149                         // Construct a condition string by starting with the least significant part
 150                         // of the index, and adding more significant parts progressively to the left
 151                         // of the string.
 152                         $nextCond = '';
 153                         foreach ( array_reverse( $index ) as $field ) {
 154                                 $encValue = $dbr->addQuotes( $row->$field );
 155                                 if ( $nextCond === '' ) {
 156                                         $nextCond = "$field > $encValue";
 157                                 } else {
 158                                         $nextCond = "$field > $encValue OR ($field = $encValue AND ($nextCond))";
 159                                 }
 160                         }
 161                         $indexConds = [ $nextCond ];
 162                 }
 163
 164                 $this->output( "Finished $table... $this->updated of $this->processed rows updated\n" );
 165         }
 166
 167         /**
 168          * @param array $matches
 169          * @return string
 170          */
 171         protected function hexChar( $matches ) {
 172                 return sprintf( "\\x%02x", ord( $matches[1] ) );
 173         }
 174 }