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 = array(
  33                 'table' => 'page',
  34                 'conds' => array(),
  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                 $wgUser = User::newFromName( 'Conversion script' );
  53                 $this->dryrun = $this->hasOption( 'dry-run' );
  54                 if ( $this->dryrun ) {
  55                         $this->output( "Checking for bad titles...\n" );
  56                 } else {
  57                         $this->output( "Checking and fixing bad titles...\n" );
  58                 }
  59                 $this->runTable( $this->defaultParams );
  60         }
  61
  62         protected function init( $count, $table ) {
  63                 $this->processed = 0;
  64                 $this->updated = 0;
  65                 $this->count = $count;
  66                 $this->startTime = microtime( true );
  67                 $this->table = $table;
  68         }
  69
  70         /**
  71          * @param int $updated
  72          */
  73         protected function progress( $updated ) {
  74                 $this->updated += $updated;
  75                 $this->processed++;
  76                 if ( $this->processed % $this->reportInterval != 0 ) {
  77                         return;
  78                 }
  79                 $portion = $this->processed / $this->count;
  80                 $updateRate = $this->updated / $this->processed;
  81
  82                 $now = microtime( true );
  83                 $delta = $now - $this->startTime;
  84                 $estimatedTotalTime = $delta / $portion;
  85                 $eta = $this->startTime + $estimatedTotalTime;
  86
  87                 $this->output(
  88                         sprintf( "%s %s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec <%.2f%% updated>\n",
  89                                 wfWikiID(),
  90                                 wfTimestamp( TS_DB, intval( $now ) ),
  91                                 $portion * 100.0,
  92                                 $this->table,
  93                                 wfTimestamp( TS_DB, intval( $eta ) ),
  94                                 $this->processed,
  95                                 $this->count,
  96                                 $this->processed / $delta,
  97                                 $updateRate * 100.0
  98                         )
  99                 );
 100                 flush();
 101         }
 102
 103         /**
 104          * @param array $params
 105          * @throws MWException
 106          */
 107         public function runTable( $params ) {
 108                 $dbr = wfGetDB( DB_SLAVE );
 109
 110                 if ( array_diff( array_keys( $params ),
 111                         array( 'table', 'conds', 'index', 'callback' ) )
 112                 ) {
 113                         throw new MWException( __METHOD__ . ': Missing parameter ' . implode( ', ', $params ) );
 114                 }
 115
 116                 $table = $params['table'];
 117                 // count(*) would melt the DB for huge tables, we can estimate here
 118                 $count = $dbr->estimateRowCount( $table, '*', '', __METHOD__ );
 119                 $this->init( $count, $table );
 120                 $this->output( "Processing $table...\n" );
 121
 122                 $index = (array)$params['index'];
 123                 $indexConds = array();
 124                 $options = array(
 125                         'ORDER BY' => implode( ',', $index ),
 126                         'LIMIT' => $this->batchSize
 127                 );
 128                 $callback = array( $this, $params['callback'] );
 129
 130                 while ( true ) {
 131                         $conds = array_merge( $params['conds'], $indexConds );
 132                         $res = $dbr->select( $table, '*', $conds, __METHOD__, $options );
 133                         if ( !$res->numRows() ) {
 134                                 // Done
 135                                 break;
 136                         }
 137
 138                         foreach ( $res as $row ) {
 139                                 call_user_func( $callback, $row );
 140                         }
 141
 142                         if ( $res->numRows() < $this->batchSize ) {
 143                                 // Done
 144                                 break;
 145                         }
 146
 147                         // Update the conditions to select the next batch.
 148                         // Construct a condition string by starting with the least significant part
 149                         // of the index, and adding more significant parts progressively to the left
 150                         // of the string.
 151                         $nextCond = '';
 152                         foreach ( array_reverse( $index ) as $field ) {
 153                                 $encValue = $dbr->addQuotes( $row->$field );
 154                                 if ( $nextCond === '' ) {
 155                                         $nextCond = "$field > $encValue";
 156                                 } else {
 157                                         $nextCond = "$field > $encValue OR ($field = $encValue AND ($nextCond))";
 158                                 }
 159                         }
 160                         $indexConds = array( $nextCond );
 161                 }
 162
 163                 $this->output( "Finished $table... $this->updated of $this->processed rows updated\n" );
 164         }
 165
 166         /**
 167          * @param array $matches
 168          * @return string
 169          */
 170         protected function hexChar( $matches ) {
 171                 return sprintf( "\\x%02x", ord( $matches[1] ) );
 172         }
 173 }
 174
 175 class TableCleanupTest extends TableCleanup {
 176         function processRow( $row ) {
 177                 $this->progress( mt_rand( 0, 1 ) );
 178         }
 179 }
 180