Fixes for TitleCleanup subclasses:

[lhc/web/wiklou.git] / maintenance / cleanupTable.inc

diff --git a/maintenance/cleanupTable.inc b/maintenance/cleanupTable.inc
index f67b3b1..da0ff88 100644 (file)
--- a/maintenance/cleanupTable.inc
+++ b/maintenance/cleanupTable.inc
@@ -23,10 +23,18 @@
 
 require_once( dirname(__FILE__) . '/Maintenance.php' );
 
-abstract class TableCleanup extends Maintenance {
-       protected $targetTable = 'page';
+class TableCleanup extends Maintenance {
+       protected $defaultParams = array(
+               'table' => 'page',
+               'conds' => array(),
+               'index' => 'page_id',
+               'callback' => 'processRow',
+       );
+
        protected $dryrun = false;
        protected $maxLag = 10; # if slaves are lagged more than 10 secs, wait
+       public $batchSize = 100;
+       public $reportInterval = 100;
 
        public function __construct() {
                parent::__construct();
@@ -42,9 +50,7 @@ abstract class TableCleanup extends Maintenance {
                } else {
                        $this->output( "Checking and fixing bad titles...\n" );
                }
-               $this->runTable( $this->targetTable,
-                       '', //'WHERE page_namespace=0',
-                       array( $this, 'processPage' ) );
+               $this->runTable( $this->defaultParams );
        }
 
        protected function init( $count, $table ) {
@@ -58,7 +64,7 @@ abstract class TableCleanup extends Maintenance {
        protected function progress( $updated ) {
                $this->updated += $updated;
                $this->processed++;
-               if( $this->processed % 100 != 0 ) {
+               if( $this->processed % $this->reportInterval != 0 ) {
                        return;
                }
                $portion = $this->processed / $this->count;
@@ -85,34 +91,73 @@ abstract class TableCleanup extends Maintenance {
                flush();
        }
 
-       protected function runTable( $table, $where, $callback ) {
+       public function runTable( $params ) {
                $dbr = wfGetDB( DB_SLAVE );
-               
-               $count = $dbr->selectField( $table, 'count(*)', '', __METHOD__ );
+
+               if ( array_diff( array_keys( $params ),
+                       array( 'table', 'conds', 'index', 'callback' ) ) ) 
+               {
+                       throw new MWException( __METHOD__.': Missing parameter ' . implode( ', ', $params ) );
+               }
+
+               $table = $params['table'];
+               $count = $dbr->selectField( $table, 'count(*)', $params['conds'], __METHOD__ );
                $this->init( $count, $table );
-               $this->output( "Processing $table..." );
+               $this->output( "Processing $table...\n" );
+
+
+               $index = (array)$params['index'];
+               $indexConds = array();
+               $options = array(
+                       'ORDER BY' => implode( ',', $index ),
+                       'LIMIT' => $this->batchSize
+               );
+               $callback = array( $this, $params['callback'] );
 
-               // Unbuffered queries, avoids OOM
-               $dbr->bufferResults( false );
-               
-               $tableName = $dbr->tableName( $table );
-               $sql = "SELECT * FROM $tableName $where";
-               $result = $dbr->query( $sql, __METHOD__ );
+               while ( true ) {
+                       $conds = array_merge( $params['conds'], $indexConds );
+                       $res = $dbr->select( $table, '*', $conds, __METHOD__, $options );
+                       if ( !$res->numRows() ) {
+                               // Done
+                               break;
+                       }
 
-               foreach( $result as $row ) {
-                       call_user_func( $callback, $row );
+                       foreach ( $res as $row ) {
+                               call_user_func( $callback, $row );
+                       }
+
+                       if ( $res->numRows() < $this->batchSize ) {
+                               // Done
+                               break;
+                       }
+
+                       // Update the conditions to select the next batch.
+                       // Construct a condition string by starting with the least significant part
+                       // of the index, and adding more significant parts progressively to the left
+                       // of the string.
+                       $nextCond = '';
+                       foreach ( array_reverse( $index ) as $field ) {
+                               $encValue = $dbr->addQuotes( $row->$field );
+                               if ( $nextCond === '' ) {
+                                       $nextCond = "$field > $encValue";
+                               } else {
+                                       $nextCond = "$field > $encValue OR ($field = $encValue AND ($nextCond))";
+                               }
+                       }
+                       $indexConds = array( $nextCond );
                }
-               
+
                $this->output( "Finished $table... $this->updated of $this->processed rows updated\n" );
-               
-               $result->free();
-               
-               $dbr->bufferResults( true );
        }
 
        protected function hexChar( $matches ) {
                return sprintf( "\\x%02x", ord( $matches[1] ) );
        }
-       
-       abstract protected function processPage( $row );
 }
+
+class TableCleanupTest extends TableCleanup {
+       function processRow( $row ) {
+               $this->progress( mt_rand( 0, 1 ) );
+       }
+}
+