populateRevisionLength.php

Go to the documentation of this file.

<?php
require_once( __DIR__ . '/Maintenance.php' );

class PopulateRevisionLength extends LoggedUpdateMaintenance {
        public function __construct() {
                parent::__construct();
                $this->mDescription = "Populates the rev_len field";
                $this->setBatchSize( 200 );
        }

        protected function getUpdateKey() {
                return 'populate rev_len';
        }

        protected function updateSkippedMessage() {
                return 'rev_len column of revision table already populated.';
        }

        public function doDBUpdates() {
                $db = $this->getDB( DB_MASTER );
                if ( !$db->tableExists( 'revision' ) ) {
                        $this->error( "revision table does not exist", true );
                } else if ( !$db->fieldExists( 'revision', 'rev_sha1', __METHOD__ ) ) {
                        $this->output( "rev_sha1 column does not exist\n\n", true );
                        return false;
                }

                $this->output( "Populating rev_len column\n" );

                $start = $db->selectField( 'revision', 'MIN(rev_id)', false, __METHOD__ );
                $end = $db->selectField( 'revision', 'MAX(rev_id)', false, __METHOD__ );
                if ( !$start || !$end ) {
                        $this->output( "...revision table seems to be empty.\n" );
                        return true;
                }

                # Do remaining chunks
                $blockStart = intval( $start );
                $blockEnd = intval( $start ) + $this->mBatchSize - 1;
                $count = 0;
                $missing = 0;
                $fields = Revision::selectFields();
                while ( $blockStart <= $end ) {
                        $this->output( "...doing rev_id from $blockStart to $blockEnd\n" );
                        $res = $db->select( 'revision',
                                                $fields,
                                                array( "rev_id >= $blockStart",
                                                   "rev_id <= $blockEnd",
                                                   "rev_len IS NULL" ),
                                                __METHOD__ );
                        # Go through and update rev_len from these rows.
                        foreach ( $res as $row ) {
                                $rev = new Revision( $row );
                                $content = $rev->getContent();
                                if ( !$content ) {
                                        # This should not happen, but sometimes does (bug 20757)
                                        $this->output( "Content of revision {$row->rev_id} unavailable!\n" );
                                        $missing++;
                                }
                                else {
                                        # Update the row...
                                        $db->update( 'revision',
                                                         array( 'rev_len' => $content->getSize() ),
                                                         array( 'rev_id' => $row->rev_id ),
                                                         __METHOD__ );
                                        $count++;
                                }
                        }
                        $blockStart += $this->mBatchSize;
                        $blockEnd += $this->mBatchSize;
                        wfWaitForSlaves();
                }

                $this->output( "rev_len population complete ... {$count} rows changed ({$missing} missing)\n" );
                return true;
        }
}

$maintClass = "PopulateRevisionLength";
require_once( RUN_MAINTENANCE_IF_MAIN );