populateRevisionSha1.php

Go to the documentation of this file.

<?php
require_once( dirname( __FILE__ ) . '/Maintenance.php' );

class PopulateRevisionSha1 extends LoggedUpdateMaintenance {
        public function __construct() {
                parent::__construct();
                $this->mDescription = "Populates the rev_sha1 and ar_sha1 fields";
                $this->setBatchSize( 200 );
        }

        protected function getUpdateKey() {
                return 'populate rev_sha1';
        }

        protected function doDBUpdates() {
                $db = $this->getDB( DB_MASTER );

                if ( !$db->tableExists( 'revision' ) ) {
                        $this->error( "revision table does not exist", true );
                } elseif ( !$db->tableExists( 'archive' ) ) {
                        $this->error( "archive table does not exist", true );
                }

                $this->output( "Populating rev_sha1 column\n" );
                $rc = $this->doSha1Updates( 'revision', 'rev_id', 'rev' );

                $this->output( "Populating ar_sha1 column\n" );
                $ac = $this->doSha1Updates( 'archive', 'ar_rev_id', 'ar' );
                $this->output( "Populating ar_sha1 column legacy rows\n" );
                $ac += $this->doSha1LegacyUpdates();

                $this->output( "rev_sha1 and ar_sha1 population complete [$rc revision rows, $ac archive rows].\n" );
                return true;
        }

        protected function doSha1Updates( $table, $idCol, $prefix ) {
                $db = $this->getDB( DB_MASTER );
                $start = $db->selectField( $table, "MIN($idCol)", false, __METHOD__ );
                $end = $db->selectField( $table, "MAX($idCol)", false, __METHOD__ );
                if ( !$start || !$end ) {
                        $this->output( "...$table table seems to be empty.\n" );
                        return 0;
                }

                $count = 0;
                # Do remaining chunk
                $end += $this->mBatchSize - 1;
                $blockStart = $start;
                $blockEnd = $start + $this->mBatchSize - 1;
                while ( $blockEnd <= $end ) {
                        $this->output( "...doing $idCol from $blockStart to $blockEnd\n" );
                        $cond = "$idCol BETWEEN $blockStart AND $blockEnd
                                AND $idCol IS NOT NULL AND {$prefix}_sha1 = ''";
                        $res = $db->select( $table, '*', $cond, __METHOD__ );

                        $db->begin();
                        foreach ( $res as $row ) {
                                if ( $this->upgradeRow( $row, $table, $idCol, $prefix ) ) {
                                        $count++;
                                }
                        }
                        $db->commit();

                        $blockStart += $this->mBatchSize;
                        $blockEnd += $this->mBatchSize;
                        wfWaitForSlaves();
                }
                return $count;
        }

        protected function doSha1LegacyUpdates() {
                $count = 0;
                $db = $this->getDB( DB_MASTER );
                $res = $db->select( 'archive', '*', array( 'ar_rev_id IS NULL' ), __METHOD__ );

                $updateSize = 0;
                $db->begin();
                foreach ( $res as $row ) {
                        if ( $this->upgradeLegacyArchiveRow( $row ) ) {
                                ++$count;
                        }
                        if ( ++$updateSize >= 100 ) {
                                $updateSize = 0;
                                $db->commit();
                                $this->output( "Commited row with ar_timestamp={$row->ar_timestamp}\n" );
                                wfWaitForSlaves();
                                $db->begin();
                        }
                }
                $db->commit();
                return $count;
        }

        protected function upgradeRow( $row, $table, $idCol, $prefix ) {
                $db = $this->getDB( DB_MASTER );
                if ( $table === 'archive' ) {
                        $rev = Revision::newFromArchiveRow( $row );
                } else {
                        $rev = new Revision( $row );
                }
                $text = $rev->getRawText();
                if ( !is_string( $text ) ) {
                        # This should not happen, but sometimes does (bug 20757)
                        $this->output( "Text of revision with {$idCol}={$row->$idCol} unavailable!\n" );
                        return false;
                } else {
                        $db->update( $table,
                                array( "{$prefix}_sha1" => Revision::base36Sha1( $text ) ),
                                array( $idCol => $row->$idCol ),
                                __METHOD__ 
                        );
                        return true;
                }
        }

        protected function upgradeLegacyArchiveRow( $row ) {
                $db = $this->getDB( DB_MASTER );
                $rev = Revision::newFromArchiveRow( $row );
                $text = $rev->getRawText();
                if ( !is_string( $text ) ) {
                        # This should not happen, but sometimes does (bug 20757)
                        $this->output( "Text of revision with timestamp {$row->ar_timestamp} unavailable!\n" );
                        return false;
                } else {
                        # Archive table as no PK, but (NS,title,time) should be near unique.
                        # Any duplicates on those should also have duplicated text anyway.
                        $db->update( 'archive',
                                array( 'ar_sha1' => Revision::base36Sha1( $text ) ),
                                array(
                                        'ar_namespace' => $row->ar_namespace,
                                        'ar_title'     => $row->ar_title,
                                        'ar_timestamp' => $row->ar_timestamp,
                                        'ar_len'       => $row->ar_len // extra sanity
                                ),
                                __METHOD__ 
                        );
                        return true;
                }
        }
}

$maintClass = "PopulateRevisionSha1";
require_once( RUN_MAINTENANCE_IF_MAIN );