+ return $count;
+ }
+
+ /**
+ * @return int
+ */
+ protected function doSha1LegacyUpdates() {
+ $count = 0;
+ $db = $this->getDB( DB_MASTER );
+ $res = $db->select( 'archive', '*',
+ array( 'ar_rev_id IS NULL', 'ar_sha1' => '' ), __METHOD__ );
+
+ $updateSize = 0;
+ $db->begin( __METHOD__ );
+ foreach ( $res as $row ) {
+ if ( $this->upgradeLegacyArchiveRow( $row ) ) {
+ ++$count;
+ }
+ if ( ++$updateSize >= 100 ) {
+ $updateSize = 0;
+ $db->commit( __METHOD__ );
+ $this->output( "Commited row with ar_timestamp={$row->ar_timestamp}\n" );
+ wfWaitForSlaves();
+ $db->begin( __METHOD__ );
+ }
+ }
+ $db->commit( __METHOD__ );
+ return $count;
+ }
+
+ /**
+ * @param $row
+ * @param $table
+ * @param $idCol
+ * @param $prefix
+ * @return bool
+ */
+ protected function upgradeRow( $row, $table, $idCol, $prefix ) {
+ $db = $this->getDB( DB_MASTER );
+ try {
+ $rev = ( $table === 'archive' )
+ ? Revision::newFromArchiveRow( $row )
+ : new Revision( $row );
+ $text = $rev->getRawText();
+ } catch ( MWException $e ) {
+ $this->output( "Text of revision with {$idCol}={$row->$idCol} unavailable!\n" );
+ return false; // bug 22624?
+ }
+ 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;
+ }
+ }
+
+ /**
+ * @param $row
+ * @return bool
+ */
+ protected function upgradeLegacyArchiveRow( $row ) {
+ $db = $this->getDB( DB_MASTER );
+ try {
+ $rev = Revision::newFromArchiveRow( $row );
+ } catch ( MWException $e ) {
+ $this->output( "Text of revision with timestamp {$row->ar_timestamp} unavailable!\n" );
+ return false; // bug 22624?
+ }
+ $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;
+ }