MergeHistory: Update revactor_page too

When using MergeHistory, we need to update the denormalized
revision_actor_temp.revactor_page to match the update we do for
revision.rev_page.

Also, we need a maintenance script to clean up the rows that were broken
by our failure to do that before.

Bug: T232464
Change-Id: Ib819a9d9fc978d75d7cc7e53f361483b69ab8020

3 files changed, 90 insertions, 0 deletions

diff --git a/autoload.php b/autoload.php
index f95e0015a733..48d5b30d6db9 100644
--- a/autoload.php
+++ b/autoload.php
@@ -272,6 +272,7 @@ $wgAutoloadLocalClasses = [
 	'CleanupInvalidDbKeys' => __DIR__ . '/maintenance/cleanupInvalidDbKeys.php',
 	'CleanupPreferences' => __DIR__ . '/maintenance/cleanupPreferences.php',
 	'CleanupRemovedModules' => __DIR__ . '/maintenance/cleanupRemovedModules.php',
+	'CleanupRevActorPage' => __DIR__ . '/maintenance/cleanupRevActorPage.php',
 	'CleanupSpam' => __DIR__ . '/maintenance/cleanupSpam.php',
 	'CleanupUploadStash' => __DIR__ . '/maintenance/cleanupUploadStash.php',
 	'CleanupUsersWithNoId' => __DIR__ . '/maintenance/cleanupUsersWithNoId.php',
diff --git a/includes/MergeHistory.php b/includes/MergeHistory.php
index 4045a5436b56..1a950c5f5e1f 100644
--- a/includes/MergeHistory.php
+++ b/includes/MergeHistory.php
@@ -273,6 +273,18 @@ class MergeHistory {
 			return $status;
 		}
 
+		// Update denormalized revactor_page too
+		$this->dbw->update(
+			'revision_actor_temp',
+			[ 'revactor_page' => $this->dest->getArticleID() ],
+			[
+				'revactor_page' => $this->source->getArticleID(),
+				// Slightly hacky, but should work given the values assigned in this class
+				str_replace( 'rev_timestamp', 'revactor_timestamp', $this->timeWhere )
+			],
+			__METHOD__
+		);
+
 		// Make the source page a redirect if no revisions are left
 		$haveRevisions = $this->dbw->lockForUpdate(
 			'revision',
diff --git a/maintenance/cleanupRevActorPage.php b/maintenance/cleanupRevActorPage.php
new file mode 100644
index 000000000000..ac655fc93bb2
--- /dev/null
+++ b/maintenance/cleanupRevActorPage.php
@@ -0,0 +1,77 @@
+<?php
+
+require_once __DIR__ . '/Maintenance.php';
+
+/**
+ * Maintenance script that cleans up cases where rev_page and revactor_page
+ * became desynced, e.g. from T232464.
+ *
+ * @ingroup Maintenance
+ * @since 1.34
+ */
+class CleanupRevActorPage extends LoggedUpdateMaintenance {
+
+	public function __construct() {
+		parent::__construct();
+		$this->addDescription(
+			'Resyncs revactor_page with rev_page when they differ, e.g. from T232464.'
+		);
+		$this->setBatchSize( 1000 );
+	}
+
+	protected function getUpdateKey() {
+		return __CLASS__;
+	}
+
+	protected function doDBUpdates() {
+		$dbw = $this->getDB( DB_MASTER );
+		$max = $dbw->selectField( 'revision', 'MAX(rev_id)', '', __METHOD__ );
+		$batchSize = $this->mBatchSize;
+
+		$this->output( "Resyncing revactor_page with rev_page...\n" );
+
+		$count = 0;
+		for ( $start = 1; $start <= $max; $start += $batchSize ) {
+			$end = $start + $batchSize - 1;
+			$this->output( "... rev_id $start - $end, $count changed\n" );
+
+			// Fetch the rows needing update
+			$res = $dbw->select(
+				[ 'revision', 'revision_actor_temp' ],
+				[ 'rev_id', 'rev_page' ],
+				[
+					'rev_page != revactor_page',
+					"rev_id >= $start",
+					"rev_id <= $end",
+				],
+				__METHOD__,
+				[],
+				[ 'revision_actor_temp' => [ 'JOIN', 'rev_id = revactor_rev' ] ]
+			);
+
+			if ( !$res->numRows() ) {
+				continue;
+			}
+
+			// Update the existing rows
+			foreach ( $res as $row ) {
+				$dbw->update(
+					'revision_actor_temp',
+					[ 'revactor_page' => $row->rev_page ],
+					[ 'revactor_rev' => $row->rev_id ],
+					__METHOD__
+				);
+				$count += $dbw->affectedRows();
+			}
+
+			wfWaitForSlaves();
+		}
+
+		$this->output( "Completed resync, $count row(s) updated\n" );
+
+		return true;
+	}
+}
+
+$maintClass = CleanupRevActorPage::class;
+require_once RUN_MAINTENANCE_IF_MAIN;