Use revision table instead of recentchanges

This switches Nuke over to using the `revision` table instead
of `recentchanges`, allowing a wider timespan of pages to be
deleted. As a DoS prevention measure, this adds the
`$wgNukeMaxAge` config option (defaulted to the value of
`$wgRCMaxAge`) and a max execution time for the SELECT
(set to `$wgMaxExecutionTimeForExpensiveQueries`).

This also adds a relevant test.

Partially based off of I6c2b7e6b695d58a7dcba93ccaeba9ed35d81cf80.

Bug: T379147
Co-Authored-by: Kgraessle <kgraessle@wikimedia.org>
Change-Id: I5d68d2663751783bcc773799e951f74866ceb722

extension.json

@@ -29,6 +29,12 @@
 		"ListDefinedTags": "MediaWiki\\Extension\\Nuke\\Hooks::onRegisterTags",
 		"ChangeTagsListActive": "MediaWiki\\Extension\\Nuke\\Hooks::onRegisterTags"
 	},
+	"config": {
+		"NukeMaxAge": {
+			"value": 0,
+			"description": "The maximum age of a new page creation or file upload before it becomes ineligible for mass deletion. Defaults to the value of $wgRCMaxAge."
+		}
+	},
 	"ResourceModules": {
 		"ext.nuke.confirm": {
 			"scripts": [

includes/SpecialNuke.php

@@ -12,6 +12,7 @@ use MediaWiki\Html\Html;
 use MediaWiki\Html\ListToggle;
 use MediaWiki\HTMLForm\HTMLForm;
 use MediaWiki\Language\Language;
+use MediaWiki\MainConfigNames;
 use MediaWiki\Page\File\FileDeleteForm;
 use MediaWiki\Permissions\PermissionManager;
 use MediaWiki\Request\WebRequest;
@@ -437,21 +438,32 @@ class SpecialNuke extends SpecialPage {
 	 */
 	protected function getNewPages( $username, $limit, $namespace = null, $tempnames = [] ): array {
 		$dbr = $this->dbProvider->getReplicaDatabase();
+
+		$maxAge = $this->getConfig()->get( "NukeMaxAge" );
+		// If no Nuke-specific max age was set, this should match the value of `$wgRCMaxAge`.
+		if ( !$maxAge ) {
+			$maxAge = $this->getConfig()->get( MainConfigNames::RCMaxAge );
+		}
+
 		$queryBuilder = $dbr->newSelectQueryBuilder()
 			->select( [ 'page_title', 'page_namespace' ] )
-			->from( 'recentchanges' )
-			->join( 'actor', null, 'actor_id=rc_actor' )
-			->join( 'page', null, 'page_id=rc_cur_id' )
-			->where(
-				$dbr->expr( 'rc_source', '=', 'mw.new' )->orExpr(
-					$dbr->expr( 'rc_log_type', '=', 'upload' )
-						->and( 'rc_log_action', '=', 'upload' )
-				)
-			)
-			->orderBy( 'rc_timestamp', SelectQueryBuilder::SORT_DESC )
-			->limit( $limit );
+			->from( 'revision' )
+			->join( 'actor', null, 'actor_id=rev_actor' )
+			->join( 'page', null, 'page_id=rev_page' )
+			->where( [
+				$dbr->expr( 'rev_parent_id', '=', 0 ),
+				$dbr->expr( 'rev_timestamp', '>', $dbr->timestamp(
+					time() - $maxAge
+				) )
+			] )
+			->orderBy( 'rev_timestamp', SelectQueryBuilder::SORT_DESC )
+			->distinct()
+			->limit( $limit )
+			->setMaxExecutionTime(
+				$this->getConfig()->get( MainConfigNames::MaxExecutionTimeForExpensiveQueries )
+			);
 
-		$queryBuilder->field( 'actor_name', 'rc_user_text' );
+		$queryBuilder->field( 'actor_name' );
 		$actornames = array_filter( [ $username, ...$tempnames ] );
 		if ( $actornames ) {
 			$queryBuilder->andWhere( [ 'actor_name' => $actornames ] );
@@ -556,7 +568,7 @@ class SpecialNuke extends SpecialPage {
 		foreach ( $result as $row ) {
 			$pages[] = [
 				Title::makeTitle( $row->page_namespace, $row->page_title ),
-				$row->rc_user_text
+				$row->actor_name
 			];
 		}
 

tests/phpunit/integration/SpecialNukeTest.php

@@ -560,6 +560,77 @@ class SpecialNukeTest extends SpecialPageTestBase {
 		$this->assertEquals( 2, substr_count( $html, '<li>' ) );
 	}
 
+	public function testListMaxAge() {
+		// 1 day
+		$maxAge = 86400;
+		$this->overrideConfigValues( [
+			'NukeMaxAge' => $maxAge,
+			'RCMaxAge' => $maxAge
+		] );
+
+		// Will never show up. If it does, the max age isn't being applied at all.
+		$page1Status = $this->editPage( 'Page1', 'test' );
+		// Will show up conditionally (see below).
+		$page2Status = $this->editPage( 'Page2', 'test' );
+		// Will always show up.
+		$this->editPage( 'Page3', 'test' );
+
+		// Prepare database connection and update query builder
+		$dbw = $this->getServiceContainer()->getConnectionProvider()->getPrimaryDatabase();
+		$uqb = $dbw->newUpdateQueryBuilder()
+			->update( 'revision' )
+			->caller( __METHOD__ );
+
+		// Manually change the rev_timestamp of Page1's creation in the database.
+		( clone $uqb )
+			->set( [ 'rev_timestamp' => $dbw->timestamp( time() - ( $maxAge * 3 ) ) ] )
+			->where( [
+				'rev_id' => $page1Status->getNewRevision()->getId()
+			] )->execute();
+		// Manually change the rev_timestamp of Page2's creation in the database.
+		( clone $uqb )
+			->set( [ 'rev_timestamp' => $dbw->timestamp( time() - $maxAge - 60 ) ] )
+			->where( [
+				'rev_id' => $page2Status->getNewRevision()->getId()
+			] )->execute();
+
+		$admin = $this->getTestSysop()->getUser();
+		$request = new FauxRequest( [
+			'action' => 'submit',
+			'pattern' => 'Page%',
+			'limit' => 2
+		], true );
+		$performer = new UltimateAuthority( $admin );
+
+		[ $html ] = $this->executeSpecialPage( '', $request, 'qqx', $performer );
+		$this->assertStringNotContainsString( 'Page1', $html );
+		$this->assertStringNotContainsString( 'Page2', $html );
+		$this->assertStringContainsString( 'Page3', $html );
+
+		// Now check if resetting $wgNukeMaxAge will show Page2.
+		// This should follow $wgRCMaxAge, which will include the page.
+		$this->overrideConfigValues( [
+			'RCMaxAge' => $maxAge * 2,
+			'NukeMaxAge' => 0
+		] );
+
+		[ $html ] = $this->executeSpecialPage( '', $request, 'qqx', $performer );
+		$this->assertStringNotContainsString( 'Page1', $html );
+		$this->assertStringContainsString( 'Page2', $html );
+		$this->assertStringContainsString( 'Page3', $html );
+
+		// Now check if expanding just $wgNukeMaxAge will show Page2.
+		$this->overrideConfigValues( [
+			'RCMaxAge' => $maxAge,
+			'NukeMaxAge' => $maxAge * 2
+		] );
+
+		[ $html ] = $this->executeSpecialPage( '', $request, 'qqx', $performer );
+		$this->assertStringNotContainsString( 'Page1', $html );
+		$this->assertStringContainsString( 'Page2', $html );
+		$this->assertStringContainsString( 'Page3', $html );
+	}
+
 	public function testExecutePattern() {
 		// Test that matching wildcards works, and that escaping wildcards works as documented
 		// at https://www.mediawiki.org/wiki/Help:Extension:Nuke