Fix XML dumps for content types with non-string getNativeData()

In fdc3e9f9524d91a492bdc212486d4518991c0fe2, the code generating XML
dumps was updated to support multi-content revisions. This refactor
included a workaround for content types that are subclasses of
TextContent to use getNativeData() rather than serialize(), apparently
to satisfy the Flow extension.

However, this assumes that getNativeData() always returns a string. As
demonstrated in T155582, this is not the case, which is one of the
reasons why the method was deprecated. Notably, if a wiki has a custom
content type defined whose getNativeData() returns a non-string value,
and has pages using that content type, this breaks XML dump generation
(dumpBackup.php) for that wiki and also makes those pages unexportable
via Special:Export.

Fix it by using getText() instead of getNativeData(), which is the
recommended migration path anyways per T155582. I am somewhat perplexed
by the reference to Flow in the original code comment, because Flow's
BoardContent does not seem to extend TextContent at all.

Bug: T155582
Bug: T328503
Change-Id: I670fb53f193ec20d3d4c258e54c89e7f64cf2d1b

1 files changed, 116 insertions, 17 deletions

diff --git a/tests/phpunit/includes/ExportTest.php b/tests/phpunit/includes/ExportTest.php
index fd13c5baaba2..526306635211 100644
--- a/tests/phpunit/includes/ExportTest.php
+++ b/tests/phpunit/includes/ExportTest.php
@@ -1,6 +1,8 @@
 <?php
 
+use MediaWiki\Content\Renderer\ContentParseParams;
 use MediaWiki\MainConfigNames;
+use MediaWiki\Page\PageIdentity;
 
 /**
  * Test class for Export methods.
@@ -23,26 +25,10 @@ class ExportTest extends MediaWikiLangTestCase {
 		$pageTitle = 'UTPage';
 
 		$services = $this->getServiceContainer();
-		$exporter = $services
-			->getWikiExporterFactory()
-			->getWikiExporter( $this->db, WikiExporter::FULL );
 
 		$title = Title::newFromText( $pageTitle );
 
-		$sink = new DumpStringOutput;
-		$exporter->setOutputSink( $sink );
-		$exporter->openStream();
-		$exporter->pageByTitle( $title );
-		$exporter->closeStream();
-
-		// phpcs:ignore Generic.PHP.NoSilencedErrors -- suppress deprecation per T268847
-		$oldDisable = @libxml_disable_entity_loader( true );
-
-		// This throws error if invalid xml output
-		$xmlObject = simplexml_load_string( $sink );
-
-		// phpcs:ignore Generic.PHP.NoSilencedErrors
-		@libxml_disable_entity_loader( $oldDisable );
+		$xmlObject = $this->getXmlDumpForPage( $title );
 
 		/**
 		 * Check namespaces match xml
@@ -66,4 +52,117 @@ class ExportTest extends MediaWikiLangTestCase {
 		$this->assertNotEquals( '', $text[0] );
 	}
 
+	/**
+	 * Regression test for T328503 to verify that custom content types
+	 * with a getNativeData() override that returns a non-string value export correctly.
+	 *
+	 * @covers XmlDumpWriter::writeText
+	 */
+	public function testShouldExportContentWithNonStringNativeData(): void {
+		// Make a mock ContentHandler for a Content that has a getNativeData() method
+		// with a non-string return value.
+		$contentModelId = 'non-string-test-content-model';
+		$contentHandler = new class( $contentModelId ) extends ContentHandler {
+
+			public function __construct( $contentModelId ) {
+				parent::__construct(
+					$contentModelId,
+					[ CONTENT_FORMAT_TEXT ]
+				);
+			}
+
+			public function serializeContent( Content $content, $format = null ) {
+				return json_encode( $content->getNativeData() );
+			}
+
+			public function unserializeContent( $blob, $format = null ) {
+				return $this->getTestContent( $blob );
+			}
+
+			public function makeEmptyContent() {
+				return $this->getTestContent( '{}' );
+			}
+
+			protected function fillParserOutput(
+				Content $content,
+				ContentParseParams $cpoParams,
+				ParserOutput &$output
+			) {
+				$output->setText( json_encode( $content->getNativeData() ) );
+			}
+
+			private function getTestContent( string $blob ): Content {
+				return new class( $blob, $this->getModelID() ) extends TextContent {
+					/** @var array */
+					private $data;
+
+					public function __construct( $text, $contentModelId ) {
+						parent::__construct(
+							$text,
+							$contentModelId
+						);
+
+						$this->data = json_decode( $text, true );
+					}
+
+					public function getNativeData() {
+						return $this->data;
+					}
+				};
+			}
+		};
+
+		$this->setTemporaryHook(
+			'ContentHandlerForModelID',
+			static function (
+				string $modelId,
+				?ContentHandler &$handlerRef
+			) use ( $contentModelId, $contentHandler ): void {
+				if ( $modelId === $contentModelId ) {
+					$handlerRef = $contentHandler;
+				}
+			}
+		);
+
+		$wikiPage = $this->getNonexistingTestPage( 'NonStringNativeDataExportTest' );
+
+		$testText = json_encode( [ 'test' => 'data' ] );
+		$content = $contentHandler->unserializeContent( $testText );
+
+		$this->editPage( $wikiPage, $content );
+
+		$xmlObject = $this->getXmlDumpForPage( $wikiPage );
+
+		$this->assertSame( $contentModelId, (string)$xmlObject->page->revision->model );
+		$this->assertSame( $testText, (string)$xmlObject->page->revision->text );
+	}
+
+	/**
+	 * Convenience function to export the content of the given page in MediaWiki's XML dump format.
+	 * @param PageIdentity $page page to export
+	 * @return SimpleXMLElement root element of the generated XML
+	 */
+	private function getXmlDumpForPage( PageIdentity $page ): SimpleXMLElement {
+		$exporter = $this->getServiceContainer()
+			->getWikiExporterFactory()
+			->getWikiExporter( $this->db, WikiExporter::FULL );
+
+		$sink = new DumpStringOutput();
+		$exporter->setOutputSink( $sink );
+		$exporter->openStream();
+		$exporter->pageByTitle( $page );
+		$exporter->closeStream();
+
+		// phpcs:ignore Generic.PHP.NoSilencedErrors -- suppress deprecation per T268847
+		$oldDisable = @libxml_disable_entity_loader( true );
+
+		// This throws error if invalid xml output
+		$xmlObject = simplexml_load_string( $sink );
+
+		// phpcs:ignore Generic.PHP.NoSilencedErrors
+		@libxml_disable_entity_loader( $oldDisable );
+
+		return $xmlObject;
+	}
+
 }