Check if non-JSON-serializable data passed to ParserOutput

Bug: T264394
Change-Id: I6eedd03a81b95f6f55d25c00b31e01cbd8658d43

includes/json/FormatJson.php

@@ -319,4 +319,49 @@ class FormatJson {
 		// Add final chunk to buffer before returning
 		return $buffer . substr( $str, $mark, $maxLen - $mark );
 	}
+
+	/**
+	 * Recursive check for ability to serialize $value to JSON via FormatJson::encode().
+	 *
+	 * @note instances of JsonSerializable interface are not considered serializable
+	 * in this method. The $value passed here is a result of JsonSerializable::jsonSerialize().
+	 *
+	 *
+	 * @param mixed $value
+	 * @param string $accumulatedPath
+	 * @return string|null JSON path to first encountered non-serializable property or null.
+	 */
+	private static function detectNonSerializableDataInternal(
+		$value,
+		string $accumulatedPath
+	): ?string {
+		if ( is_array( $value ) ||
+			( is_object( $value ) && get_class( $value ) === 'stdClass' ) ) {
+			foreach ( $value as $key => $propValue ) {
+				$propValueNonSerializablePath = self::detectNonSerializableDataInternal(
+					$propValue,
+					$accumulatedPath . '.' . $key
+				);
+				if ( $propValueNonSerializablePath ) {
+					return $propValueNonSerializablePath;
+				}
+			}
+		// Instances of classes other the \stdClass can not be serialized to JSON
+		} elseif ( !is_scalar( $value ) && $value !== null ) {
+			return $accumulatedPath;
+		}
+		return null;
+	}
+
+	/**
+	 * Checks if the $value is JSON-serializable (contains only scalar values)
+	 * and returns a JSON-path to the first non-serializable property encountered.
+	 *
+	 * @since 1.36
+	 * @param mixed $value
+	 * @return string|null JSON path to first encountered non-serializable property or null.
+	 */
+	public static function detectNonSerializableData( $value ): ?string {
+		return self::detectNonSerializableDataInternal( $value, '$' );
+	}
 }

includes/parser/ParserOutput.php

@@ -1135,6 +1135,16 @@ class ParserOutput extends CacheTime {
 	 * @param mixed $value
 	 */
 	public function setProperty( $name, $value ) {
+		$unserializablePath = FormatJson::detectNonSerializableData( $value );
+		if ( $unserializablePath ) {
+			LoggerFactory::getInstance( 'ParserOutput' )->warning(
+				'Non-serializable page property set',
+				[
+					'name' => $name,
+					'path' => $unserializablePath,
+				]
+			);
+		}
 		$this->mProperties[$name] = $value;
 	}
 
@@ -1233,6 +1243,16 @@ class ParserOutput extends CacheTime {
 		if ( $value === null ) {
 			unset( $this->mExtensionData[$key] );
 		} else {
+			$unserializablePath = FormatJson::detectNonSerializableData( $value );
+			if ( $unserializablePath ) {
+				LoggerFactory::getInstance( 'ParserOutput' )->warning(
+					'Non-serializable extension data set',
+					[
+						'key' => $key,
+						'path' => $unserializablePath,
+					]
+				);
+			}
 			$this->mExtensionData[$key] = $value;
 		}
 	}

tests/phpunit/unit/includes/json/FormatJsonTest.php

@@ -79,4 +79,35 @@ class FormatJsonTest extends MediaWikiUnitTestCase {
 		}
 	}
 
+	public function provideValidateSerializable() {
+		$classInstance = new class() {
+		};
+
+		yield 'Number' => [ 1, null ];
+		yield 'Null' => [ null, null ];
+		yield 'Class' => [ $classInstance, '$' ];
+		yield 'Empty array' => [ [], null ];
+		yield 'Empty stdClass' => [ new stdClass(), null ];
+		yield 'Non-empty array' => [ [ 1, 2, 3 ], null ];
+		yield 'Non-empty map' => [ [ 'a' => 'b' ], null ];
+		yield 'Nested, serializable' => [ [ 'a' => [ 'b' => [ 'c' => 'd' ] ] ], null ];
+		yield 'Nested, serializable, with null' => [ [ 'a' => [ 'b' => null ] ], null ];
+		yield 'Nested, serializable, with stdClass' => [ [ 'a' => (object)[ 'b' => [ 'c' => 'd' ] ] ], null ];
+		yield 'Nested, serializable, with stdClass, with null' => [ [ 'a' => (object)[ 'b' => null ] ], null ];
+		yield 'Nested, non-serializable' => [ [ 'a' => [ 'b' => $classInstance ] ], '$.a.b' ];
+		yield 'Nested, non-serializable, in array' => [ [ 'a' => [ 1, 2, $classInstance ] ], '$.a.2' ];
+		yield 'Nested, non-serializable, in stdClass' => [ [ 'a' => (object)[ 1, 2, $classInstance ] ], '$.a.2' ];
+	}
+
+	/**
+	 * @dataProvider provideValidateSerializable
+	 * @covers FormatJson::detectNonSerializableData
+	 * @covers FormatJson::detectNonSerializableDataInternal
+	 * @param $value
+	 * @param string|null $result
+	 */
+	public function testValidateSerializable( $value, ?string $result ) {
+		$this->assertSame( $result, FormatJson::detectNonSerializableData( $value ) );
+	}
+
 }