fix: #3549107 Escape or strip control characters in JSON:API (8d5894c6) · Commits · project / drupal

core/misc/cspell/dictionary.txt

+1 −0

Original line number	Diff line number	Diff line
		@@ -527,6 +527,7 @@ trgm
		truecolor
		twistie
		twocol
		ufffd
		uids
		unban
		uncacheable

core/modules/jsonapi/tests/src/Unit/Encoder/JsonEncoderTest.php

0 → 100644

+70 −0

Original line number	Diff line number	Diff line
		<?php

		declare(strict_types=1);

		namespace Drupal\Tests\jsonapi\Unit\Encoder;

		use Drupal\jsonapi\Encoder\JsonEncoder;
		use Drupal\Tests\UnitTestCase;
		use PHPUnit\Framework\Attributes\CoversClass;
		use PHPUnit\Framework\Attributes\Group;

		/**
		* Tests the JSON:API encoder.
		*
		* @internal
		*/
		#[CoversClass(JsonEncoder::class)]
		#[Group('jsonapi')]
		class JsonEncoderTest extends UnitTestCase {

		/**
		* The encoder under test.
		*/
		protected JsonEncoder $encoder;

		/**
		* {@inheritdoc}
		*/
		protected function setUp(): void {
		parent::setUp();
		$this->encoder = new JsonEncoder();
		}

		/**
		* Tests the supportsEncoding() method.
		*/
		public function testSupportsEncoding(): void {
		$this->assertTrue($this->encoder->supportsEncoding('api_json'));
		$this->assertFalse($this->encoder->supportsEncoding('json'));
		$this->assertFalse($this->encoder->supportsEncoding('xml'));
		}

		/**
		* Tests JSON:API encoder inherits control character handling.
		*
		* This verifies that the JSON:API encoder correctly inherits the
		* JSON_INVALID_UTF8_SUBSTITUTE flag from the parent serialization encoder.
		* The comprehensive tests for this functionality are in the serialization
		* module's JsonEncoderTest.
		*
		* @see \Drupal\Tests\serialization\Unit\Encoder\JsonEncoderTest
		* @see https://www.drupal.org/project/drupal/issues/3549107
		*/
		public function testInheritsControlCharacterHandling(): void {
		// Test that invalid UTF-8 is handled (would fail without the flag).
		$input = "Test\x80Data";
		$encoded = $this->encoder->encode($input, 'api_json');

		// Verify it's valid JSON.
		$this->assertJson($encoded, 'Encoded output should be valid JSON even with invalid UTF-8.');

		// Verify the replacement character is present.
		$this->assertStringContainsString('\ufffd', $encoded, 'Invalid UTF-8 should be replaced with U+FFFD.');

		// Verify it can be decoded.
		json_decode($encoded);
		$this->assertSame(JSON_ERROR_NONE, json_last_error(), 'Encoded JSON should be decodable without errors.');
		}

		}

core/modules/serialization/src/Encoder/JsonEncoder.php

+4 −1

Original line number	Diff line number	Diff line
		@@ -29,7 +29,10 @@ public function __construct(?JsonEncode $encodingImpl = NULL, ?JsonDecode $decod
		// Encode <, >, ', &, and " for RFC4627-compliant JSON, which may also be
		// embedded into HTML.
		// @see \Symfony\Component\HttpFoundation\JsonResponse
		$json_encoding_options = JSON_HEX_TAG \| JSON_HEX_APOS \| JSON_HEX_AMP \| JSON_HEX_QUOT;
		// Additionally, substitute invalid UTF-8 sequences to ensure control
		// characters and malformed data are properly handled.
		// @see https://www.drupal.org/project/drupal/issues/3549107
		$json_encoding_options = JSON_HEX_TAG \| JSON_HEX_APOS \| JSON_HEX_AMP \| JSON_HEX_QUOT \| JSON_INVALID_UTF8_SUBSTITUTE;
		$this->encodingImpl = $encodingImpl ?: new JsonEncode([JsonEncode::OPTIONS => $json_encoding_options]);
		$this->decodingImpl = $decodingImpl ?: new JsonDecode([JsonDecode::ASSOCIATIVE => TRUE]);
		}

core/modules/serialization/tests/src/Unit/Encoder/JsonEncoderTest.php

+85 −4

Original line number	Diff line number	Diff line
		@@ -16,15 +16,96 @@
		#[Group('serialization')]
		class JsonEncoderTest extends UnitTestCase {

		/**
		* The encoder under test.
		*/
		protected JsonEncoder $encoder;

		/**
		* {@inheritdoc}
		*/
		protected function setUp(): void {
		parent::setUp();
		$this->encoder = new JsonEncoder();
		}

		/**
		* Tests the supportsEncoding() method.
		*/
		public function testSupportsEncoding(): void {
		$encoder = new JsonEncoder();
		$this->assertTrue($this->encoder->supportsEncoding('json'));
		$this->assertTrue($this->encoder->supportsEncoding('ajax'));
		$this->assertFalse($this->encoder->supportsEncoding('xml'));
		}

		/**
		* Tests that invalid UTF-8 is handled via JSON_INVALID_UTF8_SUBSTITUTE.
		*
		* @see https://www.drupal.org/project/drupal/issues/3549107
		*/
		public function testEncodeInvalidUtf8IsSubstituted(): void {
		// A representative invalid UTF-8 sequence that would previously cause
		// json_encode() to fail.
		$input = "Test\x80Data";
		$encoded = $this->encoder->encode($input, 'json');

		// Verify it's valid JSON (would fail without the flag).
		$this->assertJson($encoded, 'Encoded output should be valid JSON even with invalid UTF-8.');

		// Verify the replacement character is present.
		$this->assertStringContainsString('\\ufffd', $encoded, 'Invalid UTF-8 should be replaced with U+FFFD.');

		// Verify it can be decoded.
		json_decode($encoded);
		$this->assertSame(JSON_ERROR_NONE, json_last_error(), 'Encoded JSON should be decodable without errors.');
		}

		/**
		* Tests that HTML-unsafe characters are still escaped.
		*
		* This ensures that existing JSON_HEX_* behavior is preserved after adding
		* JSON_INVALID_UTF8_SUBSTITUTE.
		*/
		public function testHtmlUnsafeCharactersAreEscaped(): void {
		$input = "<script>alert('test & \"hack\"');</script>";
		$encoded = $this->encoder->encode($input, 'json');

		// Verify it's valid JSON.
		$this->assertJson($encoded, 'HTML-unsafe characters should produce valid JSON.');

		// Verify HTML-unsafe characters are escaped as hex codes.
		$this->assertStringContainsString('\\u003C', $encoded, '< should be escaped to \\u003C.');
		$this->assertStringContainsString('\\u003E', $encoded, '> should be escaped to \\u003E.');
		$this->assertStringContainsString('\\u0027', $encoded, "' should be escaped to \\u0027.");
		$this->assertStringContainsString('\\u0026', $encoded, '& should be escaped to \\u0026.');
		$this->assertStringContainsString('\\u0022', $encoded, '" should be escaped to \\u0022.');
		}

		/**
		* Simple structured data smoke test.
		*
		* This verifies that the encoder works for nested arrays and that invalid
		* UTF-8 inside the structure is still handled correctly.
		*/
		public function testStructuredDataSmokeTest(): void {
		$data = [
		'title' => 'Example',
		'body' => "Content with invalid UTF-8: \x80",
		'metadata' => [
		'tags' => ['one', 'two'],
		],
		];

		$encoded = $this->encoder->encode($data, 'json');

		$this->assertJson($encoded, 'Structured data should produce valid JSON.');
		$this->assertStringContainsString('\\ufffd', $encoded, 'Invalid UTF-8 in nested data should be replaced.');

		$this->assertTrue($encoder->supportsEncoding('json'));
		$this->assertTrue($encoder->supportsEncoding('ajax'));
		$this->assertFalse($encoder->supportsEncoding('xml'));
		$decoded = json_decode($encoded, TRUE);
		$this->assertSame(JSON_ERROR_NONE, json_last_error(), 'Structured data should be decodable.');
		$this->assertIsArray($decoded, 'Decoded data should be an array.');
		$this->assertArrayHasKey('title', $decoded, 'Decoded data should have title key.');
		$this->assertArrayHasKey('metadata', $decoded, 'Decoded data should have metadata key.');
		}

		}