Add sanitization reporting

amp.php

@@ -88,6 +88,7 @@ function amp_after_setup_theme() {
 	}
 
 	add_action( 'init', 'amp_init' );
+	add_action( 'rest_api_init', 'AMP_Mutation_Utils::amp_rest_validation' );
 	add_action( 'widgets_init', 'AMP_Theme_Support::register_widgets' );
 	add_action( 'admin_init', 'AMP_Options_Manager::register_settings' );
 	add_filter( 'amp_post_template_analytics', 'amp_add_custom_analytics' );

includes/class-amp-autoloader.php

@@ -78,6 +78,7 @@ class AMP_Autoloader {
 		'AMP_DOM_Utils'                               => 'includes/utils/class-amp-dom-utils',
 		'AMP_HTML_Utils'                              => 'includes/utils/class-amp-html-utils',
 		'AMP_Image_Dimension_Extractor'               => 'includes/utils/class-amp-image-dimension-extractor',
+		'AMP_Mutation_Utils'                          => 'includes/utils/class-amp-mutation-utils',
 		'AMP_String_Utils'                            => 'includes/utils/class-amp-string-utils',
 		'AMP_WP_Utils'                                => 'includes/utils/class-amp-wp-utils',
 		'AMP_Widget_Archives'                         => 'includes/widgets/class-amp-widget-archives',

includes/class-amp-theme-support.php

@@ -506,13 +506,25 @@ public static function get_amp_component_scripts() {
 	 * Start output buffering.
 	 */
 	public static function start_output_buffering() {
-		ob_start( array( __CLASS__, 'finish_output_buffering' ) );
+		ob_start( array( __CLASS__, 'finish_buffer_add_header' ) );
 	}
 
 	/**
-	 * Finish output buffering.
+	 * Get the result of the output buffering, and add a header.
 	 *
 	 * @todo Do this in shutdown instead of output buffering callback?
+	 * @param string $output Buffered output.
+	 * @return string Finalized output.
+	 */
+	public static function finish_buffer_add_header( $output ) {
+		$markup = self::finish_output_buffering( $output );
+		AMP_Mutation_Utils::add_header();
+		return $markup;
+	}
+
+	/**
+	 * Finish output buffering.
+	 *
 	 * @global int $content_width
 	 * @param string $output Buffered output.
 	 * @return string Finalized output.
@@ -523,6 +535,7 @@ public static function finish_output_buffering( $output ) {
 		$dom  = AMP_DOM_Utils::get_dom( $output );
 		$args = array(
 			'content_max_width' => ! empty( $content_width ) ? $content_width : AMP_Post_Template::CONTENT_MAX_WIDTH, // Back-compat.
+			'mutation_callback' => 'AMP_Mutation_Utils::track_removed',
 		);
 
 		$assets = AMP_Content_Sanitizer::sanitize_document( $dom, self::$sanitizer_classes, $args );

includes/sanitizers/class-amp-form-sanitizer.php

@@ -86,6 +86,9 @@ public function sanitize() {
 				}
 			} elseif ( 'post' === $method ) {
 				$node->removeAttribute( 'action' );
+				if ( isset( $this->args['mutation_callback'] ) ) {
+					call_user_func( $this->args['mutation_callback'], $node, 'removed_attr', 'action' );
+				}
 				if ( ! $xhr_action ) {
 					$node->setAttribute( 'action-xhr', $action_url );
 				} elseif ( 'http://' === substr( $xhr_action, 0, 7 ) ) {

includes/sanitizers/class-amp-iframe-sanitizer.php

@@ -74,7 +74,7 @@ public function sanitize() {
 			 * @see: https://github.com/ampproject/amphtml/issues/2261
 			 */
 			if ( empty( $new_attributes['src'] ) ) {
-				$node->parentNode->removeChild( $node );
+				$this->remove_child( $node );
 				continue;
 			}
 
@@ -95,11 +95,11 @@ public function sanitize() {
 				$parent_node->replaceChild( $new_node, $node );
 			} else {
 				// AMP does not like iframes in <p> tags.
-				$parent_node->removeChild( $node );
+				$this->remove_child( $node );
 				$parent_node->parentNode->insertBefore( $new_node, $parent_node->nextSibling );
 
 				if ( AMP_DOM_Utils::is_node_empty( $parent_node ) ) {
-					$parent_node->parentNode->removeChild( $parent_node );
+					$this->remove_child( $parent_node );  // phpcs:ignore WordPress.NamingConventions.ValidVariableName.NotSnakeCaseMemberVar.
 				}
 			}
 		}
@@ -192,4 +192,25 @@ private function build_placeholder( $parent_attributes ) {
 
 		return $placeholder_node;
 	}
+
+	/**
+	 * Removes a child of a node.
+	 *
+	 * Also, calls the mutation callback for it.
+	 * This tracks all the nodes that were removed.
+	 *
+	 * @since 0.7
+	 *
+	 * @param DOMNode $child The node to remove.
+	 * @return void.
+	 */
+	public function remove_child( $child ) {
+		if ( method_exists( $child->parentNode, 'removeChild' ) ) { // phpcs:ignore WordPress.NamingConventions.ValidVariableName.NotSnakeCaseMemberVar.
+			$child->parentNode->removeChild( $child ); // phpcs:ignore WordPress.NamingConventions.ValidVariableName.NotSnakeCaseMemberVar.
+			if ( isset( $this->args['mutation_callback'] ) ) {
+				call_user_func( $this->args['mutation_callback'], $child, 'removed' );
+			}
+		}
+	}
+
 }

includes/sanitizers/class-amp-style-sanitizer.php

@@ -78,6 +78,9 @@ private function collect_styles_recursive( $node ) {
 					$this->styles[ '.' . $class_name ] = $style;
 				}
 				$node->removeAttribute( 'style' );
+				if ( isset( $this->args['mutation_callback'] ) ) {
+					call_user_func( $this->args['mutation_callback'], $node, 'removed_attr', 'style' );
+				}
 			}
 		}
 

includes/sanitizers/class-amp-tag-and-attribute-sanitizer.php

@@ -595,6 +595,9 @@ private function sanitize_disallowed_attributes_in_node( $node, $attr_spec_list
 
 		foreach ( $attrs_to_remove as $attr_name ) {
 			$node->removeAttribute( $attr_name );
+			if ( isset( $this->args['mutation_callback'] ) ) {
+				call_user_func( $this->args['mutation_callback'], $node, 'removed_attr', $attr_name );
+			}
 		}
 	}
 
@@ -701,6 +704,9 @@ private function delegated_sanitize_disallowed_attribute_values_in_node( $node,
 				$node->setAttribute( $attr_name, '' );
 			} else {
 				$node->removeAttribute( $attr_name );
+				if ( isset( $this->args['mutation_callback'] ) ) {
+					call_user_func( $this->args['mutation_callback'], $node, 'removed_attr', $attr_name );
+				}
 			}
 		}
 	}
@@ -1279,12 +1285,18 @@ private function remove_node( $node ) {
 		$parent = $node->parentNode;
 		if ( $node && $parent ) {
 			$parent->removeChild( $node );
+			if ( isset( $this->args['mutation_callback'] ) ) {
+				call_user_func( $this->args['mutation_callback'], $node, 'removed' );
+			}
 		}
 		while ( $parent && ! $parent->hasChildNodes() && 'body' !== $parent->nodeName ) {
 			$node   = $parent;
 			$parent = $parent->parentNode;
 			if ( $parent ) {
 				$parent->removeChild( $node );
+				if ( isset( $this->args['mutation_callback'] ) ) {
+					call_user_func( $this->args['mutation_callback'], $node, 'removed' );
+				}
 			}
 		}
 	}

includes/sanitizers/class-amp-video-sanitizer.php

@@ -94,6 +94,9 @@ public function sanitize() {
 			 */
 			if ( 0 === $new_node->childNodes->length && empty( $new_attributes['src'] ) ) {
 				$node->parentNode->removeChild( $node );
+				if ( isset( $this->args['mutation_callback'] ) ) {
+					call_user_func( $this->args['mutation_callback'], $node, 'removed' );
+				}
 			} else {
 				$node->parentNode->replaceChild( $new_node, $node );
 			}

includes/utils/class-amp-mutation-utils.php

@@ -0,0 +1,212 @@
+<?php
+/**
+ * Class AMP_Mutation_Utils
+ *
+ * @package AMP
+ */
+
+/**
+ * Class AMP_Mutation_Utils
+ *
+ * @since 0.7
+ */
+class AMP_Mutation_Utils {
+
+	/**
+	 * The argument if an attribute was removed.
+	 *
+	 * @var array.
+	 */
+	const ATTRIBUTE_REMOVED = 'removed_attr';
+
+	/**
+	 * The argument if a node was removed.
+	 *
+	 * @var array.
+	 */
+	const NODE_REMOVED = 'removed';
+
+	/**
+	 * Key for the markup value in the REST API endpoint.
+	 *
+	 * @var string.
+	 */
+	const MARKUP_KEY = 'markup';
+
+	/**
+	 * The attributes that the sanitizer removed.
+	 *
+	 * @var array.
+	 */
+	public static $removed_attributes;
+
+	/**
+	 * The nodes that the sanitizer removed.
+	 *
+	 * @var array.
+	 */
+	public static $removed_nodes;
+
+	/**
+	 * Tracks when a sanitizer removes an attribute or node.
+	 *
+	 * @param DOMNode|DOMElement $node The node in which there was a removal.
+	 * @param string             $removal_type The removal: 'removed_attr' for an attribute, or 'removed' for a node or element.
+	 * @param string             $attr_name The name of the attribute removed (optional).
+	 * @return void.
+	 */
+	public static function track_removed( $node, $removal_type, $attr_name = null ) {
+		if ( ( self::ATTRIBUTE_REMOVED === $removal_type ) && isset( $attr_name ) ) { // phpcs:ignore WordPress.NamingConventions.ValidVariableName.NotSnakeCaseMemberVar
+			self::$removed_attributes = self::increment_count( self::$removed_attributes, $attr_name );
+		} elseif ( ( self::NODE_REMOVED === $removal_type ) && isset( $node->nodeName ) ) {
+			self::$removed_nodes = self::increment_count( self::$removed_nodes, $node->nodeName ); // phpcs:ignore WordPress.NamingConventions.ValidVariableName.NotSnakeCaseMemberVar
+		}
+	}
+
+	/**
+	 * Tracks when a sanitizer removes an attribute or node.
+	 *
+	 * @param array  $histogram The count of attributes or nodes removed.
+	 * @param string $key The attribute or node name removed.
+	 * @return array $histogram The incremented histogram.
+	 */
+	public static function increment_count( $histogram, $key ) {
+		$current_value     = isset( $histogram[ $key ] ) ? $histogram[ $key ] : 0;
+		$histogram[ $key ] = $current_value + 1;
+		return $histogram;
+	}
+
+	/**
+	 * Gets whether a node was removed in a sanitizer.
+	 *
+	 * @return boolean.
+	 */
+	public static function was_node_removed() {
+		return ! empty( self::$removed_nodes );
+	}
+
+	/**
+	 * Processes markup, to determine AMP validity.
+	 *
+	 * Passes $markup through the AMP sanitizers.
+	 * Also passes a 'mutation_callback' to keep track of stripped attributes and nodes.
+	 *
+	 * @param string $markup The markup to process.
+	 * @return void.
+	 */
+	public static function process_markup( $markup ) {
+		$args = array(
+			'content_max_width' => ! empty( $content_width ) ? $content_width : AMP_Post_Template::CONTENT_MAX_WIDTH,
+			'mutation_callback' => 'AMP_Mutation_Utils::track_removed',
+		);
+		AMP_Content_Sanitizer::sanitize( $markup, amp_get_content_sanitizers(), $args );
+	}
+
+	/**
+	 * Registers the REST API endpoint for validation.
+	 *
+	 * @return void.
+	 */
+	public static function amp_rest_validation() {
+		register_rest_route( 'amp-wp/v1', '/validate', array(
+			'methods'             => 'POST',
+			'callback'            => array( __CLASS__, 'validate_markup' ),
+			'args'                => array(
+				self::MARKUP_KEY => array(
+					'validate_callback' => array( __CLASS__, 'validate_arg' ),
+				),
+			),
+			'permission_callback' => array( __CLASS__, 'permission' ),
+		) );
+	}
+
+	/**
+	 * The permission callback for the REST request.
+	 *
+	 * @return boolean $has_permission Whether the current user has the permission.
+	 */
+	public static function permission() {
+		return current_user_can( 'edit_posts' );
+	}
+
+	/**
+	 * Validate the markup passed to the REST API.
+	 *
+	 * @param WP_REST_Request $request The REST request.
+	 * @return array|WP_Error.
+	 */
+	public static function validate_markup( WP_REST_Request $request ) {
+		$json = $request->get_json_params();
+		if ( empty( $json[ self::MARKUP_KEY ] ) ) {
+			return new WP_Error( 'no_markup', 'No markup passed to validator', array(
+				'status' => 404,
+			) );
+		}
+
+		return self::get_response( $json[ self::MARKUP_KEY ] );
+	}
+
+	/**
+	 * Gets the AMP validation response.
+	 *
+	 * If $markup isn't passed,
+	 * It will return the validation errors the sanitizers found in rendering the page.
+	 *
+	 * @param string $markup To validate for AMP compatibility (optional).
+	 * @return array $response The AMP validity of the markup.
+	 */
+	public static function get_response( $markup = null ) {
+		if ( isset( $markup ) ) {
+			self::process_markup( $markup );
+		}
+		$response = array(
+			'has_error'          => self::was_node_removed(),
+			'removed_nodes'      => self::$removed_nodes,
+			'removed_attributes' => self::$removed_attributes,
+		);
+		self::reset_removed();
+		return $response;
+	}
+
+	/**
+	 * Reset the stored removed nodes and attributes.
+	 *
+	 * After testing if the markup is valid,
+	 * these static values will remain.
+	 * So reset them in case another test is needed.
+	 *
+	 * @return void.
+	 */
+	public static function reset_removed() {
+		self::$removed_nodes      = null;
+		self::$removed_attributes = null;
+	}
+
+	/**
+	 * Validate the argument in the REST API request.
+	 *
+	 * It would be ideal to simply pass 'is_string' in register_rest_route().
+	 * But it always returned false.
+	 *
+	 * @param mixed $arg The argument to validate.
+	 * @return boolean $is_valid Whether the argument is valid.
+	 */
+	public static function validate_arg( $arg ) {
+		return is_string( $arg );
+	}
+
+	/**
+	 * Output AMP validation data in the response header of a frontend GET request.
+	 *
+	 * This must be called before the document output begins.
+	 * Because the document is buffered,
+	 * The sanitizers run after the 'send_headers' action.
+	 * So it's not possible to call this function on that hook.
+	 *
+	 * @return void. $header The filtered response header.
+	 */
+	public static function add_header() {
+		header( sprintf( 'AMP-Validation-Error: %s', wp_json_encode( self::get_response() ) ) );
+	}
+
+}

phpcs.xml

@@ -11,6 +11,9 @@
 		</properties>
 	</rule>
 
+	<rule ref="WordPress.Files.FileName.InvalidClassFileName">
+		<exclude-pattern>tests/*</exclude-pattern>
+	</rule>
 	<rule ref="WordPress.Arrays.MultipleStatementAlignment.LongIndexSpaceBeforeDoubleArrow">
 		<exclude-pattern>tests/test-tag-and-attribute-sanitizer.php</exclude-pattern>
 	</rule>