Unwrap all embeds with paragraph tags

includes/class-amp-theme-support.php

@@ -6,6 +6,7 @@
  */
 
 use AmpProject\Amp;
+use AmpProject\AmpWP\Embed\Registerable;
 use AmpProject\AmpWP\Option;
 use AmpProject\AmpWP\RemoteRequest\CachedRemoteGetRequest;
 use AmpProject\AmpWP\ConfigurationArgument;
@@ -1179,8 +1180,9 @@ public static function register_content_embed_handlers() {
 				);
 				continue;
 			}
-
-			$embed_handler->register_embed();
+			if ( $embed_handler instanceof Registerable ) {
+				$embed_handler->register_embed();
+			}
 			$embed_handlers[] = $embed_handler;
 		}
 

includes/embeds/class-amp-base-embed-handler.php

@@ -7,6 +7,8 @@
  * @package  AMP
  */
 
+use AmpProject\Dom\Document;
+
 /**
  * Class AMP_Base_Embed_Handler
  */
@@ -40,14 +42,18 @@ abstract class AMP_Base_Embed_Handler {
 	protected $did_convert_elements = false;
 
 	/**
-	 * Registers embed.
+	 * Default AMP tag to be used when sanitizing embeds.
+	 *
+	 * @var string
 	 */
-	abstract public function register_embed();
+	protected $amp_tag = 'amp-iframe';
 
 	/**
-	 * Unregisters embed.
+	 * Base URL used for identifying embeds.
+	 *
+	 * @var string
 	 */
-	abstract public function unregister_embed();
+	protected $base_embed_url = '';
 
 	/**
 	 * Constructor.
@@ -64,6 +70,56 @@ public function __construct( $args = [] ) {
 		);
 	}
 
+	/**
+	 * Sanitize all embeds on the page to be AMP compatible.
+	 *
+	 * @param Document $dom DOM.
+	 */
+	public function sanitize_raw_embeds( Document $dom ) {
+		$nodes = $this->get_raw_embed_nodes( $dom );
+
+		if ( null === $nodes ) {
+			// Bail if the embed handler returns null.
+			return;
+		}
+
+		if ( 0 === $nodes->length ) {
+			return;
+		}
+
+		foreach ( $nodes as $node ) {
+			if ( ! $this->is_raw_embed( $node ) ) {
+				continue;
+			}
+			$this->sanitize_raw_embed( $node );
+		}
+	}
+
+	/**
+	 * Determine if the node is indeed a raw embed.
+	 *
+	 * @param DOMElement $node DOM element.
+	 * @return bool True if it is a raw embed, false otherwise.
+	 */
+	protected function is_raw_embed( DOMElement $node ) {
+		return $node->parentNode && $this->amp_tag !== $node->parentNode->nodeName;
+	}
+
+	/**
+	 * Get all raw embeds from the DOM.
+	 *
+	 * @param Document $dom Document.
+	 * @return DOMNodeList|null A list of DOMElement nodes, or null if not implemented.
+	 */
+	abstract protected function get_raw_embed_nodes( Document $dom );
+
+	/**
+	 * Make embed AMP compatible.
+	 *
+	 * @param DOMElement $node DOM element.
+	 */
+	abstract protected function sanitize_raw_embed( DOMElement $node );
+
 	/**
 	 * Get mapping of AMP component names to AMP script URLs.
 	 *
@@ -107,4 +163,90 @@ function ( $attr_name ) {
 		}
 		return wp_array_slice_assoc( $matches, $attribute_names );
 	}
+
+	/**
+	 * Get all child elements of the specified element.
+	 *
+	 * @param DOMElement $node Element.
+	 * @return DOMElement[] Array of child elements for specified element.
+	 */
+	protected function get_child_elements( DOMElement $node ) {
+		return array_filter(
+			iterator_to_array( $node->childNodes ),
+			static function ( DOMNode $child ) {
+				return $child instanceof DOMElement;
+			}
+		);
+	}
+
+	/**
+	 * Replace the node's parent with itself if the parent is a <p> tag, has no attributes and has no other children.
+	 * This usually happens while `wpautop()` processes the element.
+	 *
+	 * @since 1.6
+	 *
+	 * @param DOMElement $node Node.
+	 */
+	protected function unwrap_p_element( DOMElement $node ) {
+		$parent_node = $node->parentNode;
+		while ( $parent_node && ! ( $parent_node instanceof DOMElement ) ) {
+			$parent_node = $parent_node->parentNode;
+		}
+
+		if ( $parent_node instanceof DOMElement && 'p' === $parent_node->nodeName && false === $parent_node->hasAttributes() ) {
+			$child_element_count = count( $this->get_child_elements( $parent_node ) );
+			if ( 1 === $child_element_count ) {
+				$parent_node->parentNode->replaceChild( $node, $parent_node );
+			}
+		}
+	}
+
+	/**
+	 * Removes the node's nearest <script> sibling with a `src` attribute containing the base `src` URL provided.
+	 *
+	 * @since 1.6
+	 *
+	 * @param DOMElement $node         The DOMNode to whose sibling is the script to be removed.
+	 * @param string     $base_src_url Script URL to match against.
+	 * @param string     $content      Text content of node to match against.
+	 * @param bool       $is_next_to   Whether the script sibling is next or preceding the specified element.
+	 */
+	protected function remove_script_sibling( DOMElement $node, $base_src_url, $content = '', $is_next_to = true ) {
+		$sibling_location = $is_next_to ? 'nextSibling' : 'previousSibling';
+		$element_sibling  = $node->{$sibling_location};
+
+		while ( $element_sibling && ! ( $element_sibling instanceof DOMElement ) ) {
+			$element_sibling = $element_sibling->{$sibling_location};
+		}
+
+		// Handle case where script is wrapped in paragraph by wpautop.
+		if ( $element_sibling instanceof DOMElement && 'p' === $element_sibling->nodeName ) {
+			$children_elements = array_values( $this->get_child_elements( $element_sibling ) );
+
+			if (
+				1 === count( $children_elements ) &&
+				'script' === $children_elements[0]->nodeName &&
+				(
+					( $base_src_url && false !== strpos( $children_elements[0]->getAttribute( 'src' ), $base_src_url ) ) ||
+					( $content && false !== strpos( $children_elements[0]->textContent, $content ) )
+				)
+			) {
+				$element_sibling->parentNode->removeChild( $element_sibling );
+				return;
+			}
+		}
+
+		// Handle case where script is immediately following.
+		$is_embed_script = (
+			$element_sibling instanceof DOMElement &&
+			'script' === strtolower( $element_sibling->nodeName ) &&
+			(
+				( $base_src_url && false !== strpos( $element_sibling->getAttribute( 'src' ), $base_src_url ) ) ||
+				( $content && false !== strpos( $element_sibling->textContent, $content ) )
+			)
+		);
+		if ( $is_embed_script ) {
+			$element_sibling->parentNode->removeChild( $element_sibling );
+		}
+	}
 }

includes/embeds/class-amp-core-block-handler.php

@@ -5,6 +5,7 @@
  * @package AMP
  */
 
+use AmpProject\AmpWP\Embed\Registerable;
 use AmpProject\Attribute;
 use AmpProject\Dom\Document;
 
@@ -13,7 +14,7 @@
  *
  * @since 1.0
  */
-class AMP_Core_Block_Handler extends AMP_Base_Embed_Handler {
+class AMP_Core_Block_Handler extends AMP_Base_Embed_Handler implements Registerable {
 
 	/**
 	 * Attribute to store the original width on a video or iframe just before WordPress removes it.
@@ -355,4 +356,22 @@ private function process_text_widgets( Document $dom ) {
 			}
 		}
 	}
+
+	/**
+	 * Get all raw embeds from the DOM.
+	 *
+	 * @param Document $dom Document.
+	 * @return DOMNodeList|null A list of DOMElement nodes, or null if not implemented.
+	 */
+	protected function get_raw_embed_nodes( Document $dom ) { // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
+		return null;
+	}
+
+	/**
+	 * Make embed AMP compatible.
+	 *
+	 * @param DOMElement $node DOM element.
+	 */
+	protected function sanitize_raw_embed( DOMElement $node ) { // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
+	}
 }

includes/embeds/class-amp-crowdsignal-embed-handler.php

@@ -8,59 +8,128 @@
  * @since 1.2
  */
 
+use AmpProject\AmpWP\Embed\Registerable;
+use AmpProject\Dom\Document;
+
 /**
  * Class AMP_Crowdsignal_Embed_Handler
  */
-class AMP_Crowdsignal_Embed_Handler extends AMP_Base_Embed_Handler {
+class AMP_Crowdsignal_Embed_Handler extends AMP_Base_Embed_Handler implements Registerable {
 
 	/**
-	 * Register embed.
+	 * Register the embed.
+	 *
+	 * @return void
 	 */
 	public function register_embed() {
-		add_filter( 'embed_oembed_html', [ $this, 'filter_embed_oembed_html' ], 10, 3 );
+		if ( version_compare( get_bloginfo( 'version' ), '5.2', '>=' ) ) {
+			return;
+		}
+
+		// The oEmbed providers for CrowdSignal embeds are outdated on WP < 5.1. Updating the providers here will
+		// allow for the oEmbed HTML to be fetched, and can then sanitized later below.
+		$formats = [
+			'#https?://(.+\.)?polldaddy\.com/.*#i',
+			'#https?://poll\.fm/.*#i',
+			'#https?://(.+\.)?survey\.fm/.*#i', // Not available on WP 5.2.
+		];
+
+		foreach ( $formats as $format ) {
+			wp_oembed_add_provider( $format, 'https://api.crowdsignal.com/oembed', true );
+		}
 	}
 
 	/**
-	 * Unregister embed.
+	 * Unregister the embed.
+	 *
+	 * @return void
 	 */
 	public function unregister_embed() {
-		remove_filter( 'embed_oembed_html', [ $this, 'filter_embed_oembed_html' ], 10 );
 	}
 
 	/**
-	 * Filter oEmbed HTML for Crowdsignal for AMP output.
+	 * Get all raw embeds from the DOM.
 	 *
-	 * @param string $cache Cache for oEmbed.
-	 * @param string $url   Embed URL.
-	 * @param array  $attr  Shortcode attributes.
-	 * @return string Embed.
+	 * @param Document $dom Document.
+	 * @return DOMNodeList A list of DOMElement nodes.
 	 */
-	public function filter_embed_oembed_html( $cache, $url, $attr ) {
-		$parsed_url = wp_parse_url( $url );
-		if ( empty( $parsed_url['host'] ) || empty( $parsed_url['path'] ) || ! preg_match( '#(^|\.)(?P<host>polldaddy\.com|crowdsignal\.com|survey\.fm|poll\.fm)#', $parsed_url['host'], $matches ) ) {
-			return $cache;
-		}
+	protected function get_raw_embed_nodes( Document $dom ) {
+		$queries = [
+			// For poll embeds.
+			'//iframe[ @class="cs-iframe-embed" and starts-with( @src, "https://poll.fm/" ) ]',
+			// For survey embeds.
+			'//div[ @class="pd-embed" and @data-settings ]',
+		];
 
-		$parsed_url['host'] = $matches['host'];
+		return $dom->xpath->query( implode( ' | ', $queries ) );
+	}
 
-		$output = '';
+	/**
+	 * Make embed AMP compatible.
+	 *
+	 * @param DOMElement $node DOM element.
+	 */
+	protected function sanitize_raw_embed( DOMElement $node ) {
+		$is_poll = 'cs-iframe-embed' === $node->getAttribute( 'class' );
 
-		// Poll oEmbed responses include noscript which can be used as the AMP response.
-		if ( preg_match( '#<noscript>(.+?)</noscript>#s', $cache, $matches ) ) {
-			$output = $matches[1];
+		if ( $is_poll ) {
+			$this->sanitize_poll_embed( $node );
+		} else {
+			$this->sanitize_survey_embed( $node );
 		}
+	}
 
-		if ( empty( $output ) ) {
-			if ( ! empty( $attr['title'] ) ) {
-				$name = $attr['title'];
-			} elseif ( 'survey.fm' === $parsed_url['host'] || preg_match( '#^/s/#', $parsed_url['path'] ) ) {
-				$name = __( 'View Survey', 'amp' );
-			} else {
-				$name = __( 'View Poll', 'amp' );
-			}
-			$output = sprintf( '<a href="%s" target="_blank">%s</a>', esc_url( $url ), esc_html( $name ) );
+	/**
+	 * Sanitize poll embed.
+	 *
+	 * @param DOMElement $node Poll embed.
+	 */
+	private function sanitize_poll_embed( DOMElement $node ) {
+		// Replace the `noscript` parent element with the iframe.
+		$node->parentNode->parentNode->replaceChild( $node, $node->parentNode );
+
+		$this->unwrap_p_element( $node );
+		$this->remove_script_sibling( $node, 'https://secure.polldaddy.com', '', false );
+	}
+
+	/**
+	 * Sanitize survey embed.
+	 *
+	 * @param DOMElement $node Survey embed.
+	 */
+	private function sanitize_survey_embed( DOMElement $node ) {
+		$settings = json_decode( $node->getAttribute( 'data-settings' ), false );
+
+		// We can't form the iframe URL without a domain and survey ID.
+		if ( ! ( property_exists( $settings, 'domain' ) || property_exists( $settings, 'id' ) ) ) {
+			return;
 		}
 
-		return $output;
+		// Logic for building the iframe `src` can be found in https://polldaddy.com/survey.js.
+		$iframe_src = sprintf(
+			'https://%s/%s?%s',
+			$settings->domain,
+			$settings->id,
+			property_exists( $settings, 'auto' ) && $settings->auto ? 'ft=1&iframe=' . amp_get_current_url() : 'iframe=1'
+		);
+
+		$iframe_node = AMP_DOM_Utils::create_node(
+			Document::fromNode( $node ),
+			'iframe',
+			[
+				'src'               => $iframe_src,
+				'layout'            => 'responsive',
+				'width'             => 600,
+				'height'            => 600,
+				'frameborder'       => 0,
+				'scrolling'         => 'no',
+				'allowtransparency' => 'true',
+				'sandbox'           => 'allow-scripts allow-same-origin',
+			]
+		);
+
+		$this->remove_script_sibling( $node, null, 'https://polldaddy.com/survey.js' );
+
+		$node->parentNode->replaceChild( $iframe_node, $node );
 	}
 }