listenability-wikipedia.php

<?php
/**
 * Plugin Name: Listenability: Wikipedia
 * Plugin URI: https://wordpress.org/plugins/listenability/
 * Author: Weston Ruter
 * Description: Format the text for listening (the say command).
 */

/**
 * Strip out elements from Wikipedia articles.
 *
 * @param array $args {
 *     @type DOMDocument $document
 *     @type DOMXPath    $xpath
 *     @type string      $url
 *     @type array|null  $readability_parser_response
 * }
 */
add_action( 'listenability_document', function ( $args ) {
	$is_wikipedia = ( false !== strpos( parse_url( $args['url'], PHP_URL_HOST ), 'wikipedia' ) );
	if ( ! $is_wikipedia ) {
		return;
	}

	$removed_elements = array(
		'//span[@class = "mw-editsection"]',
		'//sup[@class = "reference"]',
		'//sup[@class = "Template-Fact"]',
		'//div[@class = "reflist"]',
		'//figure',
		'//div[@class = "printfooter"]',
		'//div[@class = "catlinks"]',
		'//div[@class = "toc"]',
		'//table',
	);
	foreach ( $args['xpath']->query( join( ' | ', $removed_elements ) ) as $element ) {
		$element->parentNode->removeChild( $element );
	}
} );