[SmartCrawl Pro] Multilingual sitemap (WPML integration)

smartcrawl-multilingual-sitemap.php

<?php
/**
 * Plugin Name: [SmartCrawl Pro] Multilingual sitemap (WPML integration)
 * Description: Improves the SmartCrawl sitemaps with multilingual-ready capabilities when the WPML plugin is present
 * Author: Anderson Salas @ WPMUDEV
 * Task: SLS-4082
 * Author URI: https://premium.wpmudev.org
 * License: GPLv2 or later
 */

add_action( 'plugins_loaded', function() {
	if ( ! class_exists( 'Smartcrawl_Sitemap_Post_Fetcher' ) ) {
		return; // SmartCrawl is not installed/enabled.
	}

	if ( ! defined( 'ICL_SITEPRESS_VERSION' ) ) {
		return; // WPML is not installed/enabled.
	}

	add_filter( 'wds_partial_sitemap_items', function( $items, $type, $page ) {
		global $wpdb;

		// ---------------------------------------------------------------------------- //
		// Modify this array with the Custom Post Types to be translated in the Sitemap 
		// or leave it as a blank array to use all the public available types    
		// ---------------------------------------------------------------------------- //
		$allowed_types = array();
		// ---------------------------------------------------------------------------- //

		if ( empty( $allowed_types ) ) {
			$allowed_types = get_post_types(
				array(
					'show_ui' => true, 
					'public'  => true,
				)
			);
			unset( $allowed_types['post'] );
			unset( $allowed_types['page'] );
			unset( $allowed_types['attachment'] );
			$allowed_types = array_keys( $allowed_types );
		}

		if ( empty( $allowed_types ) || ( ! empty( $allowed_types ) && ! in_array( $type, $allowed_types ) ) ) {
			return $items; // Type not allowed.
		}

		if ( empty( $items ) ) {
			return $items; // Nothing to handle.
		}
		
		$grouped_items = array();

		foreach( $items as $item ) {
			if ( ! isset( $grouped_items[ $item->get_location() ] ) ) {
				$grouped_items[ $item->get_location() ] = array(
					'id'       => url_to_postid( $item->get_location() ),
					'variants' => array(),
					'item'     => $item,
				);
			} else {
				$grouped_items[ $item->get_location() ]['variants'][ $item->get_last_modified() ] = $item;
			}
		}

		if ( empty( $items ) ) {
			return $items;
		}

		$p      = $wpdb->prefix;
		$ids_in = implode( ',', array_column( $grouped_items, 'id' ) );
		$query  = "SELECT 
			( 
				SELECT 
					element_id 
				FROM 
					{$p}icl_translations 
				WHERE 
					trid = t.trid
					AND element_type = 'post_$type' 
					AND source_language_code IS NULL LIMIT 1 
			) parent_id,
			p.id post_id,
			p.guid post_url,
			t.language_code post_lang,
			UNIX_TIMESTAMP( p.post_modified_gmt ) post_date
		FROM 
			{$p}icl_translations t
			LEFT JOIN {$p}posts p ON p.ID = t.element_id
		WHERE 
			trid IN ( 
				SELECT 
					trid 
				FROM 
					{$p}icl_translations
				WHERE 
					element_id IN ($ids_in) 
					AND element_type  = 'post_$type'
					AND p.post_status = 'publish'
			) 
			AND t.element_id NOT IN ($ids_in)";

		$posts = $wpdb->get_results( $query, ARRAY_A ); // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared	
		
		if ( empty( $posts ) ) {
			return $items; // Empty result set (probably due to a DB query error).
		}
		
		$result       = array();
		$replacements = array();

		foreach( $grouped_items as &$item ) {
			$item_url    = $item['item']->get_location();
			$item['item']->set_location( $item_url . '%MLSM_' . $item['id'] . '%' );
			$result[]    = $item['item'];
			$replacement = '</loc>';
			foreach( $posts as $post ) {
				if( $item['id'] === (int) $post['parent_id'] ) {
					//$post_date  = (int) $post['post_date'];
					//if ( isset( $item['variants'][ $post_date ] ) ) {
						//$new_url      = apply_filters( 'wpml_permalink', $post['post_url'], $post['post_lang'] ); 
						$new_url      = apply_filters( 'wpml_permalink', $item_url, $post['post_lang'] ); 
						$replacement .=  "\n" . '<xhtml:link rel="alternate" hreflang="' . $post['post_lang'] . '" href="' . $new_url . '" />';
					//}
				}
			}
			$replacements[ '%MLSM_' . $item['id'] . '%</loc>' ] = $replacement;
		}

		$xml = Smartcrawl_Simple_Renderer::load( 'sitemap/sitemap-general-xml', array( 'items' => $result, ) );
		$xml = str_ireplace( array_keys( $replacements ), array_values( $replacements ), $xml );

		if ( false === strpos( $xml, 'xmlns:xhtml=' ) ) {
			$xml = str_ireplace( '<urlset', '<urlset xmlns:xhtml="http://www.w3.org/1999/xhtml" ', $xml );
		}

		// Cache this XML output.
		$sitemap_cache = Smartcrawl_Sitemap_Cache::get();
		$sitemap_cache->set_cached( $type, $page, $xml );

		$is_gzip_supported = stripos( (string) smartcrawl_get_array_value( $_SERVER, 'HTTP_ACCEPT_ENCODING' ), 'gzip' ) !== false;
		$is_gzip_request   = ! empty( get_query_var( 'wds_sitemap_gzip' ) );
	
		if ( ! headers_sent() ) {
			status_header( 200 );
			// Prevent the search engines from indexing the XML Sitemap.
			header( 'X-Robots-Tag: noindex, follow' );
			header( 'Content-Type: text/xml; charset=UTF-8' );

			if (
				$is_gzip_supported
				&& function_exists( 'gzencode' )
				&& $is_gzip_request
			) {
				header( 'Content-Encoding: gzip' );
				$xml = gzencode( $xml );
			}
		
			die( $xml );
		}

		return $result;
	}, 10, 3 );
});