[SmartCrawl] Separated the sitemaps by languages.

smartcrawl-separated-the-sitemap-by-languages.php

<?php
/**
 * Plugin Name: [SmartCrawl] Separated the sitemaps by languages.
 * Description: [SmartCrawl] Separated the sitemaps by languages - supported TranslatePress - Multilingual plugin
 * Task: SLS-323
 * Author: Thobk @ WPMUDEV
 * Author URI: https://premium.wpmudev.org
 * License: GPLv2 or later
 */
if ( ! defined( 'ABSPATH' ) ) {
	exit;
} elseif ( defined( 'WP_CLI' ) && WP_CLI ) {
	return;
}

/**
 * This MU will separated the sitemap by languages:
 * url: example.com/fr/sitemap.xml
 * path: wp-content/uploads/sc-languages/fr/smartcrawl/sitemap/index-sitemap0.xml
 */
add_action( 'after_setup_theme', 'wpmudev_sc_separated_the_sitemap_by_languages', 100 );

function wpmudev_sc_separated_the_sitemap_by_languages() {
	if ( defined( 'SMARTCRAWL_VERSION' ) && class_exists('Smartcrawl_Settings') ) {
		class WPMUDEV_SC_Separated_Sitemap_By_Languages{
			public $sub_folder = 'sc-languages';//wp-content/uploads/[sub-folder]/...			
			// Enter list of language codes in your site.
			public $all_lang_codes = array(
				// 'en',
				// 'fr'
			);

			private $current_language;
			public function __construct(){
				// check enabled sitemap
				if ( ! Smartcrawl_Sitemap_UI::get()->should_run() ) {
					return;
				}

				add_action( 'init', array( $this, 'add_rewrites' ) );

				// add_action( 'wds_before_sitemap_rebuild', array( $this, 'before_build_sitemap' ) );

				add_action( 'wp', array( $this, 'serve_sitemap' ), 998 );
				add_action( 'wp', array( $this, 'disable_custom_filter' ), 1000 );

				// maybe re-generate sitemaps.
				add_action( 'admin_init', array( $this, 'prime_cache_on_sitemap_settings_page_load' ), 10 );
			}

			public function add_rewrites(){
				/**
				 * @var $wp \WP
				 */
				global $wp;
				$wp->add_query_var( 'lang' );

				// add_rewrite_rule( '([a-z]{2})/sitemap\.xml(\.gz)?$', 'index.php?wds_sitemap=1&wds_sitemap_type=index&lang=$matches[1]&wds_sitemap_gzip=$matches[2]', 'top' );
				// add_rewrite_rule( '([a-z]{2})/([^/]+?)-sitemap([0-9]+)?\.xml(\.gz)?$', 'index.php?wds_sitemap=1&lang=$matches[1]&wds_sitemap_type=$matches[2]&wds_sitemap_page=$matches[3]&wds_sitemap_gzip=$matches[4]', 'top' );
				add_rewrite_rule( 'sitemap-([a-z]{2})\.xml(\.gz)?$', 'index.php?wds_sitemap=1&wds_sitemap_type=index&lang=$matches[1]&wds_sitemap_gzip=$matches[2]', 'top' );
				add_rewrite_rule( '([a-z]{2})/([^/]+?)-sitemap([0-9]+)?\.xml(\.gz)?$', 'index.php?wds_sitemap=1&lang=$matches[1]&wds_sitemap_type=$matches[2]&wds_sitemap_page=$matches[3]&wds_sitemap_gzip=$matches[4]', 'top' );

				// flush rewrite rules
				$this->maybe_flush_rewrite_rules();
			}

			protected function maybe_flush_rewrite_rules() {
				$flushed = get_option( 'wpmudev-sc-flush-rewrite-rules', false );
				if ( $flushed !== SMARTCRAWL_VERSION ) {
					flush_rewrite_rules();
					update_option( 'wpmudev-sc-flush-rewrite-rules', SMARTCRAWL_VERSION );
				}
			}

			// public function before_build_sitemap(){

			// }

			// ['en','vi'].
			public function get_all_language_codes( $exclude_default = false ){
				static $all_lang_codes;
				if( ! isset( $all_lang_codes ) ){
					$all_lang_codes = array();

					// support for TranslatePress - Multilingual plugin.
					if( class_exists('TRP_Translate_Press') ){
						$trp_settings = TRP_Translate_Press::get_trp_instance()->get_component( 'settings' );
		        $settings = $trp_settings->get_settings();
						$all_lang_codes = $settings['url-slugs'];

						// exclude default language.
						if( $exclude_default ){
							$default_language = $settings["default-language"];
							if( isset( $all_lang_codes[ $default_language ] ) ){
								unset( $all_lang_codes[ $default_language ] );
							}
						}
					} elseif ( $this->all_lang_codes ) {
						$all_lang_codes = $ths->all_lang_codes;
					}
				}

				return $all_lang_codes;
			}

			public function exist_language( $lang ){
				if( $lang ){
					// exclude default language.
					$all_lang_codes = $this->get_all_language_codes( true );

					if( ! in_array( $lang, $all_lang_codes ) ){
						$lang = false;
					}
				}
				return $lang;
			}

			public function should_custom_home_url(){
				static $should_change;
				if( ! isset( $should_change ) ){
					$should_change = false;
					// support for TranslatePress - Multilingual plugin.
					if( class_exists('TRP_Translate_Press') ){
						$should_change = true;
					}else{
						$should_change = ! empty( $this->all_lang_codes );
					}
				}
				return $should_change;
			}

			public function serve_sitemap(){
				// maybe clear sitemap cache
				if( ! Smartcrawl_Sitemap_Cache::get()->is_cache_pristine() ){
					$this->drop_cache();
				}
				if( ! empty( get_query_var( 'wds_sitemap' ) ) && ( $this->current_language = $this->exist_language( get_query_var( 'lang' ) ) ) ){
					add_filter( 'upload_dir', array( $this, 'custom_upload_dir_path_for_sitemap' ) );
					if( $this->should_custom_home_url() ){
						add_filter( 'home_url', array( $this, 'custom_home_url_by_lang'), 101, 2 );
					}
				}
			}

			public function get_sub_folder_sitemap(){
				return "/{$this->sub_folder}/";
			}

			public function custom_upload_dir_path_for_sitemap( $uploads ){
				if( $this->current_language ){
					$uploads['basedir'] .= $this->get_sub_folder_sitemap() . $this->current_language;
				}
				return $uploads;
			}

			public function custom_home_url_by_lang( $url, $path ){
				if( $this->current_language ){
					$home_lang_url = site_url( '/'. $this->current_language .'/' );
					if( false === strpos( $url, $home_lang_url ) ){
						$url = str_replace( site_url( '/' ), $home_lang_url, $url );
					}
				}

				return $url;
			}

			public function disable_custom_filter(){
				if( $this->current_language ){
					remove_filter( 'upload_dir', array( $this, 'custom_upload_dir_path_for_sitemap' ) );
					if( $this->should_custom_home_url() ){
						remove_filter( 'home_url', array( $this, 'custom_home_url_by_lang'), 101, 2 );
					}
				}
			}

			public function drop_cache() {
				$file_system = $this->fs_direct();
				$uploads = wp_upload_dir( null, false );
				$cache_dir = $uploads['basedir'] . $this->get_sub_folder_sitemap();
				if ( ! @file_exists( $cache_dir ) ) {
					Smartcrawl_Logger::error( "Sitemap cache by languages could not be dropped because they do not exist" );
					return false;
				}

				$removed = $file_system->rmdir( $cache_dir, true );
				if ( ! $removed ) {
					Smartcrawl_Logger::error( "Sitemap cache by languages directory could not be removed" );
					return false;
				}

				Smartcrawl_Logger::info( "Sitemap cache by languages dropped" );
				return true;
			}

			private function fs_direct() {
				if ( ! class_exists( 'WP_Filesystem_Direct', false ) ) {
					require_once ABSPATH . 'wp-admin/includes/class-wp-filesystem-base.php';
					require_once ABSPATH . 'wp-admin/includes/class-wp-filesystem-direct.php';
				}
				return new WP_Filesystem_Direct( null );
			}

			public function get_all_sitemap_urls(){
				$all_sitemap_urls = array();
				$all_lang_codes = $this->get_all_language_codes( true );
				if ( ! empty( $all_lang_codes ) ) {
					foreach ( $all_lang_codes as $lang ) {
						$all_sitemap_urls[] = home_url( $lang . '/sitemap.xml' );
					}
				}
				return $all_sitemap_urls;
			}

			public function prime_cache_on_sitemap_settings_page_load() {
				global $plugin_page;

				$is_sitemap_page = isset( $plugin_page ) && Smartcrawl_Settings::TAB_SITEMAP === $plugin_page;
				if ( ! $is_sitemap_page ) {
					return;
				}

				if ( Smartcrawl_Sitemap_Cache::get()->is_index_cached() ) {
					return;
				}

				$all_sitemap_urls = $this->get_all_sitemap_urls();
				if( $all_sitemap_urls ){
					foreach( $all_sitemap_urls as $sitemap_url ){
						wp_remote_get( $sitemap_url, array( 'blocking' => false ) );
					}
				}
			}
		}

		$run = new WPMUDEV_SC_Separated_Sitemap_By_Languages;
	}
}