yoast-sitemaps-cli.php

<?php // phpcs:ignore
/**
 * Plugin Name: Yoast Sitemap CLI
 * Description: Define Yoast sitemap generation CLI commands
 * Author:      The Code Co
 * Version:     1.0
 *
 * @package yoast-sitemap-cli
 */

namespace TheCodeCompany;

/**
 * Yoast sitemap CLI plugin driver class.
 *
 * This adds the following WP CLI command to generate and cache the root sitemap.
 *
 * `wp yoast-sitemap-build-root`
 *
 * This will generate the root sitemap and store it as a vanilla XML file in the uploads directory. This will be
 * automatically served when you hit the root sitemap index defined by Yoast. I.e. https://yoursite.com/sitemap_index.xml.
 *
 * ## Set Up
 *
 * This command should be configured to run via a cron every 10 mins or so, like so:
 *
 * ```
 * *\/10 * * * * cd /srv/www/mysite/current && wp yoast-sitemap-build-root --url="https://yoursite.com" 2>&1 2>/dev/null
 * ```
 */
class YoastSitemapCLI {

	const CMD_BUILD_ROOT_SITEMAP = 'yoast-sitemap-build-root';

	/** The max number of posts per individual sitemap generated by Yoast. */
	const SITEMAP_POSTS_PER_PAGE = 1000;

	/** @var WPSEO_Sitemaps */
	protected $yoast_sitemap = null;

	/**
	 * Boot the plugin by registering CLI command, routes etc.
	 *
	 * @return void
	 */
	public function boot() {

		$this->register_cli_commands();

		$this->register_routes();

		$this->apply_sitemap_optimisations();

	}

	/**
	 * Register the plugins WP CLI command/s.
	 *
	 * @return void
	 */
	public function register_cli_commands() {

		if ( class_exists( 'WP_CLI' ) ) {

			\WP_CLI::add_command(
				self::CMD_BUILD_ROOT_SITEMAP,
				array( $this, 'cmd_build_root_sitemap' ),
				array(
					'shortdesc' => 'Generate Yoast root sitemap.',
					'longdesc'  => 'Generate Yoast root sitemap and save to cache file on disk.',
				)
			);

		}

	}

	/**
	 * Handle custom sitemap routeing.
	 *
	 * @return void
	 */
	public function register_routes() {

		\add_filter( 'do_parse_request', array( $this, 'handle_routing' ) );

	}

	/**
	 * Handle custom routing.
	 * We may not have access to the TCC custom routing system so we have to do this manually via
	 * this hook into `do_parse_request`.
	 *
	 * @return boolean
	 */
	public function handle_routing( $continue ) {

		// Get the request path / URI.
		$request_path = $_SERVER['REQUEST_URI'] ?? '';
		$request_path = trim( $request_path, '/' );  // Remove leading/trailing slashes.
		$request_path = strtok( $request_path, '?' ); // Remove query string args.

		// Root sitemap.
		if ( preg_match( '{^sitemap_index.xml$}', $request_path ) ) {
			$this->handle_route_sitemap_root();
		}

		// Sitemap XSL which doesn't work otherwise for some reason.
		if ( preg_match( '{^main-sitemap.xsl$}', $request_path ) ) {
			$this->handle_route_sitemap_xsl();
		}

		return $continue;

	}

	/**
	 * Apply some optimisations to the Yoast sitemap generation process.
	 *
	 * @return void
	 */
	public function apply_sitemap_optimisations() {

		\add_filter(
			'wpseo_sitemap_entries_per_page',
			function( $per_page ) {
				return self::SITEMAP_POSTS_PER_PAGE;
			}
		);

	}

	/**
	 * WP CLI callback to run the root sitemap generation command.
	 *
	 * @return void
	 */
	public function cmd_build_root_sitemap() {

		$this->log_info(
			sprintf(
				'Building root sitemap to  %s',
				$this->get_sitemap_root_cache_file_path()
			)
		);

		$this->build_sitemap_root_cache_file();
		$this->log_info(
			sprintf(
				'Saved root sitemap to  %s',
				$this->get_sitemap_root_cache_file_path()
			)
		);

		$this->ping_search_engines();
		$this->log_info(
			sprintf(
				'Pinged search engines with updated sitemap',
				$this->get_sitemap_root_cache_file_path()
			)
		);

		$this->log_success( 'Done.' );

	}

	/**
	 * Handle the root sitemap custom route.
	 *
	 * @return void
	 */
	protected function handle_route_sitemap_root() {

		$sitemap_root_cache_file_path = $this->get_sitemap_root_cache_file_path();
		if ( file_exists( $sitemap_root_cache_file_path ) ) {

			// Set mime type & caching headers for sitemaps file.
			// NOTE This is a modified version of what Yoast does.
			if ( ! headers_sent() ) {

				header( 'HTTP/1.1 200 OK', true, 200 );

				// Prevent the search engines from indexing the XML Sitemap.
				header( 'X-Robots-Tag: noindex, follow', true );
				header( 'X-Sitemap-Served: php', true );
				header( 'Content-Type: text/xml' );

			}

			// Output the cached sitemap file.
			readfile( $this->get_sitemap_root_cache_file_path() );

		} else {

			// Cache file not present, 404 Not Found.
			header( 'HTTP/1.1 404 Not Found', true, 404 );
			echo '404 Not Found';

		}

		die;

	}

	/**
	 * Handle the sitemap custom XML file route.
	 * This replaces Yoasts built in route which breaks for some reason when we rewrite the root sitemap.
	 *
	 * @return void
	 */
	protected function handle_route_sitemap_xsl() {

		$yoast_sitemap = $this->get_yoast_sitemap_instance();
		$yoast_sitemap->xsl_output( 'main' );

		die;

	}

	/**
	 * Build the root sitemap and save the XML to the cache file on disk.
	 *
	 * @return void
	 */
	protected function build_sitemap_root_cache_file() {

		$sitemap_content = $this->build_sitemap_root();

		\file_put_contents( // phpcs:ignore
			$this->get_sitemap_root_cache_file_path(),
			$sitemap_content
		);

	}

	/**
	 * Build the root sitemap and return the raw XML output.
	 *
	 * @return string
	 */
	protected function build_sitemap_root() {

		$sitemap_content = '';

		$yoast_sitemap = $this->get_yoast_sitemap_instance();

		// Generate the root sitemap in Yoast and capture the output XML.
		ob_start();
			$yoast_sitemap->init_sitemaps_providers();
			$yoast_sitemap->build_root_map();
			$yoast_sitemap->output();
		$sitemap_content = ob_get_clean();

		$sitemap_content = \apply_filters( 'yoast_sitemap_cli_root_content', $sitemap_content );

		return $sitemap_content;

	}

	/**
	 * Ping search engines to let them know the sitemap has been updated.
	 * This is done via Yoast, we just call their function.
	 *
	 * @return void
	 */
	protected function ping_search_engines() {

		\WPSEO_Sitemaps::ping_search_engines();

	}

	/**
	 * Get the path of the root sitemap cache file.
	 *
	 * @return string
	 */
	protected function get_sitemap_root_cache_file_path() {

		$upload_dir = wp_upload_dir();

		// Set directory the sitemap will be stored in.
		$cache_dir = '.';
		if ( isset( $upload_dir['basedir'] ) ) {
			$cache_dir = $upload_dir['basedir'];
		}

		// Generate sitemap file path.
		$cache_file = sprintf(
			'%s/sitemap_index.xml',
			$cache_dir
		);

		$cache_file = \apply_filters( 'yoast_sitemap_cli_root_cache_file_path', $cache_file );

		return $cache_file;

	}

	/**
	 * Returns the `WPSEO_Sitemaps` instance which should be used by this plugin.
	 * This is kinda like a singleton instance.
	 *
	 * @return \WPSEO_Sitemaps
	 */
	protected function get_yoast_sitemap_instance() {

		if ( null === $this->yoast_sitemap ) {
			$this->yoast_sitemap = new \WPSEO_Sitemaps();
		}

		return $this->yoast_sitemap;

	}

	/**
	 * Print an info level message to the console.
	 *
	 * @param string $message Message to print.
	 * @return void
	 */
	protected function log_info( string $message ) {

		if ( class_exists( 'WP_CLI' ) ) {
			\WP_CLI::line( $message );
		}

	}

	/**
	 * Print a success level message to the console.
	 *
	 * @param string $message Message to print.
	 * @return void
	 */
	protected function log_success( string $message ) {

		if ( class_exists( 'WP_CLI' ) ) {
			\WP_CLI::success( $message );
		}

	}

	/**
	 * Print an error level message to the console.
	 *
	 * @param string $message Message to print.
	 * @return void
	 */
	protected function log_error( string $message ) {

		if ( class_exists( 'WP_CLI' ) ) {
			\WP_CLI::error( $message );
		}

	}

}

// Boot plugin driver class.
$yoast_sitemap_cli = new YoastSitemapCLI();
$yoast_sitemap_cli->boot();