sitemaps.php ➔ jetpack_sitemap_array_to_simplexml() - Code Metrics - Inspection of "Spacing" - Automattic/jetpack - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — add/double-encode-message ( 8b6530...2d4e84 )

unknown

created 2019-03-01 17:51 UTC

o_simplexml() D

↳ Parent: Jetpack_Sitemap_Manager::__construct()

Complexity

Conditions	9
Paths	13

Size

Total Lines	30
Code Lines	19

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	9
eloc	19
nc	13
nop	2
dl	0
loc	30
rs	4.909
c	0
b	0
f	0

<?php
/**
 * Generate sitemap files in base XML as well as some namespace extensions.
 *
 * This module generates two different base sitemaps.
 *
 * 1. sitemap.xml
 *    The basic sitemap is updated regularly by wp-cron. It is stored in the
 *    database and retrieved when requested. This sitemap aims to include canonical
 *    URLs for all published content and abide by the sitemap spec. This is the root
 *    of a tree of sitemap and sitemap index xml files, depending on the number of URLs.
 *
 *    By default the sitemap contains published posts of type 'post' and 'page', as
 *    well as the home url. To include other post types use the 'jetpack_sitemap_post_types'
 *    filter.
 *
 * @link http://sitemaps.org/protocol.php Base sitemaps protocol.
 * @link https://support.google.com/webmasters/answer/178636 Image sitemap extension.
 * @link https://developers.google.com/webmasters/videosearch/sitemaps Video sitemap extension.
 *
 * 2. news-sitemap.xml
 *    The news sitemap is generated on the fly when requested. It does not aim for
 *    completeness, instead including at most 1000 of the most recent published posts
 *    from the previous 2 days, per the news-sitemap spec.
 *
 * @link http://www.google.com/support/webmasters/bin/answer.py?answer=74288 News sitemap extension.
 *
 * @package Jetpack
 * @since 3.9.0
 * @since 4.8.0 Remove 1000 post limit.
 * @author Automattic
 */

/* Include all of the sitemap subclasses. */
require_once dirname( __FILE__ ) . '/sitemap-constants.php';
require_once dirname( __FILE__ ) . '/sitemap-buffer.php';
require_once dirname( __FILE__ ) . '/sitemap-stylist.php';
require_once dirname( __FILE__ ) . '/sitemap-librarian.php';
require_once dirname( __FILE__ ) . '/sitemap-finder.php';
require_once dirname( __FILE__ ) . '/sitemap-builder.php';

if ( defined( 'WP_DEBUG' ) && WP_DEBUG ) {
	require_once dirname( __FILE__ ) . '/sitemap-logger.php';
}

/**
 * Governs the generation, storage, and serving of sitemaps.
 *
 * @since 4.8.0
 */
class Jetpack_Sitemap_Manager {

	/**
	 * Librarian object for storing and retrieving sitemap data.
	 *
	 * @see Jetpack_Sitemap_Librarian
	 * @since 4.8.0
	 * @var Jetpack_Sitemap_Librarian $librarian Librarian object for storing and retrieving sitemap data.
	 */
	private $librarian;

	/**
	 * Logger object for reporting debug messages.
	 *
	 * @see Jetpack_Sitemap_Logger
	 * @since 4.8.0
	 * @var Jetpack_Sitemap_Logger $logger Logger object for reporting debug messages.
	 */
	private $logger;

	/**
	 * Finder object for handling sitemap URIs.
	 *
	 * @see Jetpack_Sitemap_Finder
	 * @since 4.8.0
	 * @var Jetpack_Sitemap_Finder $finder Finder object for handling with sitemap URIs.
	 */
	private $finder;

	/**
	 * Construct a new Jetpack_Sitemap_Manager.
	 *
	 * @access public
	 * @since 4.8.0
	 */
	public function __construct() {
		$this->librarian = new Jetpack_Sitemap_Librarian();
		$this->finder    = new Jetpack_Sitemap_Finder();

		if ( defined( 'WP_DEBUG' ) && ( true === WP_DEBUG ) ) {
			$this->logger = new Jetpack_Sitemap_Logger();
		}

		// Add callback for sitemap URL handler.
		add_action(
			'init',
			array( $this, 'callback_action_catch_sitemap_urls' ),
			defined( 'IS_WPCOM' ) && IS_WPCOM ? 100 : 10
		);

		// Add generator to wp_cron task list.
		$this->schedule_sitemap_generation();

		// Add sitemap to robots.txt.
		add_action(
			'do_robotstxt',
			array( $this, 'callback_action_do_robotstxt' ),
			20
		);

		// The news sitemap is cached; here we add a callback to
		// flush the cached news sitemap when a post is published.
		add_action(
			'publish_post',
			array( $this, 'callback_action_flush_news_sitemap_cache' ),
			10
		);

		// In case we need to purge all sitemaps, we do this.
		add_action(
			'jetpack_sitemaps_purge_data',
			array( $this, 'callback_action_purge_data' )
		);

		/*
		 * Module parameters are stored as options in the database.
		 * This allows us to avoid having to process all of init
		 * before serving the sitemap data. The following actions
		 * process and store these filters.
		 */

		// Process filters and store location string for sitemap.
		add_action(
			'init',
			array( $this, 'callback_action_filter_sitemap_location' ),
			999
		);
	}

	/**
	 * Echo a raw string of given content-type.
	 *
	 * @access private
	 * @since 4.8.0
	 *
	 * @param string $the_content_type The content type to be served.
	 * @param string $the_content The string to be echoed.
	 */
	private function serve_raw_and_die( $the_content_type, $the_content ) {
		header( 'Content-Type: ' . $the_content_type . '; charset=UTF-8' );

		global $wp_query;
		$wp_query->is_feed = true;
		set_query_var( 'feed', 'sitemap' );

		if ( '' === $the_content ) {
			$error = __( 'No sitemap found. Please try again later.', 'jetpack' );
			if ( current_user_can( 'manage_options' ) ) {
				$next = human_time_diff( wp_next_scheduled( 'jp_sitemap_cron_hook' ) );
				/* translators: %s is a human_time_diff until next sitemap generation. */
				$error = sprintf( __( 'No sitemap found. The system will try to build it again in %s.', 'jetpack' ), $next );
			}

			wp_die(
				esc_html( $error ),
				esc_html__( 'Sitemaps', 'jetpack' ),
				array(
					'response' => 404,
				)
			);
		}

		echo $the_content;

		die();
	}

	/**
	 * Callback to intercept sitemap url requests and serve sitemap files.
	 *
	 * @access public
	 * @since 4.8.0
	 */
	public function callback_action_catch_sitemap_urls() {
		// Regular expressions for sitemap URL routing.
		$regex = array(
			'master'        => '/^sitemap\.xml$/',
			'sitemap'       => '/^sitemap-[1-9][0-9]*\.xml$/',
			'index'         => '/^sitemap-index-[1-9][0-9]*\.xml$/',
			'sitemap-style' => '/^sitemap\.xsl$/',
			'index-style'   => '/^sitemap-index\.xsl$/',
			'image'         => '/^image-sitemap-[1-9][0-9]*\.xml$/',
			'image-index'   => '/^image-sitemap-index-[1-9][0-9]*\.xml$/',
			'image-style'   => '/^image-sitemap\.xsl$/',
			'video'         => '/^video-sitemap-[1-9][0-9]*\.xml$/',
			'video-index'   => '/^video-sitemap-index-[1-9][0-9]*\.xml$/',
			'video-style'   => '/^video-sitemap\.xsl$/',
			'news'          => '/^news-sitemap\.xml$/',
			'news-style'    => '/^news-sitemap\.xsl$/',
		);

		// The raw path(+query) of the requested URI.
		if ( isset( $_SERVER['REQUEST_URI'] ) ) { // WPCS: Input var okay.
			$raw_uri = sanitize_text_field(
				wp_unslash( $_SERVER['REQUEST_URI'] ) // WPCS: Input var okay.
			);
		} else {
			$raw_uri = '';
		}

		$request = $this->finder->recognize_sitemap_uri( $raw_uri );

		if ( isset( $request['sitemap_name'] ) ) {

			/**
			 * Filter the content type used to serve the sitemap XML files.
			 *
			 * @module sitemaps
			 *
			 * @since 3.9.0
			 *
			 * @param string $xml_content_type By default, it's 'text/xml'.
			 */
			$xml_content_type = apply_filters( 'jetpack_sitemap_content_type', 'text/xml' );

			// Catch master sitemap xml.
			if ( preg_match( $regex['master'], $request['sitemap_name'] ) ) {
				$sitemap_content = $this->librarian->get_sitemap_text(
					jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
					JP_MASTER_SITEMAP_TYPE
				);

				// if there is no master sitemap yet, let's just return an empty sitemap with a short TTL instead of a 404
				if ( empty( $sitemap_content ) ) {
					$builder = new Jetpack_Sitemap_Builder();
					$sitemap_content = $builder->empty_sitemap_xml();
				}

				$this->serve_raw_and_die(
					$xml_content_type,
					$sitemap_content
				);
			}

			// Catch sitemap xml.
			if ( preg_match( $regex['sitemap'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					$xml_content_type,
					$this->librarian->get_sitemap_text(
						$request['sitemap_name'],
						JP_PAGE_SITEMAP_TYPE
					)
				);
			}

			// Catch sitemap index xml.
			if ( preg_match( $regex['index'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					$xml_content_type,
					$this->librarian->get_sitemap_text(
						$request['sitemap_name'],
						JP_PAGE_SITEMAP_INDEX_TYPE
					)
				);
			}

			// Catch sitemap xsl.
			if ( preg_match( $regex['sitemap-style'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					'application/xml',
					Jetpack_Sitemap_Stylist::sitemap_xsl()
				);
			}

			// Catch sitemap index xsl.
			if ( preg_match( $regex['index-style'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					'application/xml',
					Jetpack_Sitemap_Stylist::sitemap_index_xsl()
				);
			}

			// Catch image sitemap xml.
			if ( preg_match( $regex['image'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					$xml_content_type,
					$this->librarian->get_sitemap_text(
						$request['sitemap_name'],
						JP_IMAGE_SITEMAP_TYPE
					)
				);
			}

			// Catch image sitemap index xml.
			if ( preg_match( $regex['image-index'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					$xml_content_type,
					$this->librarian->get_sitemap_text(
						$request['sitemap_name'],
						JP_IMAGE_SITEMAP_INDEX_TYPE
					)
				);
			}

			// Catch image sitemap xsl.
			if ( preg_match( $regex['image-style'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					'application/xml',
					Jetpack_Sitemap_Stylist::image_sitemap_xsl()
				);
			}

			// Catch video sitemap xml.
			if ( preg_match( $regex['video'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					$xml_content_type,
					$this->librarian->get_sitemap_text(
						$request['sitemap_name'],
						JP_VIDEO_SITEMAP_TYPE
					)
				);
			}

			// Catch video sitemap index xml.
			if ( preg_match( $regex['video-index'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					$xml_content_type,
					$this->librarian->get_sitemap_text(
						$request['sitemap_name'],
						JP_VIDEO_SITEMAP_INDEX_TYPE
					)
				);
			}

			// Catch video sitemap xsl.
			if ( preg_match( $regex['video-style'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					'application/xml',
					Jetpack_Sitemap_Stylist::video_sitemap_xsl()
				);
			}

			// Catch news sitemap xml.
			if ( preg_match( $regex['news'], $request['sitemap_name'] ) ) {
				$sitemap_builder = new Jetpack_Sitemap_Builder();
				$this->serve_raw_and_die(
					$xml_content_type,
					$sitemap_builder->news_sitemap_xml()
				);
			}

			// Catch news sitemap xsl.
			if ( preg_match( $regex['news-style'], $request['sitemap_name'] ) ) {
				$this->serve_raw_and_die(
					'application/xml',
					Jetpack_Sitemap_Stylist::news_sitemap_xsl()
				);
			}
		}
	}

	/**
	 * Callback for adding sitemap-interval to the list of schedules.
	 *
	 * @access public
	 * @since 4.8.0
	 *
	 * @param array $schedules The array of WP_Cron schedules.
	 *
	 * @return array The updated array of WP_Cron schedules.
	 */
	public function callback_add_sitemap_schedule( $schedules ) {
		$schedules['sitemap-interval'] = array(
			'interval' => JP_SITEMAP_INTERVAL,
			'display'  => __( 'Sitemap Interval', 'jetpack' ),
		);
		return $schedules;
	}

	/**
	 * Callback handler for sitemap cron hook
	 *
	 * @access public
	 */
	public function callback_sitemap_cron_hook() {
		$sitemap_builder = new Jetpack_Sitemap_Builder();
		$sitemap_builder->update_sitemap();
	}

	/**
	 * Add actions to schedule sitemap generation.
	 * Should only be called once, in the constructor.
	 *
	 * @access private
	 * @since 4.8.0
	 */
	private function schedule_sitemap_generation() {
		// Add cron schedule.
		add_filter( 'cron_schedules', array( $this, 'callback_add_sitemap_schedule' ) ); // phpcs:ignore WordPress.WP.CronInterval.ChangeDetected

		add_action(
			'jp_sitemap_cron_hook',
			array( $this, 'callback_sitemap_cron_hook' )
		);

		if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
			/**
			 * Filter the delay in seconds until sitemap generation cron job is started.
			 *
			 * This filter allows a site operator or hosting provider to potentialy spread out sitemap generation for a
			 * lot of sites over time. By default, it will be randomly done over 15 minutes.
			 *
			 * @module sitemaps
			 * @since 6.6.1
			 *
			 * @param int $delay Time to delay in seconds.
			 */
			$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
			wp_schedule_event(
				time() + $delay,
				'sitemap-interval',
				'jp_sitemap_cron_hook'
			);
		}
	}

	/**
	 * Callback to add sitemap to robots.txt.
	 *
	 * @access public
	 * @since 4.8.0
	 */
	public function callback_action_do_robotstxt() {

		/**
		 * Filter whether to make the default sitemap discoverable to robots or not. Default true.
		 *
		 * @module sitemaps
		 * @since 3.9.0
		 *
		 * @param bool $discover_sitemap Make default sitemap discoverable to robots.
		 */
		$discover_sitemap = apply_filters( 'jetpack_sitemap_generate', true );

		if ( true === $discover_sitemap ) {
			$sitemap_url = $this->finder->construct_sitemap_url( 'sitemap.xml' );
			echo 'Sitemap: ' . esc_url( $sitemap_url ) . "\n";
		}

		/**
		 * Filter whether to make the news sitemap discoverable to robots or not. Default true.
		 *
		 * @module sitemaps
		 * @since 3.9.0
		 *
		 * @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
		 */
		$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_generate', true );

		if ( true === $discover_news_sitemap ) {
			$news_sitemap_url = $this->finder->construct_sitemap_url( 'news-sitemap.xml' );
			echo 'Sitemap: ' . esc_url( $news_sitemap_url ) . "\n";
		}
	}

	/**
	 * Callback to delete the news sitemap cache.
	 *
	 * @access public
	 * @since 4.8.0
	 */
	public function callback_action_flush_news_sitemap_cache() {
		delete_transient( 'jetpack_news_sitemap_xml' );
	}

	/**
	 * Callback for resetting stored sitemap data.
	 *
	 * @access public
	 * @since 5.3.0
	 * @since 6.7.0 Schedules a regeneration.
	 */
	public function callback_action_purge_data() {
		$this->callback_action_flush_news_sitemap_cache();
		$this->librarian->delete_all_stored_sitemap_data();
		/** This filter is documented in modules/sitemaps/sitemaps.php */
		$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
		wp_schedule_single_event( time() + $delay, 'jp_sitemap_cron_hook' );
	}

	/**
	 * Callback to set the sitemap location.
	 *
	 * @access public
	 * @since 4.8.0
	 */
	public function callback_action_filter_sitemap_location() {
		update_option(
			'jetpack_sitemap_location',
			/**
			 * Additional path for sitemap URIs. Default value is empty.
			 *
			 * This string is any additional path fragment you want included between
			 * the home URL and the sitemap filenames. Exactly how this fragment is
			 * interpreted depends on your permalink settings. For example:
			 *
			 *   Pretty permalinks:
			 *     home_url() . jetpack_sitemap_location . '/sitemap.xml'
			 *
			 *   Plain ("ugly") permalinks:
			 *     home_url() . jetpack_sitemap_location . '/?jetpack-sitemap=sitemap.xml'
			 *
			 *   PATHINFO permalinks:
			 *     home_url() . '/index.php' . jetpack_sitemap_location . '/sitemap.xml'
			 *
			 * where 'sitemap.xml' is the name of a specific sitemap file.
			 * The value of this filter must be a valid path fragment per RFC 3986;
			 * in particular it must either be empty or begin with a '/'.
			 * Also take care that any restrictions on sitemap location imposed by
			 * the sitemap protocol are satisfied.
			 *
			 * The result of this filter is stored in an option, 'jetpack_sitemap_location';
			 * that option is what gets read when the sitemap location is needed.
			 * This way we don't have to wait for init to finish before building sitemaps.
			 *
			 * @link https://tools.ietf.org/html/rfc3986#section-3.3 RFC 3986
			 * @link http://www.sitemaps.org/ The sitemap protocol
			 *
			 * @since 4.8.0
			 */
			apply_filters(
				'jetpack_sitemap_location',
				''
			)
		);
	}

} // End Jetpack_Sitemap_Manager class.

new Jetpack_Sitemap_Manager();

/**
 * Absolute URL of the current blog's sitemap.
 *
 * @module sitemaps
 *
 * @since  3.9.0
 * @since  4.8.1 Code uses method found in Jetpack_Sitemap_Finder::construct_sitemap_url in 4.8.0.
 *                It has been moved here to avoid fatal errors with other plugins that were expecting to find this function.
 *
 * @param string $filename Sitemap file name. Defaults to 'sitemap.xml', the initial sitemaps page.
 *
 * @return string Sitemap URL.
 */
function jetpack_sitemap_uri( $filename = 'sitemap.xml' ) {
	global $wp_rewrite;

	$location = Jetpack_Options::get_option_and_ensure_autoload( 'jetpack_sitemap_location', '' );

	if ( $wp_rewrite->using_index_permalinks() ) {
		$sitemap_url = home_url( '/index.php' . $location . '/' . $filename );
	} elseif ( $wp_rewrite->using_permalinks() ) {
		$sitemap_url = home_url( $location . '/' . $filename );
	} else {
		$sitemap_url = home_url( $location . '/?jetpack-sitemap=' . $filename );
	}

	/**
	 * Filter sitemap URL relative to home URL.
	 *
	 * @module sitemaps
	 *
	 * @since 3.9.0
	 *
	 * @param string $sitemap_url Sitemap URL.
	 */
	return apply_filters( 'jetpack_sitemap_location', $sitemap_url );
}


1		<?php
2		/**
3		* Generate sitemap files in base XML as well as some namespace extensions.
4		*
5		* This module generates two different base sitemaps.
6		*
7		* 1. sitemap.xml
8		* The basic sitemap is updated regularly by wp-cron. It is stored in the
9		* database and retrieved when requested. This sitemap aims to include canonical
10		* URLs for all published content and abide by the sitemap spec. This is the root
11		* of a tree of sitemap and sitemap index xml files, depending on the number of URLs.
12		*
13		* By default the sitemap contains published posts of type 'post' and 'page', as
14		* well as the home url. To include other post types use the 'jetpack_sitemap_post_types'
15		* filter.
16		*
17		* @link http://sitemaps.org/protocol.php Base sitemaps protocol.
18		* @link https://support.google.com/webmasters/answer/178636 Image sitemap extension.
19		* @link https://developers.google.com/webmasters/videosearch/sitemaps Video sitemap extension.
20		*
21		* 2. news-sitemap.xml
22		* The news sitemap is generated on the fly when requested. It does not aim for
23		* completeness, instead including at most 1000 of the most recent published posts
24		* from the previous 2 days, per the news-sitemap spec.
25		*
26		* @link http://www.google.com/support/webmasters/bin/answer.py?answer=74288 News sitemap extension.
27		*
28		* @package Jetpack
29		* @since 3.9.0
30		* @since 4.8.0 Remove 1000 post limit.
31		* @author Automattic
32		*/
33
34		/* Include all of the sitemap subclasses. */
35		require_once dirname( __FILE__ ) . '/sitemap-constants.php';
36		require_once dirname( __FILE__ ) . '/sitemap-buffer.php';
37		require_once dirname( __FILE__ ) . '/sitemap-stylist.php';
38		require_once dirname( __FILE__ ) . '/sitemap-librarian.php';
39		require_once dirname( __FILE__ ) . '/sitemap-finder.php';
40		require_once dirname( __FILE__ ) . '/sitemap-builder.php';
41
42		if ( defined( 'WP_DEBUG' ) && WP_DEBUG ) {
43		require_once dirname( __FILE__ ) . '/sitemap-logger.php';
44		}
45
46		/**
47		* Governs the generation, storage, and serving of sitemaps.
48		*
49		* @since 4.8.0
50		*/
51		class Jetpack_Sitemap_Manager {
52
53		/**
54		* Librarian object for storing and retrieving sitemap data.
55		*
56		* @see Jetpack_Sitemap_Librarian
57		* @since 4.8.0
58		* @var Jetpack_Sitemap_Librarian $librarian Librarian object for storing and retrieving sitemap data.
59		*/
60		private $librarian;
61
62		/**
63		* Logger object for reporting debug messages.
64		*
65		* @see Jetpack_Sitemap_Logger
66		* @since 4.8.0
67		* @var Jetpack_Sitemap_Logger $logger Logger object for reporting debug messages.
68		*/
69		private $logger;
70
71		/**
72		* Finder object for handling sitemap URIs.
73		*
74		* @see Jetpack_Sitemap_Finder
75		* @since 4.8.0
76		* @var Jetpack_Sitemap_Finder $finder Finder object for handling with sitemap URIs.
77		*/
78		private $finder;
79
80		/**
81		* Construct a new Jetpack_Sitemap_Manager.
82		*
83		* @access public
84		* @since 4.8.0
85		*/
86		public function __construct() {
87		$this->librarian = new Jetpack_Sitemap_Librarian();
88		$this->finder = new Jetpack_Sitemap_Finder();
89
90		if ( defined( 'WP_DEBUG' ) && ( true === WP_DEBUG ) ) {
91		$this->logger = new Jetpack_Sitemap_Logger();
92		}
93
94		// Add callback for sitemap URL handler.
95		add_action(
96		'init',
97		array( $this, 'callback_action_catch_sitemap_urls' ),
98		defined( 'IS_WPCOM' ) && IS_WPCOM ? 100 : 10
99		);
100
101		// Add generator to wp_cron task list.
102		$this->schedule_sitemap_generation();
103
104		// Add sitemap to robots.txt.
105		add_action(
106		'do_robotstxt',
107		array( $this, 'callback_action_do_robotstxt' ),
108		20
109		);
110
111		// The news sitemap is cached; here we add a callback to
112		// flush the cached news sitemap when a post is published.
113		add_action(
114		'publish_post',
115		array( $this, 'callback_action_flush_news_sitemap_cache' ),
116		10
117		);
118
119		// In case we need to purge all sitemaps, we do this.
120		add_action(
121		'jetpack_sitemaps_purge_data',
122		array( $this, 'callback_action_purge_data' )
123		);
124
125		/*
126		* Module parameters are stored as options in the database.
127		* This allows us to avoid having to process all of init
128		* before serving the sitemap data. The following actions
129		* process and store these filters.
130		*/
131
132		// Process filters and store location string for sitemap.
133		add_action(
134		'init',
135		array( $this, 'callback_action_filter_sitemap_location' ),
136		999
137		);
138		}
139
140		/**
141		* Echo a raw string of given content-type.
142		*
143		* @access private
144		* @since 4.8.0
145		*
146		* @param string $the_content_type The content type to be served.
147		* @param string $the_content The string to be echoed.
148		*/
149		private function serve_raw_and_die( $the_content_type, $the_content ) {
150		header( 'Content-Type: ' . $the_content_type . '; charset=UTF-8' );
151
152		global $wp_query;
153		$wp_query->is_feed = true;
154		set_query_var( 'feed', 'sitemap' );
155
156		if ( '' === $the_content ) {
157		$error = __( 'No sitemap found. Please try again later.', 'jetpack' );
158	View Code Duplication	if ( current_user_can( 'manage_options' ) ) {
159		$next = human_time_diff( wp_next_scheduled( 'jp_sitemap_cron_hook' ) );
160		/* translators: %s is a human_time_diff until next sitemap generation. */
161		$error = sprintf( __( 'No sitemap found. The system will try to build it again in %s.', 'jetpack' ), $next );
162		}
163
164		wp_die(
165		esc_html( $error ),
166		esc_html__( 'Sitemaps', 'jetpack' ),
167		array(
168		'response' => 404,
169		)
170		);
171		}
172
173		echo $the_content;
174
175		die();
176		}
177
178		/**
179		* Callback to intercept sitemap url requests and serve sitemap files.
180		*
181		* @access public
182		* @since 4.8.0
183		*/
184		public function callback_action_catch_sitemap_urls() {
185		// Regular expressions for sitemap URL routing.
186		$regex = array(
187		'master' => '/^sitemap\.xml$/',
188		'sitemap' => '/^sitemap-[1-9][0-9]*\.xml$/',
189		'index' => '/^sitemap-index-[1-9][0-9]*\.xml$/',
190		'sitemap-style' => '/^sitemap\.xsl$/',
191		'index-style' => '/^sitemap-index\.xsl$/',
192		'image' => '/^image-sitemap-[1-9][0-9]*\.xml$/',
193		'image-index' => '/^image-sitemap-index-[1-9][0-9]*\.xml$/',
194		'image-style' => '/^image-sitemap\.xsl$/',
195		'video' => '/^video-sitemap-[1-9][0-9]*\.xml$/',
196		'video-index' => '/^video-sitemap-index-[1-9][0-9]*\.xml$/',
197		'video-style' => '/^video-sitemap\.xsl$/',
198		'news' => '/^news-sitemap\.xml$/',
199		'news-style' => '/^news-sitemap\.xsl$/',
200		);
201
202		// The raw path(+query) of the requested URI.
203		if ( isset( $_SERVER['REQUEST_URI'] ) ) { // WPCS: Input var okay.
204		$raw_uri = sanitize_text_field(
205		wp_unslash( $_SERVER['REQUEST_URI'] ) // WPCS: Input var okay.
206		);
207		} else {
208		$raw_uri = '';
209		}
210
211		$request = $this->finder->recognize_sitemap_uri( $raw_uri );
212
213		if ( isset( $request['sitemap_name'] ) ) {
214
215		/**
216		* Filter the content type used to serve the sitemap XML files.
217		*
218		* @module sitemaps
219		*
220		* @since 3.9.0
221		*
222		* @param string $xml_content_type By default, it's 'text/xml'.
223		*/
224		$xml_content_type = apply_filters( 'jetpack_sitemap_content_type', 'text/xml' );
225
226		// Catch master sitemap xml.
227		if ( preg_match( $regex['master'], $request['sitemap_name'] ) ) {
228		$sitemap_content = $this->librarian->get_sitemap_text(
229		jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
230		JP_MASTER_SITEMAP_TYPE
231		);
232
233		// if there is no master sitemap yet, let's just return an empty sitemap with a short TTL instead of a 404
234		if ( empty( $sitemap_content ) ) {
235		$builder = new Jetpack_Sitemap_Builder();
236		$sitemap_content = $builder->empty_sitemap_xml();
237		}
238
239		$this->serve_raw_and_die(
240		$xml_content_type,
241		$sitemap_content
242		);
243		}
244
245		// Catch sitemap xml.
246	View Code Duplication	if ( preg_match( $regex['sitemap'], $request['sitemap_name'] ) ) {
247		$this->serve_raw_and_die(
248		$xml_content_type,
249		$this->librarian->get_sitemap_text(
250		$request['sitemap_name'],
251		JP_PAGE_SITEMAP_TYPE
252		)
253		);
254		}
255
256		// Catch sitemap index xml.
257	View Code Duplication	if ( preg_match( $regex['index'], $request['sitemap_name'] ) ) {
258		$this->serve_raw_and_die(
259		$xml_content_type,
260		$this->librarian->get_sitemap_text(
261		$request['sitemap_name'],
262		JP_PAGE_SITEMAP_INDEX_TYPE
263		)
264		);
265		}
266
267		// Catch sitemap xsl.
268		if ( preg_match( $regex['sitemap-style'], $request['sitemap_name'] ) ) {
269		$this->serve_raw_and_die(
270		'application/xml',
271		Jetpack_Sitemap_Stylist::sitemap_xsl()
272		);
273		}
274
275		// Catch sitemap index xsl.
276		if ( preg_match( $regex['index-style'], $request['sitemap_name'] ) ) {
277		$this->serve_raw_and_die(
278		'application/xml',
279		Jetpack_Sitemap_Stylist::sitemap_index_xsl()
280		);
281		}
282
283		// Catch image sitemap xml.
284	View Code Duplication	if ( preg_match( $regex['image'], $request['sitemap_name'] ) ) {
285		$this->serve_raw_and_die(
286		$xml_content_type,
287		$this->librarian->get_sitemap_text(
288		$request['sitemap_name'],
289		JP_IMAGE_SITEMAP_TYPE
290		)
291		);
292		}
293
294		// Catch image sitemap index xml.
295	View Code Duplication	if ( preg_match( $regex['image-index'], $request['sitemap_name'] ) ) {
296		$this->serve_raw_and_die(
297		$xml_content_type,
298		$this->librarian->get_sitemap_text(
299		$request['sitemap_name'],
300		JP_IMAGE_SITEMAP_INDEX_TYPE
301		)
302		);
303		}
304
305		// Catch image sitemap xsl.
306		if ( preg_match( $regex['image-style'], $request['sitemap_name'] ) ) {
307		$this->serve_raw_and_die(
308		'application/xml',
309		Jetpack_Sitemap_Stylist::image_sitemap_xsl()
310		);
311		}
312
313		// Catch video sitemap xml.
314	View Code Duplication	if ( preg_match( $regex['video'], $request['sitemap_name'] ) ) {
315		$this->serve_raw_and_die(
316		$xml_content_type,
317		$this->librarian->get_sitemap_text(
318		$request['sitemap_name'],
319		JP_VIDEO_SITEMAP_TYPE
320		)
321		);
322		}
323
324		// Catch video sitemap index xml.
325	View Code Duplication	if ( preg_match( $regex['video-index'], $request['sitemap_name'] ) ) {
326		$this->serve_raw_and_die(
327		$xml_content_type,
328		$this->librarian->get_sitemap_text(
329		$request['sitemap_name'],
330		JP_VIDEO_SITEMAP_INDEX_TYPE
331		)
332		);
333		}
334
335		// Catch video sitemap xsl.
336		if ( preg_match( $regex['video-style'], $request['sitemap_name'] ) ) {
337		$this->serve_raw_and_die(
338		'application/xml',
339		Jetpack_Sitemap_Stylist::video_sitemap_xsl()
340		);
341		}
342
343		// Catch news sitemap xml.
344		if ( preg_match( $regex['news'], $request['sitemap_name'] ) ) {
345		$sitemap_builder = new Jetpack_Sitemap_Builder();
346		$this->serve_raw_and_die(
347		$xml_content_type,
348		$sitemap_builder->news_sitemap_xml()
349		);
350		}
351
352		// Catch news sitemap xsl.
353		if ( preg_match( $regex['news-style'], $request['sitemap_name'] ) ) {
354		$this->serve_raw_and_die(
355		'application/xml',
356		Jetpack_Sitemap_Stylist::news_sitemap_xsl()
357		);
358		}
359		}
360		}
361
362		/**
363		* Callback for adding sitemap-interval to the list of schedules.
364		*
365		* @access public
366		* @since 4.8.0
367		*
368		* @param array $schedules The array of WP_Cron schedules.
369		*
370		* @return array The updated array of WP_Cron schedules.
371		*/
372		public function callback_add_sitemap_schedule( $schedules ) {
373		$schedules['sitemap-interval'] = array(
374		'interval' => JP_SITEMAP_INTERVAL,
375		'display' => __( 'Sitemap Interval', 'jetpack' ),
376		);
377		return $schedules;
378		}
379
380		/**
381		* Callback handler for sitemap cron hook
382		*
383		* @access public
384		*/
385		public function callback_sitemap_cron_hook() {
386		$sitemap_builder = new Jetpack_Sitemap_Builder();
387		$sitemap_builder->update_sitemap();
388		}
389
390		/**
391		* Add actions to schedule sitemap generation.
392		* Should only be called once, in the constructor.
393		*
394		* @access private
395		* @since 4.8.0
396		*/
397		private function schedule_sitemap_generation() {
398		// Add cron schedule.
399		add_filter( 'cron_schedules', array( $this, 'callback_add_sitemap_schedule' ) ); // phpcs:ignore WordPress.WP.CronInterval.ChangeDetected
400
401		add_action(
402		'jp_sitemap_cron_hook',
403		array( $this, 'callback_sitemap_cron_hook' )
404		);
405
406		if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
407		/**
408		* Filter the delay in seconds until sitemap generation cron job is started.
409		*
410		* This filter allows a site operator or hosting provider to potentialy spread out sitemap generation for a
411		* lot of sites over time. By default, it will be randomly done over 15 minutes.
412		*
413		* @module sitemaps
414		* @since 6.6.1
415		*
416		* @param int $delay Time to delay in seconds.
417		*/
418		$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
419		wp_schedule_event(
420		time() + $delay,
421		'sitemap-interval',
422		'jp_sitemap_cron_hook'
423		);
424		}
425		}
426
427		/**
428		* Callback to add sitemap to robots.txt.
429		*
430		* @access public
431		* @since 4.8.0
432		*/
433		public function callback_action_do_robotstxt() {
434
435		/**
436		* Filter whether to make the default sitemap discoverable to robots or not. Default true.
437		*
438		* @module sitemaps
439		* @since 3.9.0
440		*
441		* @param bool $discover_sitemap Make default sitemap discoverable to robots.
442		*/
443		$discover_sitemap = apply_filters( 'jetpack_sitemap_generate', true );
444
445	View Code Duplication	if ( true === $discover_sitemap ) {
446		$sitemap_url = $this->finder->construct_sitemap_url( 'sitemap.xml' );
447		echo 'Sitemap: ' . esc_url( $sitemap_url ) . "\n";
448		}
449
450		/**
451		* Filter whether to make the news sitemap discoverable to robots or not. Default true.
452		*
453		* @module sitemaps
454		* @since 3.9.0
455		*
456		* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
457		*/
458		$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_generate', true );
459
460	View Code Duplication	if ( true === $discover_news_sitemap ) {
461		$news_sitemap_url = $this->finder->construct_sitemap_url( 'news-sitemap.xml' );
462		echo 'Sitemap: ' . esc_url( $news_sitemap_url ) . "\n";
463		}
464		}
465
466		/**
467		* Callback to delete the news sitemap cache.
468		*
469		* @access public
470		* @since 4.8.0
471		*/
472		public function callback_action_flush_news_sitemap_cache() {
473		delete_transient( 'jetpack_news_sitemap_xml' );
474		}
475
476		/**
477		* Callback for resetting stored sitemap data.
478		*
479		* @access public
480		* @since 5.3.0
481		* @since 6.7.0 Schedules a regeneration.
482		*/
483		public function callback_action_purge_data() {
484		$this->callback_action_flush_news_sitemap_cache();
485		$this->librarian->delete_all_stored_sitemap_data();
486		/** This filter is documented in modules/sitemaps/sitemaps.php */
487		$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
488		wp_schedule_single_event( time() + $delay, 'jp_sitemap_cron_hook' );
489		}
490
491		/**
492		* Callback to set the sitemap location.
493		*
494		* @access public
495		* @since 4.8.0
496		*/
497		public function callback_action_filter_sitemap_location() {
498		update_option(
499		'jetpack_sitemap_location',
500		/**
501		* Additional path for sitemap URIs. Default value is empty.
502		*
503		* This string is any additional path fragment you want included between
504		* the home URL and the sitemap filenames. Exactly how this fragment is
505		* interpreted depends on your permalink settings. For example:
506		*
507		* Pretty permalinks:
508		* home_url() . jetpack_sitemap_location . '/sitemap.xml'
509		*
510		* Plain ("ugly") permalinks:
511		* home_url() . jetpack_sitemap_location . '/?jetpack-sitemap=sitemap.xml'
512		*
513		* PATHINFO permalinks:
514		* home_url() . '/index.php' . jetpack_sitemap_location . '/sitemap.xml'
515		*
516		* where 'sitemap.xml' is the name of a specific sitemap file.
517		* The value of this filter must be a valid path fragment per RFC 3986;
518		* in particular it must either be empty or begin with a '/'.
519		* Also take care that any restrictions on sitemap location imposed by
520		* the sitemap protocol are satisfied.
521		*
522		* The result of this filter is stored in an option, 'jetpack_sitemap_location';
523		* that option is what gets read when the sitemap location is needed.
524		* This way we don't have to wait for init to finish before building sitemaps.
525		*
526		* @link https://tools.ietf.org/html/rfc3986#section-3.3 RFC 3986
527		* @link http://www.sitemaps.org/ The sitemap protocol
528		*
529		* @since 4.8.0
530		*/
531		apply_filters(
532		'jetpack_sitemap_location',
533		''
534		)
535		);
536		}
537
538		} // End Jetpack_Sitemap_Manager class.
539
540		new Jetpack_Sitemap_Manager();
541
542		/**
543		* Absolute URL of the current blog's sitemap.
544		*
545		* @module sitemaps
546		*
547		* @since 3.9.0
548		* @since 4.8.1 Code uses method found in Jetpack_Sitemap_Finder::construct_sitemap_url in 4.8.0.
549		* It has been moved here to avoid fatal errors with other plugins that were expecting to find this function.
550		*
551		* @param string $filename Sitemap file name. Defaults to 'sitemap.xml', the initial sitemaps page.
552		*
553		* @return string Sitemap URL.
554		*/
555		function jetpack_sitemap_uri( $filename = 'sitemap.xml' ) {
556		global $wp_rewrite;
557
558		$location = Jetpack_Options::get_option_and_ensure_autoload( 'jetpack_sitemap_location', '' );
559
560		if ( $wp_rewrite->using_index_permalinks() ) {
561		$sitemap_url = home_url( '/index.php' . $location . '/' . $filename );
562		} elseif ( $wp_rewrite->using_permalinks() ) {
563		$sitemap_url = home_url( $location . '/' . $filename );
564		} else {
565		$sitemap_url = home_url( $location . '/?jetpack-sitemap=' . $filename );
566		}
567
568		/**
569		* Filter sitemap URL relative to home URL.
570		*
571		* @module sitemaps
572		*
573		* @since 3.9.0
574		*
575		* @param string $sitemap_url Sitemap URL.
576		*/
577		return apply_filters( 'jetpack_sitemap_location', $sitemap_url );
578		}
579

Automattic / jetpack

Push — add/double-encode-message ( 8b6530...2d4e84 )

o_simplexml() D

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like