Jetpack_Sitemap_Manager::callback_action_do_robotstxt() - Code Metrics - Inspection of "Activity Log: update remote url" - Automattic/jetpack - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — update/activity-log-links ( 92fe50...3b9362 )

unknown

created 2018-09-04 11:24 UTC

callback_action_do_robotstxt() A

↳ Parent: Jetpack_Sitemap_Manager

Complexity

Conditions	3
Paths	4

Size

Total Lines

Duplication

Lines	8
Ratio	23.53 %

Importance

Changes

Metric	Value
cc	3
nc	4
nop	0
dl	8
loc	34
rs	9.376
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
 */

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 {

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

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

	/**
	 * @see Jetpack_Sitemap_Finder
	 * @since 4.8.0
	 * @var Jetpack_Sitemap_Finder $finder Finder object for dealing 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
		);

		return;
	}

	/**
	 * 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 ) {
			wp_die(
				esc_html__( "No sitemap found. Maybe it's being generated. Please try again later.", 'jetpack' ),
				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'] ) ) {
				$this->serve_raw_and_die(
					$xml_content_type,
					$this->librarian->get_sitemap_text(
						jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
						JP_MASTER_SITEMAP_TYPE
					)
				);
			}

			// 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()
				);
			}
		}

		// URL did not match any sitemap patterns.
		return;
	}

	/**
	 * 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' ) );

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

		if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
			wp_schedule_event(
				time(),
				'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";
		}

		return;
	}

	/**
	 * 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
	 */
	public function callback_action_purge_data() {
		$this->callback_action_flush_news_sitemap_cache();
		$this->librarian->delete_all_stored_sitemap_data();
	}

	/**
	 * 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',
				''
			)
		);

		return;
	}

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

Automattic / jetpack

Push — update/activity-log-links ( 92fe50...3b9362 )

callback_action_do_robotstxt() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like