sitemaps.php — Jetpack – WP Security, Backup, Speed, & Growth 13.3.3

jetpack / modules / sitemaps / sitemaps.php

jetpack / modules / sitemaps Last commit date

sitemaps.php

603 lines

1	<?php // phpcs:ignore WordPress.Files.FileName.InvalidClassFileName
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 https://www.sitemaps.org/protocol.html 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 https://support.google.com/webmasters/answer/74288 News sitemap extension.
27	*
28	* @package automattic/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 __DIR__ . '/sitemap-constants.php';
36	require_once __DIR__ . '/sitemap-buffer.php';
37	require_once __DIR__ . '/sitemap-stylist.php';
38	require_once __DIR__ . '/sitemap-librarian.php';
39	require_once __DIR__ . '/sitemap-finder.php';
40	require_once __DIR__ . '/sitemap-builder.php';
41
42	if ( defined( 'WP_DEBUG' ) && WP_DEBUG ) {
43	require_once __DIR__ . '/sitemap-logger.php';
44	}
45
46	// phpcs:disable Universal.Files.SeparateFunctionsFromOO.Mixed -- TODO: Move classes to appropriately-named class files.
47
48	/**
49	* Governs the generation, storage, and serving of sitemaps.
50	*
51	* @since 4.8.0
52	*/
53	class Jetpack_Sitemap_Manager {
54
55	/**
56	* Librarian object for storing and retrieving sitemap data.
57	*
58	* @see Jetpack_Sitemap_Librarian
59	* @since 4.8.0
60	* @var Jetpack_Sitemap_Librarian $librarian Librarian object for storing and retrieving sitemap data.
61	*/
62	private $librarian;
63
64	/**
65	* Logger object for reporting debug messages.
66	*
67	* @see Jetpack_Sitemap_Logger
68	* @since 4.8.0
69	* @var Jetpack_Sitemap_Logger $logger Logger object for reporting debug messages.
70	*/
71	private $logger;
72
73	/**
74	* Finder object for handling sitemap URIs.
75	*
76	* @see Jetpack_Sitemap_Finder
77	* @since 4.8.0
78	* @var Jetpack_Sitemap_Finder $finder Finder object for handling with sitemap URIs.
79	*/
80	private $finder;
81
82	/**
83	* Construct a new Jetpack_Sitemap_Manager.
84	*
85	* @access public
86	* @since 4.8.0
87	*/
88	public function __construct() {
89	$this->librarian = new Jetpack_Sitemap_Librarian();
90	$this->finder = new Jetpack_Sitemap_Finder();
91
92	if ( defined( 'WP_DEBUG' ) && ( true === WP_DEBUG ) ) {
93	$this->logger = new Jetpack_Sitemap_Logger();
94	}
95
96	// Add callback for sitemap URL handler.
97	add_action(
98	'wp_loaded',
99	array( $this, 'callback_action_catch_sitemap_urls' ),
100	defined( 'IS_WPCOM' ) && IS_WPCOM ? 100 : 10
101	);
102
103	// Add generator to wp_cron task list.
104	$this->schedule_sitemap_generation();
105
106	// Add sitemap to robots.txt.
107	add_action(
108	'do_robotstxt',
109	array( $this, 'callback_action_do_robotstxt' ),
110	20
111	);
112
113	// The news sitemap is cached; here we add a callback to
114	// flush the cached news sitemap when a post is published.
115	add_action(
116	'publish_post',
117	array( $this, 'callback_action_flush_news_sitemap_cache' ),
118	10
119	);
120
121	// In case we need to purge all sitemaps, we do this.
122	add_action(
123	'jetpack_sitemaps_purge_data',
124	array( $this, 'callback_action_purge_data' )
125	);
126
127	/*
128	* Module parameters are stored as options in the database.
129	* This allows us to avoid having to process all of init
130	* before serving the sitemap data. The following actions
131	* process and store these filters.
132	*/
133
134	// Process filters and store location string for sitemap.
135	add_action(
136	'init',
137	array( $this, 'callback_action_filter_sitemap_location' ),
138	999
139	);
140	}
141
142	/**
143	* Echo a raw string of given content-type.
144	*
145	* @access private
146	* @since 4.8.0
147	*
148	* @param string $the_content_type The content type to be served.
149	* @param string $the_content The string to be echoed.
150	* @return never
151	*/
152	private function serve_raw_and_die( $the_content_type, $the_content ) {
153	header( 'Content-Type: ' . $the_content_type . '; charset=UTF-8' );
154
155	global $wp_query;
156	$wp_query->is_feed = true;
157	set_query_var( 'feed', 'sitemap' );
158
159	if ( '' === $the_content ) {
160	$error = __( 'No sitemap found. Please try again later.', 'jetpack' );
161	if ( current_user_can( 'manage_options' ) ) {
162	$next = human_time_diff( wp_next_scheduled( 'jp_sitemap_cron_hook' ) );
163	/* translators: %s is a human_time_diff until next sitemap generation. */
164	$error = sprintf( __( 'No sitemap found. The system will try to build it again in %s.', 'jetpack' ), $next );
165	}
166
167	wp_die(
168	esc_html( $error ),
169	esc_html__( 'Sitemaps', 'jetpack' ),
170	array(
171	'response' => 404,
172	)
173	);
174	}
175
176	echo $the_content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- All content created by Jetpack.
177
178	die();
179	}
180
181	/**
182	* Callback to intercept sitemap url requests and serve sitemap files.
183	*
184	* @access public
185	* @since 4.8.0
186	*/
187	public function callback_action_catch_sitemap_urls() {
188	// Regular expressions for sitemap URL routing.
189	$regex = array(
190	'master' => '/^sitemap\.xml$/',
191	'sitemap' => '/^sitemap-[1-9][0-9]*\.xml$/',
192	'index' => '/^sitemap-index-[1-9][0-9]*\.xml$/',
193	'sitemap-style' => '/^sitemap\.xsl$/',
194	'index-style' => '/^sitemap-index\.xsl$/',
195	'image' => '/^image-sitemap-[1-9][0-9]*\.xml$/',
196	'image-index' => '/^image-sitemap-index-[1-9][0-9]*\.xml$/',
197	'image-style' => '/^image-sitemap\.xsl$/',
198	'video' => '/^video-sitemap-[1-9][0-9]*\.xml$/',
199	'video-index' => '/^video-sitemap-index-[1-9][0-9]*\.xml$/',
200	'video-style' => '/^video-sitemap\.xsl$/',
201	'news' => '/^news-sitemap\.xml$/',
202	'news-style' => '/^news-sitemap\.xsl$/',
203	);
204
205	// The raw path(+query) of the requested URI.
206	if ( isset( $_SERVER['REQUEST_URI'] ) ) { // WPCS: Input var okay.
207	$raw_uri = sanitize_text_field(
208	wp_unslash( $_SERVER['REQUEST_URI'] ) // WPCS: Input var okay.
209	);
210	} else {
211	$raw_uri = '';
212	}
213
214	$request = $this->finder->recognize_sitemap_uri( $raw_uri );
215
216	if ( isset( $request['sitemap_name'] ) ) {
217
218	/**
219	* Filter the content type used to serve the sitemap XML files.
220	*
221	* @module sitemaps
222	*
223	* @since 3.9.0
224	*
225	* @param string $xml_content_type By default, it's 'text/xml'.
226	*/
227	$xml_content_type = apply_filters( 'jetpack_sitemap_content_type', 'text/xml' );
228
229	// Catch master sitemap xml.
230	if ( preg_match( $regex['master'], $request['sitemap_name'] ) ) {
231	$sitemap_content = $this->librarian->get_sitemap_text(
232	jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
233	JP_MASTER_SITEMAP_TYPE
234	);
235
236	// if there is no master sitemap yet, let's just return an empty sitemap with a short TTL instead of a 404.
237	if ( empty( $sitemap_content ) ) {
238	$builder = new Jetpack_Sitemap_Builder();
239	$sitemap_content = $builder->empty_sitemap_xml();
240	}
241
242	$this->serve_raw_and_die(
243	$xml_content_type,
244	$sitemap_content
245	);
246	}
247
248	// Catch sitemap xml.
249	if ( preg_match( $regex['sitemap'], $request['sitemap_name'] ) ) {
250	$this->serve_raw_and_die(
251	$xml_content_type,
252	$this->librarian->get_sitemap_text(
253	$request['sitemap_name'],
254	JP_PAGE_SITEMAP_TYPE
255	)
256	);
257	}
258
259	// Catch sitemap index xml.
260	if ( preg_match( $regex['index'], $request['sitemap_name'] ) ) {
261	$this->serve_raw_and_die(
262	$xml_content_type,
263	$this->librarian->get_sitemap_text(
264	$request['sitemap_name'],
265	JP_PAGE_SITEMAP_INDEX_TYPE
266	)
267	);
268	}
269
270	// Catch sitemap xsl.
271	if ( preg_match( $regex['sitemap-style'], $request['sitemap_name'] ) ) {
272	$this->serve_raw_and_die(
273	'application/xml',
274	Jetpack_Sitemap_Stylist::sitemap_xsl()
275	);
276	}
277
278	// Catch sitemap index xsl.
279	if ( preg_match( $regex['index-style'], $request['sitemap_name'] ) ) {
280	$this->serve_raw_and_die(
281	'application/xml',
282	Jetpack_Sitemap_Stylist::sitemap_index_xsl()
283	);
284	}
285
286	// Catch image sitemap xml.
287	if ( preg_match( $regex['image'], $request['sitemap_name'] ) ) {
288	$this->serve_raw_and_die(
289	$xml_content_type,
290	$this->librarian->get_sitemap_text(
291	$request['sitemap_name'],
292	JP_IMAGE_SITEMAP_TYPE
293	)
294	);
295	}
296
297	// Catch image sitemap index xml.
298	if ( preg_match( $regex['image-index'], $request['sitemap_name'] ) ) {
299	$this->serve_raw_and_die(
300	$xml_content_type,
301	$this->librarian->get_sitemap_text(
302	$request['sitemap_name'],
303	JP_IMAGE_SITEMAP_INDEX_TYPE
304	)
305	);
306	}
307
308	// Catch image sitemap xsl.
309	if ( preg_match( $regex['image-style'], $request['sitemap_name'] ) ) {
310	$this->serve_raw_and_die(
311	'application/xml',
312	Jetpack_Sitemap_Stylist::image_sitemap_xsl()
313	);
314	}
315
316	// Catch video sitemap xml.
317	if ( preg_match( $regex['video'], $request['sitemap_name'] ) ) {
318	$this->serve_raw_and_die(
319	$xml_content_type,
320	$this->librarian->get_sitemap_text(
321	$request['sitemap_name'],
322	JP_VIDEO_SITEMAP_TYPE
323	)
324	);
325	}
326
327	// Catch video sitemap index xml.
328	if ( preg_match( $regex['video-index'], $request['sitemap_name'] ) ) {
329	$this->serve_raw_and_die(
330	$xml_content_type,
331	$this->librarian->get_sitemap_text(
332	$request['sitemap_name'],
333	JP_VIDEO_SITEMAP_INDEX_TYPE
334	)
335	);
336	}
337
338	// Catch video sitemap xsl.
339	if ( preg_match( $regex['video-style'], $request['sitemap_name'] ) ) {
340	$this->serve_raw_and_die(
341	'application/xml',
342	Jetpack_Sitemap_Stylist::video_sitemap_xsl()
343	);
344	}
345
346	// Catch news sitemap xml.
347	if ( preg_match( $regex['news'], $request['sitemap_name'] ) ) {
348	$sitemap_builder = new Jetpack_Sitemap_Builder();
349	$this->serve_raw_and_die(
350	$xml_content_type,
351	$sitemap_builder->news_sitemap_xml()
352	);
353	}
354
355	// Catch news sitemap xsl.
356	if ( preg_match( $regex['news-style'], $request['sitemap_name'] ) ) {
357	$this->serve_raw_and_die(
358	'application/xml',
359	Jetpack_Sitemap_Stylist::news_sitemap_xsl()
360	);
361	}
362	}
363	}
364
365	/**
366	* Callback for adding sitemap-interval to the list of schedules.
367	*
368	* @access public
369	* @since 4.8.0
370	*
371	* @param array $schedules The array of WP_Cron schedules.
372	*
373	* @return array The updated array of WP_Cron schedules.
374	*/
375	public function callback_add_sitemap_schedule( $schedules ) {
376	$schedules['sitemap-interval'] = array(
377	'interval' => JP_SITEMAP_INTERVAL,
378	'display' => __( 'Sitemap Interval', 'jetpack' ),
379	);
380	return $schedules;
381	}
382
383	/**
384	* Callback handler for sitemap cron hook
385	*
386	* @access public
387	*/
388	public function callback_sitemap_cron_hook() {
389	$sitemap_builder = new Jetpack_Sitemap_Builder();
390	$sitemap_builder->update_sitemap();
391	}
392
393	/**
394	* Add actions to schedule sitemap generation.
395	* Should only be called once, in the constructor.
396	*
397	* @access private
398	* @since 4.8.0
399	*/
400	private function schedule_sitemap_generation() {
401	// Add cron schedule.
402	add_filter( 'cron_schedules', array( $this, 'callback_add_sitemap_schedule' ) ); // phpcs:ignore WordPress.WP.CronInterval.ChangeDetected
403
404	add_action(
405	'jp_sitemap_cron_hook',
406	array( $this, 'callback_sitemap_cron_hook' )
407	);
408
409	if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
410	/**
411	* Filter the delay in seconds until sitemap generation cron job is started.
412	*
413	* This filter allows a site operator or hosting provider to potentialy spread out sitemap generation for a
414	* lot of sites over time. By default, it will be randomly done over 15 minutes.
415	*
416	* @module sitemaps
417	* @since 6.6.1
418	*
419	* @param int $delay Time to delay in seconds.
420	*/
421	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
422	wp_schedule_event(
423	time() + $delay,
424	'sitemap-interval',
425	'jp_sitemap_cron_hook'
426	);
427	}
428	}
429
430	/**
431	* Callback to add sitemap to robots.txt.
432	*
433	* @access public
434	* @since 4.8.0
435	*/
436	public function callback_action_do_robotstxt() {
437
438	/**
439	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
440	*
441	* @module sitemaps
442	* @since 3.9.0
443	* @deprecated 7.4.0
444	*
445	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
446	*/
447	$discover_sitemap = apply_filters_deprecated( 'jetpack_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_sitemap_include_in_robotstxt' );
448
449	/**
450	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
451	*
452	* @module sitemaps
453	* @since 7.4.0
454	*
455	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
456	*/
457	$discover_sitemap = apply_filters( 'jetpack_sitemap_include_in_robotstxt', $discover_sitemap );
458
459	if ( true === $discover_sitemap ) {
460	$sitemap_url = $this->finder->construct_sitemap_url( 'sitemap.xml' );
461	echo 'Sitemap: ' . esc_url( $sitemap_url ) . "\n";
462	}
463
464	/**
465	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
466	*
467	* @module sitemaps
468	* @since 3.9.0
469	* @deprecated 7.4.0
470	*
471	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
472	*/
473	$discover_news_sitemap = apply_filters_deprecated( 'jetpack_news_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_news_sitemap_include_in_robotstxt' );
474
475	/**
476	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
477	*
478	* @module sitemaps
479	* @since 7.4.0
480	*
481	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
482	*/
483	$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_include_in_robotstxt', $discover_news_sitemap );
484
485	if ( true === $discover_news_sitemap ) {
486	$news_sitemap_url = $this->finder->construct_sitemap_url( 'news-sitemap.xml' );
487	echo 'Sitemap: ' . esc_url( $news_sitemap_url ) . "\n";
488	}
489	}
490
491	/**
492	* Callback to delete the news sitemap cache.
493	*
494	* @access public
495	* @since 4.8.0
496	*/
497	public function callback_action_flush_news_sitemap_cache() {
498	delete_transient( 'jetpack_news_sitemap_xml' );
499	}
500
501	/**
502	* Callback for resetting stored sitemap data.
503	*
504	* @access public
505	* @since 5.3.0
506	* @since 6.7.0 Schedules a regeneration.
507	*/
508	public function callback_action_purge_data() {
509	$this->callback_action_flush_news_sitemap_cache();
510	$this->librarian->delete_all_stored_sitemap_data();
511	/** This filter is documented in modules/sitemaps/sitemaps.php */
512	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
513	wp_schedule_single_event( time() + $delay, 'jp_sitemap_cron_hook' );
514	}
515
516	/**
517	* Callback to set the sitemap location.
518	*
519	* @access public
520	* @since 4.8.0
521	*/
522	public function callback_action_filter_sitemap_location() {
523	update_option(
524	'jetpack_sitemap_location',
525	/**
526	* Additional path for sitemap URIs. Default value is empty.
527	*
528	* This string is any additional path fragment you want included between
529	* the home URL and the sitemap filenames. Exactly how this fragment is
530	* interpreted depends on your permalink settings. For example:
531	*
532	* Pretty permalinks:
533	* home_url() . jetpack_sitemap_location . '/sitemap.xml'
534	*
535	* Plain ("ugly") permalinks:
536	* home_url() . jetpack_sitemap_location . '/?jetpack-sitemap=sitemap.xml'
537	*
538	* PATHINFO permalinks:
539	* home_url() . '/index.php' . jetpack_sitemap_location . '/sitemap.xml'
540	*
541	* where 'sitemap.xml' is the name of a specific sitemap file.
542	* The value of this filter must be a valid path fragment per RFC 3986;
543	* in particular it must either be empty or begin with a '/'.
544	* Also take care that any restrictions on sitemap location imposed by
545	* the sitemap protocol are satisfied.
546	*
547	* The result of this filter is stored in an option, 'jetpack_sitemap_location';
548	* that option is what gets read when the sitemap location is needed.
549	* This way we don't have to wait for init to finish before building sitemaps.
550	*
551	* @link https://tools.ietf.org/html/rfc3986#section-3.3 RFC 3986
552	* @link https://www.sitemaps.org/ The sitemap protocol
553	*
554	* @since 4.8.0
555	*/
556	apply_filters(
557	'jetpack_sitemap_location',
558	''
559	)
560	);
561	}
562	} // End Jetpack_Sitemap_Manager class.
563
564	new Jetpack_Sitemap_Manager();
565
566	/**
567	* Absolute URL of the current blog's sitemap.
568	*
569	* @module sitemaps
570	*
571	* @since 3.9.0
572	* @since 4.8.1 Code uses method found in Jetpack_Sitemap_Finder::construct_sitemap_url in 4.8.0.
573	* It has been moved here to avoid fatal errors with other plugins that were expecting to find this function.
574	*
575	* @param string $filename Sitemap file name. Defaults to 'sitemap.xml', the initial sitemaps page.
576	*
577	* @return string Sitemap URL.
578	*/
579	function jetpack_sitemap_uri( $filename = 'sitemap.xml' ) {
580	global $wp_rewrite;
581
582	$location = Jetpack_Options::get_option_and_ensure_autoload( 'jetpack_sitemap_location', '' );
583
584	if ( $wp_rewrite->using_index_permalinks() ) {
585	$sitemap_url = home_url( '/index.php' . $location . '/' . $filename );
586	} elseif ( $wp_rewrite->using_permalinks() ) {
587	$sitemap_url = home_url( $location . '/' . $filename );
588	} else {
589	$sitemap_url = home_url( $location . '/?jetpack-sitemap=' . $filename );
590	}
591
592	/**
593	* Filter sitemap URL relative to home URL.
594	*
595	* @module sitemaps
596	*
597	* @since 3.9.0
598	*
599	* @param string $sitemap_url Sitemap URL.
600	*/
601	return apply_filters( 'jetpack_sitemap_location', $sitemap_url );
602	}
603