sitemaps.php — Jetpack – WP Security, Backup, Speed, & Growth 14.9.2

jetpack / modules / sitemaps / sitemaps.php

jetpack / modules / sitemaps Last commit date

sitemaps.php

606 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-buffer-fallback.php';
38	require_once __DIR__ . '/sitemap-buffer-xmlwriter.php';
39	require_once __DIR__ . '/sitemap-buffer-page-xmlwriter.php';
40	require_once __DIR__ . '/sitemap-buffer-image-xmlwriter.php';
41	require_once __DIR__ . '/sitemap-buffer-video-xmlwriter.php';
42	require_once __DIR__ . '/sitemap-buffer-news-xmlwriter.php';
43	require_once __DIR__ . '/sitemap-buffer-master-xmlwriter.php';
44	require_once __DIR__ . '/sitemap-buffer-factory.php';
45	require_once __DIR__ . '/sitemap-stylist.php';
46	require_once __DIR__ . '/sitemap-librarian.php';
47	require_once __DIR__ . '/sitemap-finder.php';
48	require_once __DIR__ . '/sitemap-builder.php';
49
50	if ( defined( 'WP_DEBUG' ) && WP_DEBUG ) {
51	require_once __DIR__ . '/sitemap-logger.php';
52	}
53
54	// phpcs:disable Universal.Files.SeparateFunctionsFromOO.Mixed -- TODO: Move classes to appropriately-named class files.
55
56	/**
57	* Governs the generation, storage, and serving of sitemaps.
58	*
59	* @since 4.8.0
60	*
61	* @phan-constructor-used-for-side-effects
62	*/
63	class Jetpack_Sitemap_Manager {
64
65	/**
66	* Librarian object for storing and retrieving sitemap data.
67	*
68	* @see Jetpack_Sitemap_Librarian
69	* @since 4.8.0
70	* @var Jetpack_Sitemap_Librarian $librarian Librarian object for storing and retrieving sitemap data.
71	*/
72	private $librarian;
73
74	/**
75	* Logger object for reporting debug messages.
76	*
77	* @see Jetpack_Sitemap_Logger
78	* @since 4.8.0
79	* @var Jetpack_Sitemap_Logger $logger Logger object for reporting debug messages.
80	*/
81	private $logger;
82
83	/**
84	* Finder object for handling sitemap URIs.
85	*
86	* @see Jetpack_Sitemap_Finder
87	* @since 4.8.0
88	* @var Jetpack_Sitemap_Finder $finder Finder object for handling with sitemap URIs.
89	*/
90	private $finder;
91
92	/**
93	* Construct a new Jetpack_Sitemap_Manager.
94	*
95	* @access public
96	* @since 4.8.0
97	*/
98	public function __construct() {
99	$this->librarian = new Jetpack_Sitemap_Librarian();
100	$this->finder = new Jetpack_Sitemap_Finder();
101
102	if ( defined( 'WP_DEBUG' ) && ( true === WP_DEBUG ) ) {
103	$this->logger = new Jetpack_Sitemap_Logger();
104	}
105
106	// Add callback for sitemap URL handler.
107	add_action(
108	'wp_loaded',
109	array( $this, 'callback_action_catch_sitemap_urls' ),
110	defined( 'IS_WPCOM' ) && IS_WPCOM ? 100 : 10
111	);
112
113	// Add generator to wp_cron task list.
114	$this->schedule_sitemap_generation();
115
116	// Add sitemap to robots.txt.
117	add_action(
118	'do_robotstxt',
119	array( $this, 'callback_action_do_robotstxt' ),
120	20
121	);
122
123	// The news sitemap is cached; here we add a callback to
124	// flush the cached news sitemap when a post is published.
125	add_action(
126	'publish_post',
127	array( $this, 'callback_action_flush_news_sitemap_cache' ),
128	10
129	);
130
131	// In case we need to purge all sitemaps, we do this.
132	add_action(
133	'jetpack_sitemaps_purge_data',
134	array( $this, 'callback_action_purge_data' )
135	);
136
137	/*
138	* Module parameters are stored as options in the database.
139	* This allows us to avoid having to process all of init
140	* before serving the sitemap data. The following actions
141	* process and store these filters.
142	*/
143
144	// Process filters and store location string for sitemap.
145	add_action(
146	'init',
147	array( $this, 'callback_action_filter_sitemap_location' ),
148	999
149	);
150	}
151
152	/**
153	* Echo a raw string of given content-type.
154	*
155	* @access private
156	* @since 4.8.0
157	*
158	* @param string $the_content_type The content type to be served.
159	* @param string $the_content The string to be echoed.
160	* @return never
161	*/
162	private function serve_raw_and_die( $the_content_type, $the_content ) {
163	header( 'Content-Type: ' . $the_content_type . '; charset=UTF-8' );
164
165	global $wp_query;
166	$wp_query->is_feed = true;
167	set_query_var( 'feed', 'sitemap' );
168
169	if ( '' === $the_content ) {
170	$error = __( 'No sitemap found. Please try again later.', 'jetpack' );
171	if ( current_user_can( 'manage_options' ) ) {
172	$next = human_time_diff( wp_next_scheduled( 'jp_sitemap_cron_hook' ) );
173	/* translators: %s is a human_time_diff until next sitemap generation. */
174	$error = sprintf( __( 'No sitemap found. The system will try to build it again in %s.', 'jetpack' ), $next );
175	}
176
177	wp_die(
178	esc_html( $error ),
179	esc_html__( 'Sitemaps', 'jetpack' ),
180	array(
181	'response' => 404,
182	)
183	);
184	}
185
186	echo $the_content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- All content created by Jetpack.
187
188	die( 0 );
189	}
190
191	/**
192	* Callback to intercept sitemap url requests and serve sitemap files.
193	*
194	* @access public
195	* @since 4.8.0
196	*/
197	public function callback_action_catch_sitemap_urls() {
198	// Regular expressions for sitemap URL routing.
199	$regex = array(
200	'sitemap' => '/^sitemap-[1-9][0-9]*\.xml$/',
201	'index' => '/^sitemap-index-[1-9][0-9]*\.xml$/',
202	'image' => '/^image-sitemap-[1-9][0-9]*\.xml$/',
203	'image-index' => '/^image-sitemap-index-[1-9][0-9]*\.xml$/',
204	'video' => '/^video-sitemap-[1-9][0-9]*\.xml$/',
205	'video-index' => '/^video-sitemap-index-[1-9][0-9]*\.xml$/',
206	);
207
208	// The raw path(+query) of the requested URI.
209	if ( isset( $_SERVER['REQUEST_URI'] ) ) { // WPCS: Input var okay.
210	$raw_uri = sanitize_text_field(
211	wp_unslash( $_SERVER['REQUEST_URI'] ) // WPCS: Input var okay.
212	);
213	} else {
214	$raw_uri = '';
215	}
216
217	$request = $this->finder->recognize_sitemap_uri( $raw_uri );
218
219	if ( isset( $request['sitemap_name'] ) ) {
220
221	/**
222	* Filter the content type used to serve the sitemap XML files.
223	*
224	* @module sitemaps
225	*
226	* @since 3.9.0
227	*
228	* @param string $xml_content_type By default, it's 'text/xml'.
229	*/
230	$xml_content_type = apply_filters( 'jetpack_sitemap_content_type', 'text/xml' );
231
232	// Catch master sitemap xml.
233	if ( 'sitemap.xml' === $request['sitemap_name'] ) {
234	$sitemap_content = $this->librarian->get_sitemap_text(
235	jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
236	JP_MASTER_SITEMAP_TYPE
237	);
238
239	// if there is no master sitemap yet, let's just return an empty sitemap with a short TTL instead of a 404.
240	if ( empty( $sitemap_content ) ) {
241	$builder = new Jetpack_Sitemap_Builder();
242	$sitemap_content = $builder->empty_sitemap_xml();
243	}
244
245	$this->serve_raw_and_die(
246	$xml_content_type,
247	$sitemap_content
248	);
249	}
250
251	// Catch sitemap xsl.
252	if ( 'sitemap.xsl' === $request['sitemap_name'] ) {
253	$this->serve_raw_and_die(
254	'application/xml',
255	Jetpack_Sitemap_Stylist::sitemap_xsl()
256	);
257	}
258
259	// Catch sitemap index xsl.
260	if ( 'sitemap-index.xsl' === $request['sitemap_name'] ) {
261	$this->serve_raw_and_die(
262	'application/xml',
263	Jetpack_Sitemap_Stylist::sitemap_index_xsl()
264	);
265	}
266
267	// Catch image sitemap xsl.
268	if ( 'image-sitemap.xsl' === $request['sitemap_name'] ) {
269	$this->serve_raw_and_die(
270	'application/xml',
271	Jetpack_Sitemap_Stylist::image_sitemap_xsl()
272	);
273	}
274
275	// Catch video sitemap xsl.
276	if ( 'video-sitemap.xsl' === $request['sitemap_name'] ) {
277	$this->serve_raw_and_die(
278	'application/xml',
279	Jetpack_Sitemap_Stylist::video_sitemap_xsl()
280	);
281	}
282
283	// Catch news sitemap xml.
284	if ( 'news-sitemap.xml' === $request['sitemap_name'] ) {
285	$sitemap_builder = new Jetpack_Sitemap_Builder();
286	$this->serve_raw_and_die(
287	$xml_content_type,
288	$sitemap_builder->news_sitemap_xml()
289	);
290	}
291
292	// Catch news sitemap xsl.
293	if ( 'news-sitemap.xsl' === $request['sitemap_name'] ) {
294	$this->serve_raw_and_die(
295	'application/xml',
296	Jetpack_Sitemap_Stylist::news_sitemap_xsl()
297	);
298	}
299
300	// Catch sitemap xml.
301	if ( preg_match( $regex['sitemap'], $request['sitemap_name'] ) ) {
302	$this->serve_raw_and_die(
303	$xml_content_type,
304	$this->librarian->get_sitemap_text(
305	$request['sitemap_name'],
306	JP_PAGE_SITEMAP_TYPE
307	)
308	);
309	}
310
311	// Catch sitemap index xml.
312	if ( preg_match( $regex['index'], $request['sitemap_name'] ) ) {
313	$this->serve_raw_and_die(
314	$xml_content_type,
315	$this->librarian->get_sitemap_text(
316	$request['sitemap_name'],
317	JP_PAGE_SITEMAP_INDEX_TYPE
318	)
319	);
320	}
321
322	// Catch image sitemap xml.
323	if ( preg_match( $regex['image'], $request['sitemap_name'] ) ) {
324	$this->serve_raw_and_die(
325	$xml_content_type,
326	$this->librarian->get_sitemap_text(
327	$request['sitemap_name'],
328	JP_IMAGE_SITEMAP_TYPE
329	)
330	);
331	}
332
333	// Catch image sitemap index xml.
334	if ( preg_match( $regex['image-index'], $request['sitemap_name'] ) ) {
335	$this->serve_raw_and_die(
336	$xml_content_type,
337	$this->librarian->get_sitemap_text(
338	$request['sitemap_name'],
339	JP_IMAGE_SITEMAP_INDEX_TYPE
340	)
341	);
342	}
343
344	// Catch video sitemap xml.
345	if ( preg_match( $regex['video'], $request['sitemap_name'] ) ) {
346	$this->serve_raw_and_die(
347	$xml_content_type,
348	$this->librarian->get_sitemap_text(
349	$request['sitemap_name'],
350	JP_VIDEO_SITEMAP_TYPE
351	)
352	);
353	}
354
355	// Catch video sitemap index xml.
356	if ( preg_match( $regex['video-index'], $request['sitemap_name'] ) ) {
357	$this->serve_raw_and_die(
358	$xml_content_type,
359	$this->librarian->get_sitemap_text(
360	$request['sitemap_name'],
361	JP_VIDEO_SITEMAP_INDEX_TYPE
362	)
363	);
364	}
365	}
366	}
367
368	/**
369	* Callback for adding sitemap-interval to the list of schedules.
370	*
371	* @access public
372	* @since 4.8.0
373	*
374	* @param array $schedules The array of WP_Cron schedules.
375	*
376	* @return array The updated array of WP_Cron schedules.
377	*/
378	public function callback_add_sitemap_schedule( $schedules ) {
379	$schedules['sitemap-interval'] = array(
380	'interval' => JP_SITEMAP_INTERVAL,
381	'display' => __( 'Sitemap Interval', 'jetpack' ),
382	);
383	return $schedules;
384	}
385
386	/**
387	* Callback handler for sitemap cron hook
388	*
389	* @access public
390	*/
391	public function callback_sitemap_cron_hook() {
392	$sitemap_builder = new Jetpack_Sitemap_Builder();
393	$sitemap_builder->update_sitemap();
394	}
395
396	/**
397	* Add actions to schedule sitemap generation.
398	* Should only be called once, in the constructor.
399	*
400	* @access private
401	* @since 4.8.0
402	*/
403	private function schedule_sitemap_generation() {
404	// Add cron schedule.
405	add_filter( 'cron_schedules', array( $this, 'callback_add_sitemap_schedule' ) ); // phpcs:ignore WordPress.WP.CronInterval.ChangeDetected
406
407	add_action(
408	'jp_sitemap_cron_hook',
409	array( $this, 'callback_sitemap_cron_hook' )
410	);
411
412	if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
413	/**
414	* Filter the delay in seconds until sitemap generation cron job is started.
415	*
416	* This filter allows a site operator or hosting provider to potentialy spread out sitemap generation for a
417	* lot of sites over time. By default, it will be randomly done over 15 minutes.
418	*
419	* @module sitemaps
420	* @since 6.6.1
421	*
422	* @param int $delay Time to delay in seconds.
423	*/
424	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
425	wp_schedule_event(
426	time() + $delay,
427	'sitemap-interval',
428	'jp_sitemap_cron_hook'
429	);
430	}
431	}
432
433	/**
434	* Callback to add sitemap to robots.txt.
435	*
436	* @access public
437	* @since 4.8.0
438	*/
439	public function callback_action_do_robotstxt() {
440
441	/**
442	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
443	*
444	* @module sitemaps
445	* @since 3.9.0
446	* @deprecated 7.4.0
447	*
448	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
449	*/
450	$discover_sitemap = apply_filters_deprecated( 'jetpack_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_sitemap_include_in_robotstxt' );
451
452	/**
453	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
454	*
455	* @module sitemaps
456	* @since 7.4.0
457	*
458	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
459	*/
460	$discover_sitemap = apply_filters( 'jetpack_sitemap_include_in_robotstxt', $discover_sitemap );
461
462	if ( true === $discover_sitemap ) {
463	$sitemap_url = $this->finder->construct_sitemap_url( 'sitemap.xml' );
464	echo 'Sitemap: ' . esc_url( $sitemap_url ) . "\n";
465	}
466
467	/**
468	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
469	*
470	* @module sitemaps
471	* @since 3.9.0
472	* @deprecated 7.4.0
473	*
474	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
475	*/
476	$discover_news_sitemap = apply_filters_deprecated( 'jetpack_news_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_news_sitemap_include_in_robotstxt' );
477
478	/**
479	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
480	*
481	* @module sitemaps
482	* @since 7.4.0
483	*
484	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
485	*/
486	$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_include_in_robotstxt', $discover_news_sitemap );
487
488	if ( true === $discover_news_sitemap ) {
489	$news_sitemap_url = $this->finder->construct_sitemap_url( 'news-sitemap.xml' );
490	echo 'Sitemap: ' . esc_url( $news_sitemap_url ) . "\n";
491	}
492	}
493
494	/**
495	* Callback to delete the news sitemap cache.
496	*
497	* @access public
498	* @since 4.8.0
499	*/
500	public function callback_action_flush_news_sitemap_cache() {
501	delete_transient( 'jetpack_news_sitemap_xml' );
502	}
503
504	/**
505	* Callback for resetting stored sitemap data.
506	*
507	* @access public
508	* @since 5.3.0
509	* @since 6.7.0 Schedules a regeneration.
510	*/
511	public function callback_action_purge_data() {
512	$this->callback_action_flush_news_sitemap_cache();
513	$this->librarian->delete_all_stored_sitemap_data();
514	/** This filter is documented in modules/sitemaps/sitemaps.php */
515	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
516	wp_schedule_single_event( time() + $delay, 'jp_sitemap_cron_hook' );
517	}
518
519	/**
520	* Callback to set the sitemap location.
521	*
522	* @access public
523	* @since 4.8.0
524	*/
525	public function callback_action_filter_sitemap_location() {
526	update_option(
527	'jetpack_sitemap_location',
528	/**
529	* Additional path for sitemap URIs. Default value is empty.
530	*
531	* This string is any additional path fragment you want included between
532	* the home URL and the sitemap filenames. Exactly how this fragment is
533	* interpreted depends on your permalink settings. For example:
534	*
535	* Pretty permalinks:
536	* home_url() . jetpack_sitemap_location . '/sitemap.xml'
537	*
538	* Plain ("ugly") permalinks:
539	* home_url() . jetpack_sitemap_location . '/?jetpack-sitemap=sitemap.xml'
540	*
541	* PATHINFO permalinks:
542	* home_url() . '/index.php' . jetpack_sitemap_location . '/sitemap.xml'
543	*
544	* where 'sitemap.xml' is the name of a specific sitemap file.
545	* The value of this filter must be a valid path fragment per RFC 3986;
546	* in particular it must either be empty or begin with a '/'.
547	* Also take care that any restrictions on sitemap location imposed by
548	* the sitemap protocol are satisfied.
549	*
550	* The result of this filter is stored in an option, 'jetpack_sitemap_location';
551	* that option is what gets read when the sitemap location is needed.
552	* This way we don't have to wait for init to finish before building sitemaps.
553	*
554	* @link https://tools.ietf.org/html/rfc3986#section-3.3 RFC 3986
555	* @link https://www.sitemaps.org/ The sitemap protocol
556	*
557	* @since 4.8.0
558	*/
559	apply_filters(
560	'jetpack_sitemap_location',
561	''
562	)
563	);
564	}
565	} // End Jetpack_Sitemap_Manager class.
566
567	new Jetpack_Sitemap_Manager();
568
569	/**
570	* Absolute URL of the current blog's sitemap.
571	*
572	* @module sitemaps
573	*
574	* @since 3.9.0
575	* @since 4.8.1 Code uses method found in Jetpack_Sitemap_Finder::construct_sitemap_url in 4.8.0.
576	* It has been moved here to avoid fatal errors with other plugins that were expecting to find this function.
577	*
578	* @param string $filename Sitemap file name. Defaults to 'sitemap.xml', the initial sitemaps page.
579	*
580	* @return string Sitemap URL.
581	*/
582	function jetpack_sitemap_uri( $filename = 'sitemap.xml' ) {
583	global $wp_rewrite;
584
585	$location = Jetpack_Options::get_option_and_ensure_autoload( 'jetpack_sitemap_location', '' );
586
587	if ( $wp_rewrite->using_index_permalinks() ) {
588	$sitemap_url = home_url( '/index.php' . $location . '/' . $filename );
589	} elseif ( $wp_rewrite->using_permalinks() ) {
590	$sitemap_url = home_url( $location . '/' . $filename );
591	} else {
592	$sitemap_url = home_url( $location . '/?jetpack-sitemap=' . $filename );
593	}
594
595	/**
596	* Filter sitemap URL relative to home URL.
597	*
598	* @module sitemaps
599	*
600	* @since 3.9.0
601	*
602	* @param string $sitemap_url Sitemap URL.
603	*/
604	return apply_filters( 'jetpack_sitemap_location', $sitemap_url );
605	}
606