sitemaps.php — Jetpack – WP Security, Backup, Speed, & Growth 14.8.1

jetpack / modules / sitemaps / sitemaps.php

jetpack / modules / sitemaps Last commit date

sitemaps.php

611 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	class Jetpack_Sitemap_Manager {
62
63	/**
64	* Librarian object for storing and retrieving sitemap data.
65	*
66	* @see Jetpack_Sitemap_Librarian
67	* @since 4.8.0
68	* @var Jetpack_Sitemap_Librarian $librarian Librarian object for storing and retrieving sitemap data.
69	*/
70	private $librarian;
71
72	/**
73	* Logger object for reporting debug messages.
74	*
75	* @see Jetpack_Sitemap_Logger
76	* @since 4.8.0
77	* @var Jetpack_Sitemap_Logger $logger Logger object for reporting debug messages.
78	*/
79	private $logger;
80
81	/**
82	* Finder object for handling sitemap URIs.
83	*
84	* @see Jetpack_Sitemap_Finder
85	* @since 4.8.0
86	* @var Jetpack_Sitemap_Finder $finder Finder object for handling with sitemap URIs.
87	*/
88	private $finder;
89
90	/**
91	* Construct a new Jetpack_Sitemap_Manager.
92	*
93	* @access public
94	* @since 4.8.0
95	*/
96	public function __construct() {
97	$this->librarian = new Jetpack_Sitemap_Librarian();
98	$this->finder = new Jetpack_Sitemap_Finder();
99
100	if ( defined( 'WP_DEBUG' ) && ( true === WP_DEBUG ) ) {
101	$this->logger = new Jetpack_Sitemap_Logger();
102	}
103
104	// Add callback for sitemap URL handler.
105	add_action(
106	'wp_loaded',
107	array( $this, 'callback_action_catch_sitemap_urls' ),
108	defined( 'IS_WPCOM' ) && IS_WPCOM ? 100 : 10
109	);
110
111	// Add generator to wp_cron task list.
112	$this->schedule_sitemap_generation();
113
114	// Add sitemap to robots.txt.
115	add_action(
116	'do_robotstxt',
117	array( $this, 'callback_action_do_robotstxt' ),
118	20
119	);
120
121	// The news sitemap is cached; here we add a callback to
122	// flush the cached news sitemap when a post is published.
123	add_action(
124	'publish_post',
125	array( $this, 'callback_action_flush_news_sitemap_cache' ),
126	10
127	);
128
129	// In case we need to purge all sitemaps, we do this.
130	add_action(
131	'jetpack_sitemaps_purge_data',
132	array( $this, 'callback_action_purge_data' )
133	);
134
135	/*
136	* Module parameters are stored as options in the database.
137	* This allows us to avoid having to process all of init
138	* before serving the sitemap data. The following actions
139	* process and store these filters.
140	*/
141
142	// Process filters and store location string for sitemap.
143	add_action(
144	'init',
145	array( $this, 'callback_action_filter_sitemap_location' ),
146	999
147	);
148	}
149
150	/**
151	* Echo a raw string of given content-type.
152	*
153	* @access private
154	* @since 4.8.0
155	*
156	* @param string $the_content_type The content type to be served.
157	* @param string $the_content The string to be echoed.
158	* @return never
159	*/
160	private function serve_raw_and_die( $the_content_type, $the_content ) {
161	header( 'Content-Type: ' . $the_content_type . '; charset=UTF-8' );
162
163	global $wp_query;
164	$wp_query->is_feed = true;
165	set_query_var( 'feed', 'sitemap' );
166
167	if ( '' === $the_content ) {
168	$error = __( 'No sitemap found. Please try again later.', 'jetpack' );
169	if ( current_user_can( 'manage_options' ) ) {
170	$next = human_time_diff( wp_next_scheduled( 'jp_sitemap_cron_hook' ) );
171	/* translators: %s is a human_time_diff until next sitemap generation. */
172	$error = sprintf( __( 'No sitemap found. The system will try to build it again in %s.', 'jetpack' ), $next );
173	}
174
175	wp_die(
176	esc_html( $error ),
177	esc_html__( 'Sitemaps', 'jetpack' ),
178	array(
179	'response' => 404,
180	)
181	);
182	}
183
184	echo $the_content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- All content created by Jetpack.
185
186	die( 0 );
187	}
188
189	/**
190	* Callback to intercept sitemap url requests and serve sitemap files.
191	*
192	* @access public
193	* @since 4.8.0
194	*/
195	public function callback_action_catch_sitemap_urls() {
196	// Regular expressions for sitemap URL routing.
197	$regex = array(
198	'master' => '/^sitemap\.xml$/',
199	'sitemap' => '/^sitemap-[1-9][0-9]*\.xml$/',
200	'index' => '/^sitemap-index-[1-9][0-9]*\.xml$/',
201	'sitemap-style' => '/^sitemap\.xsl$/',
202	'index-style' => '/^sitemap-index\.xsl$/',
203	'image' => '/^image-sitemap-[1-9][0-9]*\.xml$/',
204	'image-index' => '/^image-sitemap-index-[1-9][0-9]*\.xml$/',
205	'image-style' => '/^image-sitemap\.xsl$/',
206	'video' => '/^video-sitemap-[1-9][0-9]*\.xml$/',
207	'video-index' => '/^video-sitemap-index-[1-9][0-9]*\.xml$/',
208	'video-style' => '/^video-sitemap\.xsl$/',
209	'news' => '/^news-sitemap\.xml$/',
210	'news-style' => '/^news-sitemap\.xsl$/',
211	);
212
213	// The raw path(+query) of the requested URI.
214	if ( isset( $_SERVER['REQUEST_URI'] ) ) { // WPCS: Input var okay.
215	$raw_uri = sanitize_text_field(
216	wp_unslash( $_SERVER['REQUEST_URI'] ) // WPCS: Input var okay.
217	);
218	} else {
219	$raw_uri = '';
220	}
221
222	$request = $this->finder->recognize_sitemap_uri( $raw_uri );
223
224	if ( isset( $request['sitemap_name'] ) ) {
225
226	/**
227	* Filter the content type used to serve the sitemap XML files.
228	*
229	* @module sitemaps
230	*
231	* @since 3.9.0
232	*
233	* @param string $xml_content_type By default, it's 'text/xml'.
234	*/
235	$xml_content_type = apply_filters( 'jetpack_sitemap_content_type', 'text/xml' );
236
237	// Catch master sitemap xml.
238	if ( preg_match( $regex['master'], $request['sitemap_name'] ) ) {
239	$sitemap_content = $this->librarian->get_sitemap_text(
240	jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
241	JP_MASTER_SITEMAP_TYPE
242	);
243
244	// if there is no master sitemap yet, let's just return an empty sitemap with a short TTL instead of a 404.
245	if ( empty( $sitemap_content ) ) {
246	$builder = new Jetpack_Sitemap_Builder();
247	$sitemap_content = $builder->empty_sitemap_xml();
248	}
249
250	$this->serve_raw_and_die(
251	$xml_content_type,
252	$sitemap_content
253	);
254	}
255
256	// Catch sitemap xml.
257	if ( preg_match( $regex['sitemap'], $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_TYPE
263	)
264	);
265	}
266
267	// Catch sitemap index xml.
268	if ( preg_match( $regex['index'], $request['sitemap_name'] ) ) {
269	$this->serve_raw_and_die(
270	$xml_content_type,
271	$this->librarian->get_sitemap_text(
272	$request['sitemap_name'],
273	JP_PAGE_SITEMAP_INDEX_TYPE
274	)
275	);
276	}
277
278	// Catch sitemap xsl.
279	if ( preg_match( $regex['sitemap-style'], $request['sitemap_name'] ) ) {
280	$this->serve_raw_and_die(
281	'application/xml',
282	Jetpack_Sitemap_Stylist::sitemap_xsl()
283	);
284	}
285
286	// Catch sitemap index xsl.
287	if ( preg_match( $regex['index-style'], $request['sitemap_name'] ) ) {
288	$this->serve_raw_and_die(
289	'application/xml',
290	Jetpack_Sitemap_Stylist::sitemap_index_xsl()
291	);
292	}
293
294	// Catch image sitemap xml.
295	if ( preg_match( $regex['image'], $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_TYPE
301	)
302	);
303	}
304
305	// Catch image sitemap index xml.
306	if ( preg_match( $regex['image-index'], $request['sitemap_name'] ) ) {
307	$this->serve_raw_and_die(
308	$xml_content_type,
309	$this->librarian->get_sitemap_text(
310	$request['sitemap_name'],
311	JP_IMAGE_SITEMAP_INDEX_TYPE
312	)
313	);
314	}
315
316	// Catch image sitemap xsl.
317	if ( preg_match( $regex['image-style'], $request['sitemap_name'] ) ) {
318	$this->serve_raw_and_die(
319	'application/xml',
320	Jetpack_Sitemap_Stylist::image_sitemap_xsl()
321	);
322	}
323
324	// Catch video sitemap xml.
325	if ( preg_match( $regex['video'], $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_TYPE
331	)
332	);
333	}
334
335	// Catch video sitemap index xml.
336	if ( preg_match( $regex['video-index'], $request['sitemap_name'] ) ) {
337	$this->serve_raw_and_die(
338	$xml_content_type,
339	$this->librarian->get_sitemap_text(
340	$request['sitemap_name'],
341	JP_VIDEO_SITEMAP_INDEX_TYPE
342	)
343	);
344	}
345
346	// Catch video sitemap xsl.
347	if ( preg_match( $regex['video-style'], $request['sitemap_name'] ) ) {
348	$this->serve_raw_and_die(
349	'application/xml',
350	Jetpack_Sitemap_Stylist::video_sitemap_xsl()
351	);
352	}
353
354	// Catch news sitemap xml.
355	if ( preg_match( $regex['news'], $request['sitemap_name'] ) ) {
356	$sitemap_builder = new Jetpack_Sitemap_Builder();
357	$this->serve_raw_and_die(
358	$xml_content_type,
359	$sitemap_builder->news_sitemap_xml()
360	);
361	}
362
363	// Catch news sitemap xsl.
364	if ( preg_match( $regex['news-style'], $request['sitemap_name'] ) ) {
365	$this->serve_raw_and_die(
366	'application/xml',
367	Jetpack_Sitemap_Stylist::news_sitemap_xsl()
368	);
369	}
370	}
371	}
372
373	/**
374	* Callback for adding sitemap-interval to the list of schedules.
375	*
376	* @access public
377	* @since 4.8.0
378	*
379	* @param array $schedules The array of WP_Cron schedules.
380	*
381	* @return array The updated array of WP_Cron schedules.
382	*/
383	public function callback_add_sitemap_schedule( $schedules ) {
384	$schedules['sitemap-interval'] = array(
385	'interval' => JP_SITEMAP_INTERVAL,
386	'display' => __( 'Sitemap Interval', 'jetpack' ),
387	);
388	return $schedules;
389	}
390
391	/**
392	* Callback handler for sitemap cron hook
393	*
394	* @access public
395	*/
396	public function callback_sitemap_cron_hook() {
397	$sitemap_builder = new Jetpack_Sitemap_Builder();
398	$sitemap_builder->update_sitemap();
399	}
400
401	/**
402	* Add actions to schedule sitemap generation.
403	* Should only be called once, in the constructor.
404	*
405	* @access private
406	* @since 4.8.0
407	*/
408	private function schedule_sitemap_generation() {
409	// Add cron schedule.
410	add_filter( 'cron_schedules', array( $this, 'callback_add_sitemap_schedule' ) ); // phpcs:ignore WordPress.WP.CronInterval.ChangeDetected
411
412	add_action(
413	'jp_sitemap_cron_hook',
414	array( $this, 'callback_sitemap_cron_hook' )
415	);
416
417	if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
418	/**
419	* Filter the delay in seconds until sitemap generation cron job is started.
420	*
421	* This filter allows a site operator or hosting provider to potentialy spread out sitemap generation for a
422	* lot of sites over time. By default, it will be randomly done over 15 minutes.
423	*
424	* @module sitemaps
425	* @since 6.6.1
426	*
427	* @param int $delay Time to delay in seconds.
428	*/
429	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
430	wp_schedule_event(
431	time() + $delay,
432	'sitemap-interval',
433	'jp_sitemap_cron_hook'
434	);
435	}
436	}
437
438	/**
439	* Callback to add sitemap to robots.txt.
440	*
441	* @access public
442	* @since 4.8.0
443	*/
444	public function callback_action_do_robotstxt() {
445
446	/**
447	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
448	*
449	* @module sitemaps
450	* @since 3.9.0
451	* @deprecated 7.4.0
452	*
453	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
454	*/
455	$discover_sitemap = apply_filters_deprecated( 'jetpack_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_sitemap_include_in_robotstxt' );
456
457	/**
458	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
459	*
460	* @module sitemaps
461	* @since 7.4.0
462	*
463	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
464	*/
465	$discover_sitemap = apply_filters( 'jetpack_sitemap_include_in_robotstxt', $discover_sitemap );
466
467	if ( true === $discover_sitemap ) {
468	$sitemap_url = $this->finder->construct_sitemap_url( 'sitemap.xml' );
469	echo 'Sitemap: ' . esc_url( $sitemap_url ) . "\n";
470	}
471
472	/**
473	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
474	*
475	* @module sitemaps
476	* @since 3.9.0
477	* @deprecated 7.4.0
478	*
479	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
480	*/
481	$discover_news_sitemap = apply_filters_deprecated( 'jetpack_news_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_news_sitemap_include_in_robotstxt' );
482
483	/**
484	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
485	*
486	* @module sitemaps
487	* @since 7.4.0
488	*
489	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
490	*/
491	$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_include_in_robotstxt', $discover_news_sitemap );
492
493	if ( true === $discover_news_sitemap ) {
494	$news_sitemap_url = $this->finder->construct_sitemap_url( 'news-sitemap.xml' );
495	echo 'Sitemap: ' . esc_url( $news_sitemap_url ) . "\n";
496	}
497	}
498
499	/**
500	* Callback to delete the news sitemap cache.
501	*
502	* @access public
503	* @since 4.8.0
504	*/
505	public function callback_action_flush_news_sitemap_cache() {
506	delete_transient( 'jetpack_news_sitemap_xml' );
507	}
508
509	/**
510	* Callback for resetting stored sitemap data.
511	*
512	* @access public
513	* @since 5.3.0
514	* @since 6.7.0 Schedules a regeneration.
515	*/
516	public function callback_action_purge_data() {
517	$this->callback_action_flush_news_sitemap_cache();
518	$this->librarian->delete_all_stored_sitemap_data();
519	/** This filter is documented in modules/sitemaps/sitemaps.php */
520	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
521	wp_schedule_single_event( time() + $delay, 'jp_sitemap_cron_hook' );
522	}
523
524	/**
525	* Callback to set the sitemap location.
526	*
527	* @access public
528	* @since 4.8.0
529	*/
530	public function callback_action_filter_sitemap_location() {
531	update_option(
532	'jetpack_sitemap_location',
533	/**
534	* Additional path for sitemap URIs. Default value is empty.
535	*
536	* This string is any additional path fragment you want included between
537	* the home URL and the sitemap filenames. Exactly how this fragment is
538	* interpreted depends on your permalink settings. For example:
539	*
540	* Pretty permalinks:
541	* home_url() . jetpack_sitemap_location . '/sitemap.xml'
542	*
543	* Plain ("ugly") permalinks:
544	* home_url() . jetpack_sitemap_location . '/?jetpack-sitemap=sitemap.xml'
545	*
546	* PATHINFO permalinks:
547	* home_url() . '/index.php' . jetpack_sitemap_location . '/sitemap.xml'
548	*
549	* where 'sitemap.xml' is the name of a specific sitemap file.
550	* The value of this filter must be a valid path fragment per RFC 3986;
551	* in particular it must either be empty or begin with a '/'.
552	* Also take care that any restrictions on sitemap location imposed by
553	* the sitemap protocol are satisfied.
554	*
555	* The result of this filter is stored in an option, 'jetpack_sitemap_location';
556	* that option is what gets read when the sitemap location is needed.
557	* This way we don't have to wait for init to finish before building sitemaps.
558	*
559	* @link https://tools.ietf.org/html/rfc3986#section-3.3 RFC 3986
560	* @link https://www.sitemaps.org/ The sitemap protocol
561	*
562	* @since 4.8.0
563	*/
564	apply_filters(
565	'jetpack_sitemap_location',
566	''
567	)
568	);
569	}
570	} // End Jetpack_Sitemap_Manager class.
571
572	new Jetpack_Sitemap_Manager();
573
574	/**
575	* Absolute URL of the current blog's sitemap.
576	*
577	* @module sitemaps
578	*
579	* @since 3.9.0
580	* @since 4.8.1 Code uses method found in Jetpack_Sitemap_Finder::construct_sitemap_url in 4.8.0.
581	* It has been moved here to avoid fatal errors with other plugins that were expecting to find this function.
582	*
583	* @param string $filename Sitemap file name. Defaults to 'sitemap.xml', the initial sitemaps page.
584	*
585	* @return string Sitemap URL.
586	*/
587	function jetpack_sitemap_uri( $filename = 'sitemap.xml' ) {
588	global $wp_rewrite;
589
590	$location = Jetpack_Options::get_option_and_ensure_autoload( 'jetpack_sitemap_location', '' );
591
592	if ( $wp_rewrite->using_index_permalinks() ) {
593	$sitemap_url = home_url( '/index.php' . $location . '/' . $filename );
594	} elseif ( $wp_rewrite->using_permalinks() ) {
595	$sitemap_url = home_url( $location . '/' . $filename );
596	} else {
597	$sitemap_url = home_url( $location . '/?jetpack-sitemap=' . $filename );
598	}
599
600	/**
601	* Filter sitemap URL relative to home URL.
602	*
603	* @module sitemaps
604	*
605	* @since 3.9.0
606	*
607	* @param string $sitemap_url Sitemap URL.
608	*/
609	return apply_filters( 'jetpack_sitemap_location', $sitemap_url );
610	}
611