sitemaps.php — Jetpack – WP Security, Backup, Speed, & Growth 15.9-a.7

jetpack / modules / sitemaps / sitemaps.php

jetpack / modules / sitemaps Last commit date

sitemaps.php

614 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	if ( ! defined( 'ABSPATH' ) ) {
35	exit( 0 );
36	}
37
38	/* Include all of the sitemap subclasses. */
39	require_once __DIR__ . '/sitemap-constants.php';
40	require_once __DIR__ . '/sitemap-buffer.php';
41	require_once __DIR__ . '/sitemap-buffer-fallback.php';
42	require_once __DIR__ . '/sitemap-buffer-xmlwriter.php';
43	require_once __DIR__ . '/sitemap-buffer-page-xmlwriter.php';
44	require_once __DIR__ . '/sitemap-buffer-image-xmlwriter.php';
45	require_once __DIR__ . '/sitemap-buffer-video-xmlwriter.php';
46	require_once __DIR__ . '/sitemap-buffer-news-xmlwriter.php';
47	require_once __DIR__ . '/sitemap-buffer-master-xmlwriter.php';
48	require_once __DIR__ . '/sitemap-buffer-factory.php';
49	require_once __DIR__ . '/sitemap-stylist.php';
50	require_once __DIR__ . '/sitemap-librarian.php';
51	require_once __DIR__ . '/sitemap-finder.php';
52	require_once __DIR__ . '/sitemap-builder.php';
53
54	if ( defined( 'WP_DEBUG' ) && WP_DEBUG ) {
55	require_once __DIR__ . '/sitemap-logger.php';
56	}
57
58	// phpcs:disable Universal.Files.SeparateFunctionsFromOO.Mixed -- TODO: Move classes to appropriately-named class files.
59
60	/**
61	* Governs the generation, storage, and serving of sitemaps.
62	*
63	* @since 4.8.0
64	*
65	* @phan-constructor-used-for-side-effects
66	*/
67	class Jetpack_Sitemap_Manager {
68
69	/**
70	* Librarian object for storing and retrieving sitemap data.
71	*
72	* @see Jetpack_Sitemap_Librarian
73	* @since 4.8.0
74	* @var Jetpack_Sitemap_Librarian $librarian Librarian object for storing and retrieving sitemap data.
75	*/
76	private $librarian;
77
78	/**
79	* Logger object for reporting debug messages.
80	*
81	* @see Jetpack_Sitemap_Logger
82	* @since 4.8.0
83	* @var Jetpack_Sitemap_Logger $logger Logger object for reporting debug messages.
84	*/
85	private $logger;
86
87	/**
88	* Finder object for handling sitemap URIs.
89	*
90	* @see Jetpack_Sitemap_Finder
91	* @since 4.8.0
92	* @var Jetpack_Sitemap_Finder $finder Finder object for handling with sitemap URIs.
93	*/
94	private $finder;
95
96	/**
97	* Construct a new Jetpack_Sitemap_Manager.
98	*
99	* @access public
100	* @since 4.8.0
101	*/
102	public function __construct() {
103	$this->librarian = new Jetpack_Sitemap_Librarian();
104	$this->finder = new Jetpack_Sitemap_Finder();
105
106	if ( defined( 'WP_DEBUG' ) && ( true === WP_DEBUG ) ) {
107	$this->logger = new Jetpack_Sitemap_Logger();
108	}
109
110	// Add callback for sitemap URL handler.
111	add_action(
112	'wp_loaded',
113	array( $this, 'callback_action_catch_sitemap_urls' ),
114	defined( 'IS_WPCOM' ) && IS_WPCOM ? 100 : 10
115	);
116
117	// Add generator to wp_cron task list.
118	$this->schedule_sitemap_generation();
119
120	// Add sitemap to robots.txt.
121	add_action(
122	'do_robotstxt',
123	array( $this, 'callback_action_do_robotstxt' ),
124	20
125	);
126
127	// The news sitemap is cached; here we add a callback to
128	// flush the cached news sitemap when a post is published.
129	add_action(
130	'publish_post',
131	array( $this, 'callback_action_flush_news_sitemap_cache' ),
132	10
133	);
134
135	// In case we need to purge all sitemaps, we do this.
136	add_action(
137	'jetpack_sitemaps_purge_data',
138	array( $this, 'callback_action_purge_data' )
139	);
140
141	/*
142	* Module parameters are stored as options in the database.
143	* This allows us to avoid having to process all of init
144	* before serving the sitemap data. The following actions
145	* process and store these filters.
146	*/
147
148	// Process filters and store location string for sitemap.
149	add_action(
150	'init',
151	array( $this, 'callback_action_filter_sitemap_location' ),
152	999
153	);
154	}
155
156	/**
157	* Echo a raw string of given content-type.
158	*
159	* @access private
160	* @since 4.8.0
161	*
162	* @param string $the_content_type The content type to be served.
163	* @param string $the_content The string to be echoed.
164	* @return never
165	*/
166	private function serve_raw_and_die( $the_content_type, $the_content ) {
167	header( 'Content-Type: ' . $the_content_type . '; charset=UTF-8' );
168
169	global $wp_query;
170	$wp_query->is_feed = true;
171	set_query_var( 'feed', 'sitemap' );
172
173	if ( '' === $the_content ) {
174	$error = __( 'No sitemap found. Please try again later.', 'jetpack' );
175	if ( current_user_can( 'manage_options' ) ) {
176	$next = human_time_diff( wp_next_scheduled( 'jp_sitemap_cron_hook' ) );
177	/* translators: %s is a human_time_diff until next sitemap generation. */
178	$error = sprintf( __( 'No sitemap found. The system will try to build it again in %s.', 'jetpack' ), $next );
179	}
180
181	wp_die(
182	esc_html( $error ),
183	esc_html__( 'Sitemaps', 'jetpack' ),
184	array(
185	'response' => 404,
186	)
187	);
188	}
189
190	echo $the_content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- All content created by Jetpack.
191
192	die( 0 );
193	}
194
195	/**
196	* Callback to intercept sitemap url requests and serve sitemap files.
197	*
198	* @access public
199	* @since 4.8.0
200	*/
201	public function callback_action_catch_sitemap_urls() {
202	// Regular expressions for sitemap URL routing.
203	$regex = array(
204	'sitemap' => '/^sitemap-[1-9][0-9]*\.xml$/',
205	'index' => '/^sitemap-index-[1-9][0-9]*\.xml$/',
206	'image' => '/^image-sitemap-[1-9][0-9]*\.xml$/',
207	'image-index' => '/^image-sitemap-index-[1-9][0-9]*\.xml$/',
208	'video' => '/^video-sitemap-[1-9][0-9]*\.xml$/',
209	'video-index' => '/^video-sitemap-index-[1-9][0-9]*\.xml$/',
210	);
211
212	// The raw path(+query) of the requested URI.
213	if ( isset( $_SERVER['REQUEST_URI'] ) ) { // WPCS: Input var okay.
214	$raw_uri = sanitize_text_field(
215	wp_unslash( $_SERVER['REQUEST_URI'] ) // WPCS: Input var okay.
216	);
217	} else {
218	$raw_uri = '';
219	}
220
221	$request = $this->finder->recognize_sitemap_uri( $raw_uri );
222
223	if ( isset( $request['sitemap_name'] ) ) {
224
225	/**
226	* Filter the content type used to serve the sitemap XML files.
227	*
228	* @module sitemaps
229	*
230	* @since 3.9.0
231	*
232	* @param string $xml_content_type By default, it's 'text/xml'.
233	*/
234	$xml_content_type = apply_filters( 'jetpack_sitemap_content_type', 'text/xml' );
235
236	// Catch master sitemap xml.
237	if ( 'sitemap.xml' === $request['sitemap_name'] ) {
238	$sitemap_content = $this->librarian->get_sitemap_text(
239	jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
240	JP_MASTER_SITEMAP_TYPE
241	);
242
243	// if there is no master sitemap yet, let's just return an empty sitemap with a short TTL instead of a 404.
244	if ( empty( $sitemap_content ) ) {
245	$builder = new Jetpack_Sitemap_Builder();
246	$sitemap_content = $builder->empty_sitemap_xml();
247	}
248
249	$this->serve_raw_and_die(
250	$xml_content_type,
251	$sitemap_content
252	);
253	}
254
255	// Catch sitemap xsl.
256	if ( 'sitemap.xsl' === $request['sitemap_name'] ) {
257	$this->serve_raw_and_die(
258	'application/xml',
259	Jetpack_Sitemap_Stylist::sitemap_xsl()
260	);
261	}
262
263	// Catch sitemap index xsl.
264	if ( 'sitemap-index.xsl' === $request['sitemap_name'] ) {
265	$this->serve_raw_and_die(
266	'application/xml',
267	Jetpack_Sitemap_Stylist::sitemap_index_xsl()
268	);
269	}
270
271	// Catch image sitemap xsl.
272	if ( 'image-sitemap.xsl' === $request['sitemap_name'] ) {
273	$this->serve_raw_and_die(
274	'application/xml',
275	Jetpack_Sitemap_Stylist::image_sitemap_xsl()
276	);
277	}
278
279	// Catch video sitemap xsl.
280	if ( 'video-sitemap.xsl' === $request['sitemap_name'] ) {
281	$this->serve_raw_and_die(
282	'application/xml',
283	Jetpack_Sitemap_Stylist::video_sitemap_xsl()
284	);
285	}
286
287	// Catch news sitemap xml.
288	if ( 'news-sitemap.xml' === $request['sitemap_name'] ) {
289	$sitemap_builder = new Jetpack_Sitemap_Builder();
290	$this->serve_raw_and_die(
291	$xml_content_type,
292	$sitemap_builder->news_sitemap_xml()
293	);
294	}
295
296	// Catch news sitemap xsl.
297	if ( 'news-sitemap.xsl' === $request['sitemap_name'] ) {
298	$this->serve_raw_and_die(
299	'application/xml',
300	Jetpack_Sitemap_Stylist::news_sitemap_xsl()
301	);
302	}
303
304	// Catch sitemap xml.
305	if ( preg_match( $regex['sitemap'], $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_PAGE_SITEMAP_TYPE
311	)
312	);
313	}
314
315	// Catch sitemap index xml.
316	if ( preg_match( $regex['index'], $request['sitemap_name'] ) ) {
317	$this->serve_raw_and_die(
318	$xml_content_type,
319	$this->librarian->get_sitemap_text(
320	$request['sitemap_name'],
321	JP_PAGE_SITEMAP_INDEX_TYPE
322	)
323	);
324	}
325
326	// Catch image sitemap xml.
327	if ( preg_match( $regex['image'], $request['sitemap_name'] ) ) {
328	$this->serve_raw_and_die(
329	$xml_content_type,
330	$this->librarian->get_sitemap_text(
331	$request['sitemap_name'],
332	JP_IMAGE_SITEMAP_TYPE
333	)
334	);
335	}
336
337	// Catch image sitemap index xml.
338	if ( preg_match( $regex['image-index'], $request['sitemap_name'] ) ) {
339	$this->serve_raw_and_die(
340	$xml_content_type,
341	$this->librarian->get_sitemap_text(
342	$request['sitemap_name'],
343	JP_IMAGE_SITEMAP_INDEX_TYPE
344	)
345	);
346	}
347
348	// Catch video sitemap xml.
349	if ( preg_match( $regex['video'], $request['sitemap_name'] ) ) {
350	$this->serve_raw_and_die(
351	$xml_content_type,
352	$this->librarian->get_sitemap_text(
353	$request['sitemap_name'],
354	JP_VIDEO_SITEMAP_TYPE
355	)
356	);
357	}
358
359	// Catch video sitemap index xml.
360	if ( preg_match( $regex['video-index'], $request['sitemap_name'] ) ) {
361	$this->serve_raw_and_die(
362	$xml_content_type,
363	$this->librarian->get_sitemap_text(
364	$request['sitemap_name'],
365	JP_VIDEO_SITEMAP_INDEX_TYPE
366	)
367	);
368	}
369	}
370	}
371
372	/**
373	* Callback for adding sitemap-interval to the list of schedules.
374	*
375	* @access public
376	* @since 4.8.0
377	*
378	* @param array $schedules The array of WP_Cron schedules.
379	*
380	* @return array The updated array of WP_Cron schedules.
381	*/
382	public function callback_add_sitemap_schedule( $schedules ) {
383	$schedules['sitemap-interval'] = array(
384	'interval' => JP_SITEMAP_INTERVAL,
385	'display' => __( 'Sitemap Interval', 'jetpack' ),
386	);
387	return $schedules;
388	}
389
390	/**
391	* Callback handler for sitemap cron hook
392	*
393	* @access public
394	*/
395	public function callback_sitemap_cron_hook() {
396	$sitemap_builder = new Jetpack_Sitemap_Builder();
397	$sitemap_builder->update_sitemap();
398	}
399
400	/**
401	* Add actions to schedule sitemap generation.
402	* Should only be called once, in the constructor.
403	*
404	* @access private
405	* @since 4.8.0
406	*/
407	private function schedule_sitemap_generation() {
408	// Add cron schedule.
409	add_filter( 'cron_schedules', array( $this, 'callback_add_sitemap_schedule' ) ); // phpcs:ignore WordPress.WP.CronInterval.ChangeDetected
410
411	add_action(
412	'jp_sitemap_cron_hook',
413	array( $this, 'callback_sitemap_cron_hook' )
414	);
415
416	if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
417	/**
418	* Filter the delay in seconds until sitemap generation cron job is started.
419	*
420	* This filter allows a site operator or hosting provider to potentialy spread out sitemap generation for a
421	* lot of sites over time. By default, it will be randomly done over 15 minutes.
422	*
423	* @module sitemaps
424	* @since 6.6.1
425	*
426	* @param int $delay Time to delay in seconds.
427	*/
428	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
429	wp_schedule_event(
430	time() + $delay,
431	'sitemap-interval',
432	'jp_sitemap_cron_hook'
433	);
434	}
435	}
436
437	/**
438	* Callback to add sitemap to robots.txt.
439	*
440	* @access public
441	* @since 4.8.0
442	*/
443	public function callback_action_do_robotstxt() {
444
445	/**
446	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
447	*
448	* @module sitemaps
449	* @since 3.9.0
450	* @deprecated 7.4.0
451	*
452	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
453	*/
454	$discover_sitemap = apply_filters_deprecated( 'jetpack_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_sitemap_include_in_robotstxt' );
455
456	/**
457	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
458	*
459	* @module sitemaps
460	* @since 7.4.0
461	*
462	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
463	*/
464	$discover_sitemap = apply_filters( 'jetpack_sitemap_include_in_robotstxt', $discover_sitemap );
465
466	if ( true === $discover_sitemap ) {
467	$sitemap_url = $this->finder->construct_sitemap_url( 'sitemap.xml' );
468	echo 'Sitemap: ' . esc_url( $sitemap_url ) . "\n";
469	}
470
471	/**
472	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
473	*
474	* @module sitemaps
475	* @since 3.9.0
476	* @deprecated 7.4.0
477	*
478	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
479	*/
480	$discover_news_sitemap = apply_filters_deprecated( 'jetpack_news_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_news_sitemap_include_in_robotstxt' );
481
482	/**
483	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
484	*
485	* @module sitemaps
486	* @since 7.4.0
487	*
488	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
489	*/
490	$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_include_in_robotstxt', $discover_news_sitemap );
491
492	if ( true === $discover_news_sitemap ) {
493	$news_sitemap_url = $this->finder->construct_sitemap_url( 'news-sitemap.xml' );
494	echo 'Sitemap: ' . esc_url( $news_sitemap_url ) . "\n";
495	}
496	}
497
498	/**
499	* Callback to delete the news sitemap cache.
500	*
501	* @access public
502	* @since 4.8.0
503	*/
504	public function callback_action_flush_news_sitemap_cache() {
505	delete_transient( 'jetpack_news_sitemap_xml' );
506	}
507
508	/**
509	* Callback for resetting stored sitemap data.
510	*
511	* @access public
512	* @since 5.3.0
513	* @since 6.7.0 Schedules a regeneration.
514	*/
515	public function callback_action_purge_data() {
516	$this->callback_action_flush_news_sitemap_cache();
517	$this->librarian->delete_all_stored_sitemap_data();
518	/** This filter is documented in modules/sitemaps/sitemaps.php */
519	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
520	wp_schedule_single_event( time() + $delay, 'jp_sitemap_cron_hook' );
521	}
522
523	/**
524	* Callback to set the sitemap location.
525	*
526	* @access public
527	* @since 4.8.0
528	*/
529	public function callback_action_filter_sitemap_location() {
530	update_option(
531	'jetpack_sitemap_location',
532	/**
533	* Additional path for sitemap URIs. Default value is empty.
534	*
535	* This string is any additional path fragment you want included between
536	* the home URL and the sitemap filenames. Exactly how this fragment is
537	* interpreted depends on your permalink settings. For example:
538	*
539	* Pretty permalinks:
540	* home_url() . jetpack_sitemap_location . '/sitemap.xml'
541	*
542	* Plain ("ugly") permalinks:
543	* home_url() . jetpack_sitemap_location . '/?jetpack-sitemap=sitemap.xml'
544	*
545	* PATHINFO permalinks:
546	* home_url() . '/index.php' . jetpack_sitemap_location . '/sitemap.xml'
547	*
548	* where 'sitemap.xml' is the name of a specific sitemap file.
549	* The value of this filter must be a valid path fragment per RFC 3986;
550	* in particular it must either be empty or begin with a '/'.
551	* Also take care that any restrictions on sitemap location imposed by
552	* the sitemap protocol are satisfied.
553	*
554	* The result of this filter is stored in an option, 'jetpack_sitemap_location';
555	* that option is what gets read when the sitemap location is needed.
556	* This way we don't have to wait for init to finish before building sitemaps.
557	*
558	* @link https://tools.ietf.org/html/rfc3986#section-3.3 RFC 3986
559	* @link https://www.sitemaps.org/ The sitemap protocol
560	*
561	* @since 4.8.0
562	*/
563	apply_filters(
564	'jetpack_sitemap_location',
565	''
566	)
567	);
568	}
569	} // End Jetpack_Sitemap_Manager class.
570
571	new Jetpack_Sitemap_Manager();
572
573	/**
574	* Absolute URL of the current blog's sitemap.
575	*
576	* @module sitemaps
577	*
578	* @since 3.9.0
579	* @since 4.8.1 Code uses method found in Jetpack_Sitemap_Finder::construct_sitemap_url in 4.8.0.
580	* It has been moved here to avoid fatal errors with other plugins that were expecting to find this function.
581	*
582	* @param string $filename Sitemap file name. Defaults to 'sitemap.xml', the initial sitemaps page.
583	*
584	* @return string Sitemap URL.
585	*/
586	function jetpack_sitemap_uri( $filename = 'sitemap.xml' ) {
587	global $wp_rewrite;
588
589	$location = Jetpack_Options::get_option_and_ensure_autoload( 'jetpack_sitemap_location', '' );
590
591	if ( $wp_rewrite->using_index_permalinks() ) {
592	$sitemap_url = home_url( '/index.php' . $location . '/' . $filename );
593	} elseif ( $wp_rewrite->using_permalinks() ) {
594	$sitemap_url = home_url( $location . '/' . $filename );
595	} else {
596	$sitemap_url = home_url( $location . '/?jetpack-sitemap=' . $filename );
597	}
598
599	/**
600	* Filter sitemap URL relative to home URL.
601	*
602	* @module sitemaps
603	*
604	* @since 3.9.0
605	*
606	* @param string $sitemap_url Sitemap URL.
607	*/
608	return apply_filters( 'jetpack_sitemap_location', $sitemap_url );
609	}
610
611	// Register Jetpack Sitemaps abilities (WordPress Abilities API, WP 6.9+).
612	require_once __DIR__ . '/abilities/class-sitemaps-abilities.php';
613	\Automattic\Jetpack\Plugin\Abilities\Sitemaps_Abilities::init();
614