sitemaps.php — Jetpack – WP Security, Backup, Speed, & Growth 12.3.2

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	'init',
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	*/
151	private function serve_raw_and_die( $the_content_type, $the_content ) {
152	header( 'Content-Type: ' . $the_content_type . '; charset=UTF-8' );
153
154	global $wp_query;
155	$wp_query->is_feed = true;
156	set_query_var( 'feed', 'sitemap' );
157
158	if ( '' === $the_content ) {
159	$error = __( 'No sitemap found. Please try again later.', 'jetpack' );
160	if ( current_user_can( 'manage_options' ) ) {
161	$next = human_time_diff( wp_next_scheduled( 'jp_sitemap_cron_hook' ) );
162	/* translators: %s is a human_time_diff until next sitemap generation. */
163	$error = sprintf( __( 'No sitemap found. The system will try to build it again in %s.', 'jetpack' ), $next );
164	}
165
166	wp_die(
167	esc_html( $error ),
168	esc_html__( 'Sitemaps', 'jetpack' ),
169	array(
170	'response' => 404,
171	)
172	);
173	}
174
175	echo $the_content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- All content created by Jetpack.
176
177	die();
178	}
179
180	/**
181	* Callback to intercept sitemap url requests and serve sitemap files.
182	*
183	* @access public
184	* @since 4.8.0
185	*/
186	public function callback_action_catch_sitemap_urls() {
187	// Regular expressions for sitemap URL routing.
188	$regex = array(
189	'master' => '/^sitemap\.xml$/',
190	'sitemap' => '/^sitemap-[1-9][0-9]*\.xml$/',
191	'index' => '/^sitemap-index-[1-9][0-9]*\.xml$/',
192	'sitemap-style' => '/^sitemap\.xsl$/',
193	'index-style' => '/^sitemap-index\.xsl$/',
194	'image' => '/^image-sitemap-[1-9][0-9]*\.xml$/',
195	'image-index' => '/^image-sitemap-index-[1-9][0-9]*\.xml$/',
196	'image-style' => '/^image-sitemap\.xsl$/',
197	'video' => '/^video-sitemap-[1-9][0-9]*\.xml$/',
198	'video-index' => '/^video-sitemap-index-[1-9][0-9]*\.xml$/',
199	'video-style' => '/^video-sitemap\.xsl$/',
200	'news' => '/^news-sitemap\.xml$/',
201	'news-style' => '/^news-sitemap\.xsl$/',
202	);
203
204	// The raw path(+query) of the requested URI.
205	if ( isset( $_SERVER['REQUEST_URI'] ) ) { // WPCS: Input var okay.
206	$raw_uri = sanitize_text_field(
207	wp_unslash( $_SERVER['REQUEST_URI'] ) // WPCS: Input var okay.
208	);
209	} else {
210	$raw_uri = '';
211	}
212
213	$request = $this->finder->recognize_sitemap_uri( $raw_uri );
214
215	if ( isset( $request['sitemap_name'] ) ) {
216
217	/**
218	* Filter the content type used to serve the sitemap XML files.
219	*
220	* @module sitemaps
221	*
222	* @since 3.9.0
223	*
224	* @param string $xml_content_type By default, it's 'text/xml'.
225	*/
226	$xml_content_type = apply_filters( 'jetpack_sitemap_content_type', 'text/xml' );
227
228	// Catch master sitemap xml.
229	if ( preg_match( $regex['master'], $request['sitemap_name'] ) ) {
230	$sitemap_content = $this->librarian->get_sitemap_text(
231	jp_sitemap_filename( JP_MASTER_SITEMAP_TYPE, 0 ),
232	JP_MASTER_SITEMAP_TYPE
233	);
234
235	// if there is no master sitemap yet, let's just return an empty sitemap with a short TTL instead of a 404.
236	if ( empty( $sitemap_content ) ) {
237	$builder = new Jetpack_Sitemap_Builder();
238	$sitemap_content = $builder->empty_sitemap_xml();
239	}
240
241	$this->serve_raw_and_die(
242	$xml_content_type,
243	$sitemap_content
244	);
245	}
246
247	// Catch sitemap xml.
248	if ( preg_match( $regex['sitemap'], $request['sitemap_name'] ) ) {
249	$this->serve_raw_and_die(
250	$xml_content_type,
251	$this->librarian->get_sitemap_text(
252	$request['sitemap_name'],
253	JP_PAGE_SITEMAP_TYPE
254	)
255	);
256	}
257
258	// Catch sitemap index xml.
259	if ( preg_match( $regex['index'], $request['sitemap_name'] ) ) {
260	$this->serve_raw_and_die(
261	$xml_content_type,
262	$this->librarian->get_sitemap_text(
263	$request['sitemap_name'],
264	JP_PAGE_SITEMAP_INDEX_TYPE
265	)
266	);
267	}
268
269	// Catch sitemap xsl.
270	if ( preg_match( $regex['sitemap-style'], $request['sitemap_name'] ) ) {
271	$this->serve_raw_and_die(
272	'application/xml',
273	Jetpack_Sitemap_Stylist::sitemap_xsl()
274	);
275	}
276
277	// Catch sitemap index xsl.
278	if ( preg_match( $regex['index-style'], $request['sitemap_name'] ) ) {
279	$this->serve_raw_and_die(
280	'application/xml',
281	Jetpack_Sitemap_Stylist::sitemap_index_xsl()
282	);
283	}
284
285	// Catch image sitemap xml.
286	if ( preg_match( $regex['image'], $request['sitemap_name'] ) ) {
287	$this->serve_raw_and_die(
288	$xml_content_type,
289	$this->librarian->get_sitemap_text(
290	$request['sitemap_name'],
291	JP_IMAGE_SITEMAP_TYPE
292	)
293	);
294	}
295
296	// Catch image sitemap index xml.
297	if ( preg_match( $regex['image-index'], $request['sitemap_name'] ) ) {
298	$this->serve_raw_and_die(
299	$xml_content_type,
300	$this->librarian->get_sitemap_text(
301	$request['sitemap_name'],
302	JP_IMAGE_SITEMAP_INDEX_TYPE
303	)
304	);
305	}
306
307	// Catch image sitemap xsl.
308	if ( preg_match( $regex['image-style'], $request['sitemap_name'] ) ) {
309	$this->serve_raw_and_die(
310	'application/xml',
311	Jetpack_Sitemap_Stylist::image_sitemap_xsl()
312	);
313	}
314
315	// Catch video sitemap xml.
316	if ( preg_match( $regex['video'], $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_VIDEO_SITEMAP_TYPE
322	)
323	);
324	}
325
326	// Catch video sitemap index xml.
327	if ( preg_match( $regex['video-index'], $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_VIDEO_SITEMAP_INDEX_TYPE
333	)
334	);
335	}
336
337	// Catch video sitemap xsl.
338	if ( preg_match( $regex['video-style'], $request['sitemap_name'] ) ) {
339	$this->serve_raw_and_die(
340	'application/xml',
341	Jetpack_Sitemap_Stylist::video_sitemap_xsl()
342	);
343	}
344
345	// Catch news sitemap xml.
346	if ( preg_match( $regex['news'], $request['sitemap_name'] ) ) {
347	$sitemap_builder = new Jetpack_Sitemap_Builder();
348	$this->serve_raw_and_die(
349	$xml_content_type,
350	$sitemap_builder->news_sitemap_xml()
351	);
352	}
353
354	// Catch news sitemap xsl.
355	if ( preg_match( $regex['news-style'], $request['sitemap_name'] ) ) {
356	$this->serve_raw_and_die(
357	'application/xml',
358	Jetpack_Sitemap_Stylist::news_sitemap_xsl()
359	);
360	}
361	}
362	}
363
364	/**
365	* Callback for adding sitemap-interval to the list of schedules.
366	*
367	* @access public
368	* @since 4.8.0
369	*
370	* @param array $schedules The array of WP_Cron schedules.
371	*
372	* @return array The updated array of WP_Cron schedules.
373	*/
374	public function callback_add_sitemap_schedule( $schedules ) {
375	$schedules['sitemap-interval'] = array(
376	'interval' => JP_SITEMAP_INTERVAL,
377	'display' => __( 'Sitemap Interval', 'jetpack' ),
378	);
379	return $schedules;
380	}
381
382	/**
383	* Callback handler for sitemap cron hook
384	*
385	* @access public
386	*/
387	public function callback_sitemap_cron_hook() {
388	$sitemap_builder = new Jetpack_Sitemap_Builder();
389	$sitemap_builder->update_sitemap();
390	}
391
392	/**
393	* Add actions to schedule sitemap generation.
394	* Should only be called once, in the constructor.
395	*
396	* @access private
397	* @since 4.8.0
398	*/
399	private function schedule_sitemap_generation() {
400	// Add cron schedule.
401	add_filter( 'cron_schedules', array( $this, 'callback_add_sitemap_schedule' ) ); // phpcs:ignore WordPress.WP.CronInterval.ChangeDetected
402
403	add_action(
404	'jp_sitemap_cron_hook',
405	array( $this, 'callback_sitemap_cron_hook' )
406	);
407
408	if ( ! wp_next_scheduled( 'jp_sitemap_cron_hook' ) ) {
409	/**
410	* Filter the delay in seconds until sitemap generation cron job is started.
411	*
412	* This filter allows a site operator or hosting provider to potentialy spread out sitemap generation for a
413	* lot of sites over time. By default, it will be randomly done over 15 minutes.
414	*
415	* @module sitemaps
416	* @since 6.6.1
417	*
418	* @param int $delay Time to delay in seconds.
419	*/
420	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
421	wp_schedule_event(
422	time() + $delay,
423	'sitemap-interval',
424	'jp_sitemap_cron_hook'
425	);
426	}
427	}
428
429	/**
430	* Callback to add sitemap to robots.txt.
431	*
432	* @access public
433	* @since 4.8.0
434	*/
435	public function callback_action_do_robotstxt() {
436
437	/**
438	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
439	*
440	* @module sitemaps
441	* @since 3.9.0
442	* @deprecated 7.4.0
443	*
444	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
445	*/
446	$discover_sitemap = apply_filters_deprecated( 'jetpack_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_sitemap_include_in_robotstxt' );
447
448	/**
449	* Filter whether to make the default sitemap discoverable to robots or not. Default true.
450	*
451	* @module sitemaps
452	* @since 7.4.0
453	*
454	* @param bool $discover_sitemap Make default sitemap discoverable to robots.
455	*/
456	$discover_sitemap = apply_filters( 'jetpack_sitemap_include_in_robotstxt', $discover_sitemap );
457
458	if ( true === $discover_sitemap ) {
459	$sitemap_url = $this->finder->construct_sitemap_url( 'sitemap.xml' );
460	echo 'Sitemap: ' . esc_url( $sitemap_url ) . "\n";
461	}
462
463	/**
464	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
465	*
466	* @module sitemaps
467	* @since 3.9.0
468	* @deprecated 7.4.0
469	*
470	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
471	*/
472	$discover_news_sitemap = apply_filters_deprecated( 'jetpack_news_sitemap_generate', array( true ), 'jetpack-7.4.0', 'jetpack_news_sitemap_include_in_robotstxt' );
473
474	/**
475	* Filter whether to make the news sitemap discoverable to robots or not. Default true.
476	*
477	* @module sitemaps
478	* @since 7.4.0
479	*
480	* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots.
481	*/
482	$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_include_in_robotstxt', $discover_news_sitemap );
483
484	if ( true === $discover_news_sitemap ) {
485	$news_sitemap_url = $this->finder->construct_sitemap_url( 'news-sitemap.xml' );
486	echo 'Sitemap: ' . esc_url( $news_sitemap_url ) . "\n";
487	}
488	}
489
490	/**
491	* Callback to delete the news sitemap cache.
492	*
493	* @access public
494	* @since 4.8.0
495	*/
496	public function callback_action_flush_news_sitemap_cache() {
497	delete_transient( 'jetpack_news_sitemap_xml' );
498	}
499
500	/**
501	* Callback for resetting stored sitemap data.
502	*
503	* @access public
504	* @since 5.3.0
505	* @since 6.7.0 Schedules a regeneration.
506	*/
507	public function callback_action_purge_data() {
508	$this->callback_action_flush_news_sitemap_cache();
509	$this->librarian->delete_all_stored_sitemap_data();
510	/** This filter is documented in modules/sitemaps/sitemaps.php */
511	$delay = apply_filters( 'jetpack_sitemap_generation_delay', MINUTE_IN_SECONDS * wp_rand( 1, 15 ) ); // Randomly space it out to start within next fifteen minutes.
512	wp_schedule_single_event( time() + $delay, 'jp_sitemap_cron_hook' );
513	}
514
515	/**
516	* Callback to set the sitemap location.
517	*
518	* @access public
519	* @since 4.8.0
520	*/
521	public function callback_action_filter_sitemap_location() {
522	update_option(
523	'jetpack_sitemap_location',
524	/**
525	* Additional path for sitemap URIs. Default value is empty.
526	*
527	* This string is any additional path fragment you want included between
528	* the home URL and the sitemap filenames. Exactly how this fragment is
529	* interpreted depends on your permalink settings. For example:
530	*
531	* Pretty permalinks:
532	* home_url() . jetpack_sitemap_location . '/sitemap.xml'
533	*
534	* Plain ("ugly") permalinks:
535	* home_url() . jetpack_sitemap_location . '/?jetpack-sitemap=sitemap.xml'
536	*
537	* PATHINFO permalinks:
538	* home_url() . '/index.php' . jetpack_sitemap_location . '/sitemap.xml'
539	*
540	* where 'sitemap.xml' is the name of a specific sitemap file.
541	* The value of this filter must be a valid path fragment per RFC 3986;
542	* in particular it must either be empty or begin with a '/'.
543	* Also take care that any restrictions on sitemap location imposed by
544	* the sitemap protocol are satisfied.
545	*
546	* The result of this filter is stored in an option, 'jetpack_sitemap_location';
547	* that option is what gets read when the sitemap location is needed.
548	* This way we don't have to wait for init to finish before building sitemaps.
549	*
550	* @link https://tools.ietf.org/html/rfc3986#section-3.3 RFC 3986
551	* @link https://www.sitemaps.org/ The sitemap protocol
552	*
553	* @since 4.8.0
554	*/
555	apply_filters(
556	'jetpack_sitemap_location',
557	''
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