wordpress-seo / inc / sitemaps / class-sitemaps.php

class-sitemaps.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 18.1, at inc/sitemaps/class-sitemaps.php

648 lines 16.1 KB

Raw Download Zip

1	<?php
2	/**
3	* WPSEO plugin file.
4	*
5	* @package WPSEO\XML_Sitemaps
6	*/
7
8	/**
9	* Class WPSEO_Sitemaps.
10	*
11	* @todo This class could use a general description with some explanation on sitemaps. OR.
12	*/
13	class WPSEO_Sitemaps {
14
15	/**
16	* Sitemap index identifier.
17	*
18	* @var string
19	*/
20	const SITEMAP_INDEX_TYPE = '1';
21
22	/**
23	* Content of the sitemap to output.
24	*
25	* @var string
26	*/
27	protected $sitemap = '';
28
29	/**
30	* Flag to indicate if this is an invalid or empty sitemap.
31	*
32	* @var bool
33	*/
34	public $bad_sitemap = false;
35
36	/**
37	* Whether or not the XML sitemap was served from a transient or not.
38	*
39	* @var bool
40	*/
41	private $transient = false;
42
43	/**
44	* HTTP protocol to use in headers.
45	*
46	* @since 3.2
47	*
48	* @var string
49	*/
50	protected $http_protocol = 'HTTP/1.1';
51
52	/**
53	* Holds the n variable.
54	*
55	* @var int
56	*/
57	private $current_page = 1;
58
59	/**
60	* The sitemaps router.
61	*
62	* @since 3.2
63	*
64	* @var WPSEO_Sitemaps_Router
65	*/
66	public $router;
67
68	/**
69	* The sitemap renderer.
70	*
71	* @since 3.2
72	*
73	* @var WPSEO_Sitemaps_Renderer
74	*/
75	public $renderer;
76
77	/**
78	* The sitemap cache.
79	*
80	* @since 3.2
81	*
82	* @var WPSEO_Sitemaps_Cache
83	*/
84	public $cache;
85
86	/**
87	* The sitemap providers.
88	*
89	* @since 3.2
90	*
91	* @var WPSEO_Sitemap_Provider[]
92	*/
93	public $providers;
94
95	/**
96	* Class constructor.
97	*/
98	public function __construct() {
99
100	add_action( 'after_setup_theme', [ $this, 'init_sitemaps_providers' ] );
101	add_action( 'after_setup_theme', [ $this, 'reduce_query_load' ], 99 );
102	add_action( 'pre_get_posts', [ $this, 'redirect' ], 1 );
103	add_action( 'wpseo_hit_sitemap_index', [ $this, 'hit_sitemap_index' ] );
104	add_action( 'wpseo_ping_search_engines', [ __CLASS__, 'ping_search_engines' ] );
105
106	$this->router = new WPSEO_Sitemaps_Router();
107	$this->renderer = new WPSEO_Sitemaps_Renderer();
108	$this->cache = new WPSEO_Sitemaps_Cache();
109
110	if ( ! empty( $_SERVER['SERVER_PROTOCOL'] ) ) {
111	$this->http_protocol = sanitize_text_field( wp_unslash( $_SERVER['SERVER_PROTOCOL'] ) );
112	}
113	}
114
115	/**
116	* Initialize sitemap providers classes.
117	*
118	* @since 5.3
119	*/
120	public function init_sitemaps_providers() {
121
122	$this->providers = [
123	new WPSEO_Post_Type_Sitemap_Provider(),
124	new WPSEO_Taxonomy_Sitemap_Provider(),
125	new WPSEO_Author_Sitemap_Provider(),
126	];
127
128	$external_providers = apply_filters( 'wpseo_sitemaps_providers', [] );
129
130	foreach ( $external_providers as $provider ) {
131	if ( is_object( $provider ) && $provider instanceof WPSEO_Sitemap_Provider ) {
132	$this->providers[] = $provider;
133	}
134	}
135	}
136
137	/**
138	* Check the current request URI, if we can determine it's probably an XML sitemap, kill loading the widgets.
139	*/
140	public function reduce_query_load() {
141	if ( ! isset( $_SERVER['REQUEST_URI'] ) ) {
142	return;
143	}
144	$request_uri = sanitize_text_field( wp_unslash( $_SERVER['REQUEST_URI'] ) );
145	$extension = substr( $request_uri, -4 );
146	if ( stripos( $request_uri, 'sitemap' ) !== false && in_array( $extension, [ '.xml', '.xsl' ], true ) ) {
147	remove_all_actions( 'widgets_init' );
148	}
149	}
150
151	/**
152	* Register your own sitemap. Call this during 'init'.
153	*
154	* @param string $name The name of the sitemap.
155	* @param callback $building_function Function to build your sitemap.
156	* @param string $rewrite Optional. Regular expression to match your sitemap with.
157	*/
158	public function register_sitemap( $name, $building_function, $rewrite = '' ) {
159	add_action( 'wpseo_do_sitemap_' . $name, $building_function );
160	if ( ! empty( $rewrite ) ) {
161	add_rewrite_rule( $rewrite, 'index.php?sitemap=' . $name, 'top' );
162	}
163	}
164
165	/**
166	* Register your own XSL file. Call this during 'init'.
167	*
168	* @since 1.4.23
169	*
170	* @param string $name The name of the XSL file.
171	* @param callback $building_function Function to build your XSL file.
172	* @param string $rewrite Optional. Regular expression to match your sitemap with.
173	*/
174	public function register_xsl( $name, $building_function, $rewrite = '' ) {
175	add_action( 'wpseo_xsl_' . $name, $building_function );
176	if ( ! empty( $rewrite ) ) {
177	add_rewrite_rule( $rewrite, 'index.php?yoast-sitemap-xsl=' . $name, 'top' );
178	}
179	}
180
181	/**
182	* Set the sitemap current page to allow creating partial sitemaps with WP-CLI
183	* in a one-off process.
184	*
185	* @param int $current_page The part that should be generated.
186	*/
187	public function set_n( $current_page ) {
188	if ( is_scalar( $current_page ) && intval( $current_page ) > 0 ) {
189	$this->current_page = intval( $current_page );
190	}
191	}
192
193	/**
194	* Set the sitemap content to display after you have generated it.
195	*
196	* @param string $sitemap The generated sitemap to output.
197	*/
198	public function set_sitemap( $sitemap ) {
199	$this->sitemap = $sitemap;
200	}
201
202	/**
203	* Set as true to make the request 404. Used stop the display of empty sitemaps or invalid requests.
204	*
205	* @param bool $is_bad Is this a bad request. True or false.
206	*/
207	public function set_bad_sitemap( $is_bad ) {
208	$this->bad_sitemap = (bool) $is_bad;
209	}
210
211	/**
212	* Prevent stupid plugins from running shutdown scripts when we're obviously not outputting HTML.
213	*
214	* @since 1.4.16
215	*/
216	public function sitemap_close() {
217	remove_all_actions( 'wp_footer' );
218	die();
219	}
220
221	/**
222	* Hijack requests for potential sitemaps and XSL files.
223	*
224	* @param \WP_Query $query Main query instance.
225	*/
226	public function redirect( $query ) {
227
228	if ( ! $query->is_main_query() ) {
229	return;
230	}
231
232	$yoast_sitemap_xsl = get_query_var( 'yoast-sitemap-xsl' );
233
234	if ( ! empty( $yoast_sitemap_xsl ) ) {
235	/*
236	* This is a method to provide the XSL via the home_url.
237	* Needed when the site_url and home_url are not the same.
238	* Loading the XSL needs to come from the same domain, protocol and port as the XML.
239	*
240	* Whenever home_url and site_url are the same, the file can be loaded directly.
241	*/
242	$this->xsl_output( $yoast_sitemap_xsl );
243	$this->sitemap_close();
244
245	return;
246	}
247
248	$type = get_query_var( 'sitemap' );
249
250	if ( empty( $type ) ) {
251	return;
252	}
253
254	$this->set_n( get_query_var( 'sitemap_n' ) );
255
256	if ( ! $this->get_sitemap_from_cache( $type, $this->current_page ) ) {
257	$this->build_sitemap( $type );
258	}
259
260	if ( $this->bad_sitemap ) {
261	$query->set_404();
262	status_header( 404 );
263
264	return;
265	}
266
267	$this->output();
268	$this->sitemap_close();
269	}
270
271	/**
272	* Try to get the sitemap from cache.
273	*
274	* @param string $type Sitemap type.
275	* @param int $page_number The page number to retrieve.
276	*
277	* @return bool If the sitemap has been retrieved from cache.
278	*/
279	private function get_sitemap_from_cache( $type, $page_number ) {
280
281	$this->transient = false;
282
283	if ( $this->cache->is_enabled() !== true ) {
284	return false;
285	}
286
287	/**
288	* Fires before the attempt to retrieve XML sitemap from the transient cache.
289	*
290	* @param WPSEO_Sitemaps $sitemaps Sitemaps object.
291	*/
292	do_action( 'wpseo_sitemap_stylesheet_cache_' . $type, $this );
293
294	$sitemap_cache_data = $this->cache->get_sitemap_data( $type, $page_number );
295
296	// No cache was found, refresh it because cache is enabled.
297	if ( empty( $sitemap_cache_data ) ) {
298	return $this->refresh_sitemap_cache( $type, $page_number );
299	}
300
301	// Cache object was found, parse information.
302	$this->transient = true;
303
304	$this->sitemap = $sitemap_cache_data->get_sitemap();
305	$this->bad_sitemap = ! $sitemap_cache_data->is_usable();
306
307	return true;
308	}
309
310	/**
311	* Build and save sitemap to cache.
312	*
313	* @param string $type Sitemap type.
314	* @param int $page_number The page number to save to.
315	*
316	* @return bool
317	*/
318	private function refresh_sitemap_cache( $type, $page_number ) {
319	$this->set_n( $page_number );
320	$this->build_sitemap( $type );
321
322	return $this->cache->store_sitemap( $type, $page_number, $this->sitemap, ! $this->bad_sitemap );
323	}
324
325	/**
326	* Attempts to build the requested sitemap.
327	*
328	* Sets $bad_sitemap if this isn't for the root sitemap, a post type or taxonomy.
329	*
330	* @param string $type The requested sitemap's identifier.
331	*/
332	public function build_sitemap( $type ) {
333
334	/**
335	* Filter the type of sitemap to build.
336	*
337	* @param string $type Sitemap type, determined by the request.
338	*/
339	$type = apply_filters( 'wpseo_build_sitemap_post_type', $type );
340
341	if ( $type === '1' ) {
342	$this->build_root_map();
343
344	return;
345	}
346
347	$entries_per_page = $this->get_entries_per_page();
348
349	foreach ( $this->providers as $provider ) {
350	if ( ! $provider->handles_type( $type ) ) {
351	continue;
352	}
353
354	try {
355	$links = $provider->get_sitemap_links( $type, $entries_per_page, $this->current_page );
356	} catch ( OutOfBoundsException $exception ) {
357	$this->bad_sitemap = true;
358
359	return;
360	}
361
362	$this->sitemap = $this->renderer->get_sitemap( $links, $type, $this->current_page );
363
364	return;
365	}
366
367	if ( has_action( 'wpseo_do_sitemap_' . $type ) ) {
368	/**
369	* Fires custom handler, if hooked to generate sitemap for the type.
370	*/
371	do_action( 'wpseo_do_sitemap_' . $type );
372
373	return;
374	}
375
376	$this->bad_sitemap = true;
377	}
378
379	/**
380	* Build the root sitemap (example.com/sitemap_index.xml) which lists sub-sitemaps for other content types.
381	*/
382	public function build_root_map() {
383
384	$links = [];
385	$entries_per_page = $this->get_entries_per_page();
386
387	foreach ( $this->providers as $provider ) {
388	$links = array_merge( $links, $provider->get_index_links( $entries_per_page ) );
389	}
390
391	/**
392	* Filter the sitemap links array before the index sitemap is built.
393	*
394	* @param array $links Array of sitemap links
395	*/
396	$links = apply_filters( 'wpseo_sitemap_index_links', $links );
397
398	if ( empty( $links ) ) {
399	$this->bad_sitemap = true;
400	$this->sitemap = '';
401
402	return;
403	}
404
405	$this->sitemap = $this->renderer->get_index( $links );
406	}
407
408	/**
409	* Spits out the XSL for the XML sitemap.
410	*
411	* @since 1.4.13
412	*
413	* @param string $type Type to output.
414	*/
415	public function xsl_output( $type ) {
416
417	if ( $type !== 'main' ) {
418
419	/**
420	* Fires for the output of XSL for XML sitemaps, other than type "main".
421	*/
422	do_action( 'wpseo_xsl_' . $type );
423
424	return;
425	}
426
427	header( $this->http_protocol . ' 200 OK', true, 200 );
428	// Prevent the search engines from indexing the XML Sitemap.
429	header( 'X-Robots-Tag: noindex, follow', true );
430	header( 'Content-Type: text/xml' );
431
432	// Make the browser cache this file properly.
433	$expires = YEAR_IN_SECONDS;
434	header( 'Pragma: public' );
435	header( 'Cache-Control: max-age=' . $expires );
436	header( 'Expires: ' . YoastSEO()->helpers->date->format_timestamp( ( time() + $expires ), 'D, d M Y H:i:s' ) . ' GMT' );
437
438	// Don't use WP_Filesystem() here because that's not initialized yet. See https://yoast.atlassian.net/browse/QAK-2043.
439	readfile( WPSEO_PATH . 'css/main-sitemap.xsl' );
440	}
441
442	/**
443	* Spit out the generated sitemap.
444	*/
445	public function output() {
446	$this->send_headers();
447	// phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- Escaping sitemap as either xml or html results in empty document.
448	echo $this->renderer->get_output( $this->sitemap );
449	}
450
451	/**
452	* Makes a request to the sitemap index to cache it before the arrival of the search engines.
453	*
454	* @return void
455	*/
456	public function hit_sitemap_index() {
457	if ( ! $this->cache->is_enabled() ) {
458	return;
459	}
460
461	wp_remote_get( WPSEO_Sitemaps_Router::get_base_url( 'sitemap_index.xml' ) );
462	}
463
464	/**
465	* Get the GMT modification date for the last modified post in the post type.
466	*
467	* @since 3.2
468	*
469	* @param string\|array $post_types Post type or array of types.
470	* @param bool $return_all Flag to return array of values.
471	*
472	* @return string\|array\|false
473	*/
474	public static function get_last_modified_gmt( $post_types, $return_all = false ) {
475
476	global $wpdb;
477
478	static $post_type_dates = null;
479
480	if ( ! is_array( $post_types ) ) {
481	$post_types = [ $post_types ];
482	}
483
484	foreach ( $post_types as $post_type ) {
485	if ( ! isset( $post_type_dates[ $post_type ] ) ) { // If we hadn't seen post type before. R.
486	$post_type_dates = null;
487	break;
488	}
489	}
490
491	if ( is_null( $post_type_dates ) ) {
492
493	$post_type_dates = [];
494	$post_type_names = WPSEO_Post_Type::get_accessible_post_types();
495
496	if ( ! empty( $post_type_names ) ) {
497	$post_statuses = array_map( 'esc_sql', self::get_post_statuses() );
498
499	$sql = "
500	SELECT post_type, MAX(post_modified_gmt) AS date
501	FROM $wpdb->posts
502	WHERE post_status IN ('" . implode( "','", $post_statuses ) . "')
503	AND post_type IN ('" . implode( "','", $post_type_names ) . "')
504	GROUP BY post_type
505	ORDER BY date DESC
506	";
507
508	foreach ( $wpdb->get_results( $sql ) as $obj ) {
509	$post_type_dates[ $obj->post_type ] = $obj->date;
510	}
511	}
512	}
513
514	$dates = array_intersect_key( $post_type_dates, array_flip( $post_types ) );
515
516	if ( count( $dates ) > 0 ) {
517	if ( $return_all ) {
518	return $dates;
519	}
520
521	return max( $dates );
522	}
523
524	return false;
525	}
526
527	/**
528	* Get the modification date for the last modified post in the post type.
529	*
530	* @param array $post_types Post types to get the last modification date for.
531	*
532	* @return string
533	*/
534	public function get_last_modified( $post_types ) {
535	return YoastSEO()->helpers->date->format( self::get_last_modified_gmt( $post_types ) );
536	}
537
538	/**
539	* Notify search engines of the updated sitemap.
540	*
541	* @param string\|null $url Optional URL to make the ping for.
542	*/
543	public static function ping_search_engines( $url = null ) {
544
545	/**
546	* Filter: 'wpseo_allow_xml_sitemap_ping' - Check if pinging is not allowed (allowed by default)
547	*
548	* @api boolean $allow_ping The boolean that is set to true by default.
549	*/
550	if ( apply_filters( 'wpseo_allow_xml_sitemap_ping', true ) === false ) {
551	return;
552	}
553
554	if ( get_option( 'blog_public' ) === '0' ) { // Don't ping if blog is not public.
555	return;
556	}
557
558	if ( empty( $url ) ) {
559	$url = rawurlencode( WPSEO_Sitemaps_Router::get_base_url( 'sitemap_index.xml' ) );
560	}
561
562	// Ping Google and Bing.
563	wp_remote_get( 'https://www.google.com/ping?sitemap=' . $url, [ 'blocking' => false ] );
564	wp_remote_get( 'https://www.bing.com/ping?sitemap=' . $url, [ 'blocking' => false ] );
565	}
566
567	/**
568	* Get the maximum number of entries per XML sitemap.
569	*
570	* @return int The maximum number of entries.
571	*/
572	protected function get_entries_per_page() {
573	/**
574	* Filter the maximum number of entries per XML sitemap.
575	*
576	* After changing the output of the filter, make sure that you disable and enable the
577	* sitemaps to make sure the value is picked up for the sitemap cache.
578	*
579	* @param int $entries The maximum number of entries per XML sitemap.
580	*/
581	$entries = (int) apply_filters( 'wpseo_sitemap_entries_per_page', 1000 );
582
583	return $entries;
584	}
585
586	/**
587	* Get post statuses for post_type or the root sitemap.
588	*
589	* @since 10.2
590	*
591	* @param string $type Provide a type for a post_type sitemap, SITEMAP_INDEX_TYPE for the root sitemap.
592	*
593	* @return array List of post statuses.
594	*/
595	public static function get_post_statuses( $type = self::SITEMAP_INDEX_TYPE ) {
596	/**
597	* Filter post status list for sitemap query for the post type.
598	*
599	* @param array $post_statuses Post status list, defaults to array( 'publish' ).
600	* @param string $type Post type or SITEMAP_INDEX_TYPE.
601	*/
602	$post_statuses = apply_filters( 'wpseo_sitemap_post_statuses', [ 'publish' ], $type );
603
604	if ( ! is_array( $post_statuses ) \|\| empty( $post_statuses ) ) {
605	$post_statuses = [ 'publish' ];
606	}
607
608	if ( ( $type === self::SITEMAP_INDEX_TYPE \|\| $type === 'attachment' )
609	&& ! in_array( 'inherit', $post_statuses, true )
610	) {
611	$post_statuses[] = 'inherit';
612	}
613
614	return $post_statuses;
615	}
616
617	/**
618	* Sends all the required HTTP Headers.
619	*/
620	private function send_headers() {
621	if ( headers_sent() ) {
622	return;
623	}
624
625	$headers = [
626	$this->http_protocol . ' 200 OK' => 200,
627	// Prevent the search engines from indexing the XML Sitemap.
628	'X-Robots-Tag: noindex, follow' => '',
629	'Content-Type: text/xml; charset=' . esc_attr( $this->renderer->get_output_charset() ) => '',
630	];
631
632	/**
633	* Filter the HTTP headers we send before an XML sitemap.
634	*
635	* @param array $headers The HTTP headers we're going to send out.
636	*/
637	$headers = apply_filters( 'wpseo_sitemap_http_headers', $headers );
638
639	foreach ( $headers as $header => $status ) {
640	if ( is_numeric( $status ) ) {
641	header( $header, true, $status );
642	continue;
643	}
644	header( $header, true );
645	}
646	}
647	}
648