wordpress-seo / inc / sitemaps / class-post-type-sitemap-provider.php

class-post-type-sitemap-provider.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 28.4, at inc/sitemaps/class-post-type-sitemap-provider.php

785 lines 20.0 KB

Raw Download Zip

1	<?php
2	/**
3	* WPSEO plugin file.
4	*
5	* @package WPSEO\XML_Sitemaps
6	*/
7
8	use Yoast\WP\SEO\Models\SEO_Links;
9
10	/**
11	* Sitemap provider for author archives.
12	*/
13	class WPSEO_Post_Type_Sitemap_Provider implements WPSEO_Sitemap_Provider {
14
15	/**
16	* Holds image parser instance.
17	*
18	* @var WPSEO_Sitemap_Image_Parser
19	*/
20	protected static $image_parser;
21
22	/**
23	* Holds the parsed home url.
24	*
25	* @var array
26	*/
27	protected static $parsed_home_url;
28
29	/**
30	* Determines whether images should be included in the XML sitemap.
31	*
32	* @var bool
33	*/
34	private $include_images;
35
36	/**
37	* Set up object properties for data reuse.
38	*/
39	public function __construct() {
40	add_action( 'save_post', [ $this, 'save_post' ] );
41
42	/**
43	* Filter - Allows excluding images from the XML sitemap.
44	*
45	* @param bool $include True to include, false to exclude.
46	*/
47	$this->include_images = apply_filters( 'wpseo_xml_sitemap_include_images', true );
48	}
49
50	/**
51	* Get the Image Parser.
52	*
53	* @return WPSEO_Sitemap_Image_Parser
54	*/
55	protected function get_image_parser() {
56	if ( ! isset( self::$image_parser ) ) {
57	self::$image_parser = new WPSEO_Sitemap_Image_Parser();
58	}
59
60	return self::$image_parser;
61	}
62
63	/**
64	* Gets the parsed home url.
65	*
66	* @return array The home url, as parsed by wp_parse_url.
67	*/
68	protected function get_parsed_home_url() {
69	if ( ! isset( self::$parsed_home_url ) ) {
70	self::$parsed_home_url = wp_parse_url( home_url() );
71	}
72
73	return self::$parsed_home_url;
74	}
75
76	/**
77	* Check if provider supports given item type.
78	*
79	* @param string $type Type string to check for.
80	*
81	* @return bool
82	*/
83	public function handles_type( $type ) {
84
85	return post_type_exists( $type );
86	}
87
88	/**
89	* Retrieves the sitemap links.
90	*
91	* @param int $max_entries Entries per sitemap.
92	*
93	* @return array
94	*/
95	public function get_index_links( $max_entries ) {
96	global $wpdb;
97	$post_types = WPSEO_Post_Type::get_accessible_post_types();
98	$post_types = array_filter( $post_types, [ $this, 'is_valid_post_type' ] );
99	$last_modified_times = WPSEO_Sitemaps::get_last_modified_gmt( $post_types, true );
100	$index = [];
101
102	foreach ( $post_types as $post_type ) {
103
104	$total_count = $this->get_post_type_count( $post_type );
105
106	if ( $total_count === 0 ) {
107	continue;
108	}
109
110	$max_pages = 1;
111	if ( $total_count > $max_entries ) {
112	$max_pages = (int) ceil( $total_count / $max_entries );
113	}
114
115	$all_dates = [];
116
117	if ( $max_pages > 1 ) {
118	$all_dates = version_compare( $wpdb->db_version(), '8.0', '>=' ) ? $this->get_all_dates_using_with_clause( $post_type, $max_entries ) : $this->get_all_dates( $post_type, $max_entries );
119	}
120
121	for ( $page_counter = 0; $page_counter < $max_pages; $page_counter++ ) {
122
123	$current_page = ( $page_counter === 0 ) ? '' : ( $page_counter + 1 );
124	$date = false;
125
126	if ( empty( $current_page ) \|\| $current_page === $max_pages ) {
127
128	if ( ! empty( $last_modified_times[ $post_type ] ) ) {
129	$date = $last_modified_times[ $post_type ];
130	}
131	}
132	else {
133	$date = $all_dates[ $page_counter ];
134	}
135
136	$index[] = [
137	'loc' => WPSEO_Sitemaps_Router::get_base_url( $post_type . '-sitemap' . $current_page . '.xml' ),
138	'lastmod' => $date,
139	];
140	}
141	}
142
143	return $index;
144	}
145
146	/**
147	* Get set of sitemap link data.
148	*
149	* @param string $type Sitemap type.
150	* @param int $max_entries Entries per sitemap.
151	* @param int $current_page Current page of the sitemap.
152	*
153	* @return array
154	*
155	* @throws OutOfBoundsException When an invalid page is requested.
156	*/
157	public function get_sitemap_links( $type, $max_entries, $current_page ) {
158
159	$links = [];
160	$post_type = $type;
161
162	if ( ! $this->is_valid_post_type( $post_type ) ) {
163	throw new OutOfBoundsException( 'Invalid sitemap page requested' );
164	}
165
166	$steps = min( 100, $max_entries );
167	$offset = ( $current_page > 1 ) ? ( ( $current_page - 1 ) * $max_entries ) : 0;
168	$total = ( $offset + $max_entries );
169
170	$post_type_entries = $this->get_post_type_count( $post_type );
171
172	if ( $total > $post_type_entries ) {
173	$total = $post_type_entries;
174	}
175
176	if ( $current_page === 1 ) {
177	$links = array_merge( $links, $this->get_first_links( $post_type ) );
178	}
179
180	// If total post type count is lower than the offset, an invalid page is requested.
181	if ( $post_type_entries < $offset ) {
182	throw new OutOfBoundsException( 'Invalid sitemap page requested' );
183	}
184
185	if ( $post_type_entries === 0 ) {
186	return $links;
187	}
188
189	$posts_to_exclude = $this->get_excluded_posts( $type );
190
191	while ( $total > $offset ) {
192
193	$posts = $this->get_posts( $post_type, $steps, $offset );
194
195	$offset += $steps;
196
197	if ( empty( $posts ) ) {
198	continue;
199	}
200
201	foreach ( $posts as $post ) {
202
203	if ( in_array( $post->ID, $posts_to_exclude, true ) ) {
204	continue;
205	}
206
207	if ( WPSEO_Meta::get_value( 'meta-robots-noindex', $post->ID ) === '1' ) {
208	continue;
209	}
210
211	$url = $this->get_url( $post );
212
213	if ( ! isset( $url['loc'] ) ) {
214	continue;
215	}
216
217	/**
218	* Filter URL entry before it gets added to the sitemap.
219	*
220	* @param array $url Array of URL parts.
221	* @param string $type URL type.
222	* @param object $post Data object for the URL.
223	*/
224	$url = apply_filters( 'wpseo_sitemap_entry', $url, 'post', $post );
225
226	if ( ! empty( $url ) ) {
227	$links[] = $url;
228	}
229	}
230
231	unset( $post, $url );
232	}
233
234	return $links;
235	}
236
237	/**
238	* Check for relevant post type before invalidation.
239	*
240	* @param int $post_id Post ID to possibly invalidate for.
241	*
242	* @return void
243	*/
244	public function save_post( $post_id ) {
245
246	if ( $this->is_valid_post_type( get_post_type( $post_id ) ) ) {
247	WPSEO_Sitemaps_Cache::invalidate_post( $post_id );
248	}
249	}
250
251	/**
252	* Check if post type should be present in sitemaps.
253	*
254	* @param string $post_type Post type string to check for.
255	*
256	* @return bool
257	*/
258	public function is_valid_post_type( $post_type ) {
259	if ( ! WPSEO_Post_Type::is_post_type_accessible( $post_type ) \|\| ! WPSEO_Post_Type::is_post_type_indexable( $post_type ) ) {
260	return false;
261	}
262
263	/**
264	* Filter decision if post type is excluded from the XML sitemap.
265	*
266	* @param bool $exclude Default false.
267	* @param string $post_type Post type name.
268	*/
269	if ( apply_filters( 'wpseo_sitemap_exclude_post_type', false, $post_type ) ) {
270	return false;
271	}
272
273	return true;
274	}
275
276	/**
277	* Retrieves a list with the excluded post ids.
278	*
279	* @param string $post_type Post type.
280	*
281	* @return array Array with post ids to exclude.
282	*/
283	protected function get_excluded_posts( $post_type ) {
284	$excluded_posts_ids = [];
285
286	$page_on_front_id = ( $post_type === 'page' ) ? (int) get_option( 'page_on_front' ) : 0;
287	if ( $page_on_front_id > 0 ) {
288	$excluded_posts_ids[] = $page_on_front_id;
289	}
290
291	/**
292	* Filter: 'wpseo_exclude_from_sitemap_by_post_ids' - Allow extending and modifying the posts to exclude.
293	*
294	* @param array $posts_to_exclude The posts to exclude.
295	*/
296	$excluded_posts_ids = apply_filters( 'wpseo_exclude_from_sitemap_by_post_ids', $excluded_posts_ids );
297	if ( ! is_array( $excluded_posts_ids ) ) {
298	$excluded_posts_ids = [];
299	}
300
301	$excluded_posts_ids = array_map( 'intval', $excluded_posts_ids );
302
303	$page_for_posts_id = ( $post_type === 'page' ) ? (int) get_option( 'page_for_posts' ) : 0;
304	if ( $page_for_posts_id > 0 ) {
305	$excluded_posts_ids[] = $page_for_posts_id;
306	}
307
308	return array_unique( $excluded_posts_ids );
309	}
310
311	/**
312	* Get count of posts for post type.
313	*
314	* @param string $post_type Post type to retrieve count for.
315	*
316	* @return int
317	*/
318	protected function get_post_type_count( $post_type ) {
319
320	global $wpdb;
321
322	/**
323	* Filter JOIN query part for type count of post type.
324	*
325	* @param string $join SQL part, defaults to empty string.
326	* @param string $post_type Post type name.
327	*/
328	$join_filter = apply_filters( 'wpseo_typecount_join', '', $post_type );
329
330	/**
331	* Filter WHERE query part for type count of post type.
332	*
333	* @param string $where SQL part, defaults to empty string.
334	* @param string $post_type Post type name.
335	*/
336	$where_filter = apply_filters( 'wpseo_typecount_where', '', $post_type );
337
338	$where = $this->get_sql_where_clause( $post_type );
339
340	$sql = "
341	SELECT COUNT({$wpdb->posts}.ID)
342	FROM {$wpdb->posts}
343	{$join_filter}
344	{$where}
345	{$where_filter}
346	";
347
348	return (int) $wpdb->get_var( $sql );
349	}
350
351	/**
352	* Produces set of links to prepend at start of first sitemap page.
353	*
354	* @param string $post_type Post type to produce links for.
355	*
356	* @return array
357	*/
358	protected function get_first_links( $post_type ) {
359
360	$links = [];
361	$archive_url = false;
362
363	if ( $post_type === 'page' ) {
364
365	$page_on_front_id = (int) get_option( 'page_on_front' );
366	if ( $page_on_front_id > 0 ) {
367	$front_page = $this->get_url(
368	get_post( $page_on_front_id ),
369	);
370	}
371
372	if ( empty( $front_page ) ) {
373	$front_page = [
374	'loc' => YoastSEO()->helpers->url->home(),
375	];
376	}
377
378	// Deprecated, kept for backwards data compat. R.
379	$front_page['chf'] = 'daily';
380	$front_page['pri'] = 1;
381
382	$images = ( $front_page['images'] ?? [] );
383
384	/**
385	* Filter images to be included for the term in XML sitemap.
386	*
387	* @param array $images Array of image items.
388	* @return array $image_list Array of image items.
389	*/
390	$image_list = apply_filters( 'wpseo_sitemap_urlimages_front_page', $images );
391	if ( is_array( $image_list ) ) {
392	$front_page['images'] = $image_list;
393	}
394
395	$links[] = $front_page;
396	}
397	elseif ( $post_type !== 'page' ) {
398	/**
399	* Filter the URL Yoast SEO uses in the XML sitemap for this post type archive.
400	*
401	* @param string $archive_url The URL of this archive
402	* @param string $post_type The post type this archive is for.
403	*/
404	$archive_url = apply_filters(
405	'wpseo_sitemap_post_type_archive_link',
406	$this->get_post_type_archive_link( $post_type ),
407	$post_type,
408	);
409	}
410
411	if ( $archive_url ) {
412
413	$links[] = [
414	'loc' => $archive_url,
415	'mod' => WPSEO_Sitemaps::get_last_modified_gmt( $post_type ),
416
417	// Deprecated, kept for backwards data compat. R.
418	'chf' => 'daily',
419	'pri' => 1,
420	];
421	}
422
423	/**
424	* Filters the first post type links.
425	*
426	* @param array $links The first post type links.
427	* @param string $post_type The post type this archive is for.
428	*/
429	return apply_filters( 'wpseo_sitemap_post_type_first_links', $links, $post_type );
430	}
431
432	/**
433	* Get URL for a post type archive.
434	*
435	* @since 5.3
436	*
437	* @param string $post_type Post type.
438	*
439	* @return string\|bool URL or false if it should be excluded.
440	*/
441	protected function get_post_type_archive_link( $post_type ) {
442
443	$pt_archive_page_id = -1;
444
445	if ( $post_type === 'post' ) {
446
447	if ( get_option( 'show_on_front' ) === 'posts' ) {
448	return YoastSEO()->helpers->url->home();
449	}
450
451	$pt_archive_page_id = (int) get_option( 'page_for_posts' );
452
453	// Post archive should be excluded if posts page isn't set.
454	if ( $pt_archive_page_id <= 0 ) {
455	return false;
456	}
457	}
458
459	if ( ! $this->is_post_type_archive_indexable( $post_type, $pt_archive_page_id ) ) {
460	return false;
461	}
462
463	return get_post_type_archive_link( $post_type );
464	}
465
466	/**
467	* Determines whether a post type archive is indexable.
468	*
469	* @since 11.5
470	*
471	* @param string $post_type Post type.
472	* @param int $archive_page_id The page id.
473	*
474	* @return bool True when post type archive is indexable.
475	*/
476	protected function is_post_type_archive_indexable( $post_type, $archive_page_id = -1 ) {
477
478	if ( WPSEO_Options::get( 'noindex-ptarchive-' . $post_type, false ) ) {
479	return false;
480	}
481
482	/**
483	* Filter the page which is dedicated to this post type archive.
484	*
485	* @since 9.3
486	*
487	* @param string $archive_page_id The post_id of the page.
488	* @param string $post_type The post type this archive is for.
489	*/
490	$archive_page_id = (int) apply_filters( 'wpseo_sitemap_page_for_post_type_archive', $archive_page_id, $post_type );
491
492	if ( $archive_page_id > 0 && WPSEO_Meta::get_value( 'meta-robots-noindex', $archive_page_id ) === '1' ) {
493	return false;
494	}
495
496	return true;
497	}
498
499	/**
500	* Retrieve set of posts with optimized query routine.
501	*
502	* @param string $post_type Post type to retrieve.
503	* @param int $count Count of posts to retrieve.
504	* @param int $offset Starting offset.
505	*
506	* @return object[]
507	*/
508	protected function get_posts( $post_type, $count, $offset ) {
509
510	global $wpdb;
511
512	static $filters = [];
513
514	if ( ! isset( $filters[ $post_type ] ) ) {
515	// Make sure you're wpdb->preparing everything you throw into this!!
516	$filters[ $post_type ] = [
517	/**
518	* Filter JOIN query part for the post type.
519	*
520	* @param string $join SQL part, defaults to false.
521	* @param string $post_type Post type name.
522	*/
523	'join' => apply_filters( 'wpseo_posts_join', false, $post_type ),
524
525	/**
526	* Filter WHERE query part for the post type.
527	*
528	* @param string $where SQL part, defaults to false.
529	* @param string $post_type Post type name.
530	*/
531	'where' => apply_filters( 'wpseo_posts_where', false, $post_type ),
532	];
533	}
534
535	$join_filter = $filters[ $post_type ]['join'];
536	$where_filter = $filters[ $post_type ]['where'];
537	$where = $this->get_sql_where_clause( $post_type );
538
539	/*
540	* Optimized query per this thread:
541	* {@link http://wordpress.org/support/topic/plugin-wordpress-seo-by-yoast-performance-suggestion}.
542	* Also see {@link http://explainextended.com/2009/10/23/mysql-order-by-limit-performance-late-row-lookups/}.
543	*/
544	$sql = "
545	SELECT l.ID, post_title, post_content, post_name, post_parent, post_author, post_status, post_modified_gmt, post_date, post_date_gmt
546	FROM (
547	SELECT {$wpdb->posts}.ID
548	FROM {$wpdb->posts}
549	{$join_filter}
550	{$where}
551	{$where_filter}
552	ORDER BY {$wpdb->posts}.post_modified ASC LIMIT %d OFFSET %d
553	)
554	o JOIN {$wpdb->posts} l ON l.ID = o.ID
555	";
556
557	$posts = $wpdb->get_results( $wpdb->prepare( $sql, $count, $offset ) );
558
559	$post_ids = [];
560
561	foreach ( $posts as $post_index => $post ) {
562	$post->post_type = $post_type;
563	$sanitized_post = sanitize_post( $post, 'raw' );
564	$posts[ $post_index ] = new WP_Post( $sanitized_post );
565
566	$post_ids[] = $sanitized_post->ID;
567	}
568
569	/**
570	* Filter to disable priming the post, term and featured-image caches for the sitemap.
571	*
572	* @since 28.3
573	*
574	* @param bool $disable_cache_priming Whether to disable priming the caches. Defaults to false.
575	*/
576	$disable_priming = apply_filters( 'wpseo_disable_xml_sitemap_cache_priming', false );
577
578	if ( ! $disable_priming ) {
579	// Warm the post and term caches in bulk, so permalink and image building doesn't query per post.
580	_prime_post_caches( $post_ids, true, false );
581	}
582
583	update_meta_cache( 'post', $post_ids );
584
585	if ( ! $disable_priming && $this->include_images ) {
586	$this->get_image_parser()->prime_thumbnail_caches( $posts );
587	}
588
589	return $posts;
590	}
591
592	/**
593	* Constructs an SQL where clause for a given post type.
594	*
595	* @param string $post_type Post type slug.
596	*
597	* @return string
598	*/
599	protected function get_sql_where_clause( $post_type ) {
600
601	global $wpdb;
602
603	$join = '';
604	$post_statuses = array_map( 'esc_sql', WPSEO_Sitemaps::get_post_statuses( $post_type ) );
605	$status_where = "{$wpdb->posts}.post_status IN ('" . implode( "','", $post_statuses ) . "')";
606
607	// Based on WP_Query->get_posts(). R.
608	if ( $post_type === 'attachment' ) {
609	$join = " LEFT JOIN {$wpdb->posts} AS p2 ON ({$wpdb->posts}.post_parent = p2.ID) ";
610	$parent_statuses = array_diff( $post_statuses, [ 'inherit' ] );
611	$status_where = "p2.post_status IN ('" . implode( "','", $parent_statuses ) . "') AND p2.post_password = ''";
612	}
613
614	$where_clause = "
615	{$join}
616	WHERE {$status_where}
617	AND {$wpdb->posts}.post_type = %s
618	AND {$wpdb->posts}.post_password = ''
619	AND {$wpdb->posts}.post_date != '0000-00-00 00:00:00'
620	";
621
622	return $wpdb->prepare( $where_clause, $post_type );
623	}
624
625	/**
626	* Produce array of URL parts for given post object.
627	*
628	* @param object $post Post object to get URL parts for.
629	*
630	* @return array\|bool
631	*/
632	protected function get_url( $post ) {
633
634	$url = [];
635
636	/**
637	* Filter the URL Yoast SEO uses in the XML sitemap.
638	*
639	* Note that only absolute local URLs are allowed as the check after this removes external URLs.
640	*
641	* @param string $url URL to use in the XML sitemap
642	* @param object $post Post object for the URL.
643	*/
644	$url['loc'] = apply_filters( 'wpseo_xml_sitemap_post_url', get_permalink( $post ), $post );
645	$link_type = YoastSEO()->helpers->url->get_link_type(
646	wp_parse_url( $url['loc'] ),
647	$this->get_parsed_home_url(),
648	);
649
650	/*
651	* Do not include external URLs.
652	*
653	* {@link https://wordpress.org/plugins/page-links-to/} can rewrite permalinks to external URLs.
654	*/
655	if ( $link_type === SEO_Links::TYPE_EXTERNAL ) {
656	return false;
657	}
658
659	$modified = max( $post->post_modified_gmt, $post->post_date_gmt );
660
661	if ( $modified !== '0000-00-00 00:00:00' ) {
662	$url['mod'] = $modified;
663	}
664
665	$url['chf'] = 'daily'; // Deprecated, kept for backwards data compat. R.
666
667	$canonical = WPSEO_Meta::get_value( 'canonical', $post->ID );
668
669	if ( $canonical !== '' && $canonical !== $url['loc'] ) {
670	/*
671	* Let's assume that if a canonical is set for this page and it's different from
672	* the URL of this post, that page is either already in the XML sitemap OR is on
673	* an external site, either way, we shouldn't include it here.
674	*/
675	return false;
676	}
677	unset( $canonical );
678
679	$url['pri'] = 1; // Deprecated, kept for backwards data compat. R.
680
681	if ( $this->include_images ) {
682	$url['images'] = $this->get_image_parser()->get_images( $post );
683	}
684
685	return $url;
686	}
687
688	/**
689	* Get all dates for a post type by using the WITH clause for performance.
690	*
691	* @param string $post_type Post type to retrieve dates for.
692	* @param int $max_entries Maximum number of entries to retrieve.
693	*
694	* @return array Array of dates.
695	*/
696	private function get_all_dates_using_with_clause( $post_type, $max_entries ) {
697	global $wpdb;
698
699	$post_statuses = array_map( 'esc_sql', WPSEO_Sitemaps::get_post_statuses( $post_type ) );
700
701	$replacements = array_merge(
702	[
703	'ordering',
704	'post_modified_gmt',
705	$wpdb->posts,
706	'type_status_date',
707	'post_status',
708	],
709	$post_statuses,
710	[
711	'post_type',
712	$post_type,
713	'post_modified_gmt',
714	'post_modified_gmt',
715	'ordering',
716	$max_entries,
717	],
718	);
719
720	//phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- We need to use a direct query here.
721	//phpcs:disable WordPress.DB.DirectDatabaseQuery.NoCaching -- Reason: No relevant caches.
722	return $wpdb->get_col(
723	//phpcs:disable WordPress.DB.PreparedSQLPlaceholders -- %i placeholder is still not recognized.
724	$wpdb->prepare(
725	'
726	WITH %i AS (SELECT ROW_NUMBER() OVER (ORDER BY %i) AS n, post_modified_gmt
727	FROM %i USE INDEX ( %i )
728	WHERE %i IN (' . implode( ', ', array_fill( 0, count( $post_statuses ), '%s' ) ) . ')
729	AND %i = %s
730	ORDER BY %i)
731	SELECT %i
732	FROM %i
733	WHERE MOD(n, %d) = 0;
734	',
735	$replacements,
736	),
737	);
738	}
739
740	/**
741	* Get all dates for a post type.
742	*
743	* @param string $post_type Post type to retrieve dates for.
744	* @param int $max_entries Maximum number of entries to retrieve.
745	*
746	* @return array Array of dates.
747	*/
748	private function get_all_dates( $post_type, $max_entries ) {
749	global $wpdb;
750
751	$post_statuses = array_map( 'esc_sql', WPSEO_Sitemaps::get_post_statuses( $post_type ) );
752	$replacements = array_merge(
753	[
754	'post_modified_gmt',
755	$wpdb->posts,
756	'type_status_date',
757	'post_status',
758	],
759	$post_statuses,
760	[
761	'post_type',
762	$post_type,
763	$max_entries,
764	'post_modified_gmt',
765	],
766	);
767
768	return $wpdb->get_col(
769	//phpcs:disable WordPress.DB.PreparedSQLPlaceholders -- %i placeholder is still not recognized.
770	$wpdb->prepare(
771	'
772	SELECT %i
773	FROM ( SELECT @rownum:=0 ) init
774	JOIN %i USE INDEX( %i )
775	WHERE %i IN (' . implode( ', ', array_fill( 0, count( $post_statuses ), '%s' ) ) . ')
776	AND %i = %s
777	AND ( @rownum:=@rownum+1 ) %% %d = 0
778	ORDER BY %i ASC
779	',
780	$replacements,
781	),
782	);
783	}
784	}
785