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 18.6, at inc/sitemaps/class-post-type-sitemap-provider.php

658 lines 16.5 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_filter( '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
98	$post_types = WPSEO_Post_Type::get_accessible_post_types();
99	$post_types = array_filter( $post_types, [ $this, 'is_valid_post_type' ] );
100	$last_modified_times = WPSEO_Sitemaps::get_last_modified_gmt( $post_types, true );
101	$index = [];
102
103	foreach ( $post_types as $post_type ) {
104
105	$total_count = $this->get_post_type_count( $post_type );
106
107	$max_pages = 1;
108	if ( $total_count > $max_entries ) {
109	$max_pages = (int) ceil( $total_count / $max_entries );
110	}
111
112	$all_dates = [];
113
114	if ( $max_pages > 1 ) {
115	$post_statuses = array_map( 'esc_sql', WPSEO_Sitemaps::get_post_statuses( $post_type ) );
116
117	$sql = "
118	SELECT post_modified_gmt
119	FROM ( SELECT @rownum:=0 ) init
120	JOIN {$wpdb->posts} USE INDEX( type_status_date )
121	WHERE post_status IN ('" . implode( "','", $post_statuses ) . "')
122	AND post_type = %s
123	AND ( @rownum:=@rownum+1 ) %% %d = 0
124	ORDER BY post_modified_gmt ASC
125	";
126
127	$all_dates = $wpdb->get_col( $wpdb->prepare( $sql, $post_type, $max_entries ) );
128	}
129
130	for ( $page_counter = 0; $page_counter < $max_pages; $page_counter++ ) {
131
132	$current_page = ( $page_counter === 0 ) ? '' : ( $page_counter + 1 );
133	$date = false;
134
135	if ( empty( $current_page ) \|\| $current_page === $max_pages ) {
136
137	if ( ! empty( $last_modified_times[ $post_type ] ) ) {
138	$date = $last_modified_times[ $post_type ];
139	}
140	}
141	else {
142	$date = $all_dates[ $page_counter ];
143	}
144
145	$index[] = [
146	'loc' => WPSEO_Sitemaps_Router::get_base_url( $post_type . '-sitemap' . $current_page . '.xml' ),
147	'lastmod' => $date,
148	];
149	}
150	}
151
152	return $index;
153	}
154
155	/**
156	* Get set of sitemap link data.
157	*
158	* @param string $type Sitemap type.
159	* @param int $max_entries Entries per sitemap.
160	* @param int $current_page Current page of the sitemap.
161	*
162	* @return array
163	*
164	* @throws OutOfBoundsException When an invalid page is requested.
165	*/
166	public function get_sitemap_links( $type, $max_entries, $current_page ) {
167
168	$links = [];
169	$post_type = $type;
170
171	if ( ! $this->is_valid_post_type( $post_type ) ) {
172	throw new OutOfBoundsException( 'Invalid sitemap page requested' );
173	}
174
175	$steps = min( 100, $max_entries );
176	$offset = ( $current_page > 1 ) ? ( ( $current_page - 1 ) * $max_entries ) : 0;
177	$total = ( $offset + $max_entries );
178
179	$post_type_entries = $this->get_post_type_count( $post_type );
180
181	if ( $total > $post_type_entries ) {
182	$total = $post_type_entries;
183	}
184
185	if ( $current_page === 1 ) {
186	$links = array_merge( $links, $this->get_first_links( $post_type ) );
187	}
188
189	// If total post type count is lower than the offset, an invalid page is requested.
190	if ( $post_type_entries < $offset ) {
191	throw new OutOfBoundsException( 'Invalid sitemap page requested' );
192	}
193
194	if ( $post_type_entries === 0 ) {
195	return $links;
196	}
197
198	$posts_to_exclude = $this->get_excluded_posts( $type );
199
200	while ( $total > $offset ) {
201
202	$posts = $this->get_posts( $post_type, $steps, $offset );
203
204	$offset += $steps;
205
206	if ( empty( $posts ) ) {
207	continue;
208	}
209
210	foreach ( $posts as $post ) {
211
212	if ( in_array( $post->ID, $posts_to_exclude, true ) ) {
213	continue;
214	}
215
216	if ( WPSEO_Meta::get_value( 'meta-robots-noindex', $post->ID ) === '1' ) {
217	continue;
218	}
219
220	$url = $this->get_url( $post );
221
222	if ( ! isset( $url['loc'] ) ) {
223	continue;
224	}
225
226	/**
227	* Filter URL entry before it gets added to the sitemap.
228	*
229	* @param array $url Array of URL parts.
230	* @param string $type URL type.
231	* @param object $post Data object for the URL.
232	*/
233	$url = apply_filters( 'wpseo_sitemap_entry', $url, 'post', $post );
234
235	if ( ! empty( $url ) ) {
236	$links[] = $url;
237	}
238	}
239
240	unset( $post, $url );
241	}
242
243	return $links;
244	}
245
246	/**
247	* Check for relevant post type before invalidation.
248	*
249	* @param int $post_id Post ID to possibly invalidate for.
250	*/
251	public function save_post( $post_id ) {
252
253	if ( $this->is_valid_post_type( get_post_type( $post_id ) ) ) {
254	WPSEO_Sitemaps_Cache::invalidate_post( $post_id );
255	}
256	}
257
258	/**
259	* Check if post type should be present in sitemaps.
260	*
261	* @param string $post_type Post type string to check for.
262	*
263	* @return bool
264	*/
265	public function is_valid_post_type( $post_type ) {
266	if ( ! WPSEO_Post_Type::is_post_type_accessible( $post_type ) \|\| ! WPSEO_Post_Type::is_post_type_indexable( $post_type ) ) {
267	return false;
268	}
269
270	/**
271	* Filter decision if post type is excluded from the XML sitemap.
272	*
273	* @param bool $exclude Default false.
274	* @param string $post_type Post type name.
275	*/
276	if ( apply_filters( 'wpseo_sitemap_exclude_post_type', false, $post_type ) ) {
277	return false;
278	}
279
280	return true;
281	}
282
283	/**
284	* Retrieves a list with the excluded post ids.
285	*
286	* @param string $post_type Post type.
287	*
288	* @return array Array with post ids to exclude.
289	*/
290	protected function get_excluded_posts( $post_type ) {
291	$excluded_posts_ids = [];
292
293	$page_on_front_id = ( $post_type === 'page' ) ? (int) get_option( 'page_on_front' ) : 0;
294	if ( $page_on_front_id > 0 ) {
295	$excluded_posts_ids[] = $page_on_front_id;
296	}
297
298	/**
299	* Filter: 'wpseo_exclude_from_sitemap_by_post_ids' - Allow extending and modifying the posts to exclude.
300	*
301	* @api array $posts_to_exclude The posts to exclude.
302	*/
303	$excluded_posts_ids = apply_filters( 'wpseo_exclude_from_sitemap_by_post_ids', $excluded_posts_ids );
304	if ( ! is_array( $excluded_posts_ids ) ) {
305	$excluded_posts_ids = [];
306	}
307
308	$excluded_posts_ids = array_map( 'intval', $excluded_posts_ids );
309
310	$page_for_posts_id = ( $post_type === 'page' ) ? (int) get_option( 'page_for_posts' ) : 0;
311	if ( $page_for_posts_id > 0 ) {
312	$excluded_posts_ids[] = $page_for_posts_id;
313	}
314
315	return array_unique( $excluded_posts_ids );
316	}
317
318	/**
319	* Get count of posts for post type.
320	*
321	* @param string $post_type Post type to retrieve count for.
322	*
323	* @return int
324	*/
325	protected function get_post_type_count( $post_type ) {
326
327	global $wpdb;
328
329	/**
330	* Filter JOIN query part for type count of post type.
331	*
332	* @param string $join SQL part, defaults to empty string.
333	* @param string $post_type Post type name.
334	*/
335	$join_filter = apply_filters( 'wpseo_typecount_join', '', $post_type );
336
337	/**
338	* Filter WHERE query part for type count of post type.
339	*
340	* @param string $where SQL part, defaults to empty string.
341	* @param string $post_type Post type name.
342	*/
343	$where_filter = apply_filters( 'wpseo_typecount_where', '', $post_type );
344
345	$where = $this->get_sql_where_clause( $post_type );
346
347	$sql = "
348	SELECT COUNT({$wpdb->posts}.ID)
349	FROM {$wpdb->posts}
350	{$join_filter}
351	{$where}
352	{$where_filter}
353	";
354
355	return (int) $wpdb->get_var( $sql );
356	}
357
358	/**
359	* Produces set of links to prepend at start of first sitemap page.
360	*
361	* @param string $post_type Post type to produce links for.
362	*
363	* @return array
364	*/
365	protected function get_first_links( $post_type ) {
366
367	$links = [];
368	$archive_url = false;
369
370	if ( $post_type === 'page' ) {
371
372	$page_on_front_id = (int) get_option( 'page_on_front' );
373	if ( $page_on_front_id > 0 ) {
374	$front_page = $this->get_url(
375	get_post( $page_on_front_id )
376	);
377	}
378
379	if ( empty( $front_page ) ) {
380	$front_page = [
381	'loc' => YoastSEO()->helpers->url->home(),
382	];
383	}
384
385	// Deprecated, kept for backwards data compat. R.
386	$front_page['chf'] = 'daily';
387	$front_page['pri'] = 1;
388
389	$links[] = $front_page;
390	}
391	elseif ( $post_type !== 'page' ) {
392	/**
393	* Filter the URL Yoast SEO uses in the XML sitemap for this post type archive.
394	*
395	* @param string $archive_url The URL of this archive
396	* @param string $post_type The post type this archive is for.
397	*/
398	$archive_url = apply_filters(
399	'wpseo_sitemap_post_type_archive_link',
400	$this->get_post_type_archive_link( $post_type ),
401	$post_type
402	);
403	}
404
405	if ( $archive_url ) {
406
407	$links[] = [
408	'loc' => $archive_url,
409	'mod' => WPSEO_Sitemaps::get_last_modified_gmt( $post_type ),
410
411	// Deprecated, kept for backwards data compat. R.
412	'chf' => 'daily',
413	'pri' => 1,
414	];
415	}
416
417	return $links;
418	}
419
420	/**
421	* Get URL for a post type archive.
422	*
423	* @since 5.3
424	*
425	* @param string $post_type Post type.
426	*
427	* @return string\|bool URL or false if it should be excluded.
428	*/
429	protected function get_post_type_archive_link( $post_type ) {
430
431	$pt_archive_page_id = -1;
432
433	if ( $post_type === 'post' ) {
434
435	if ( get_option( 'show_on_front' ) === 'posts' ) {
436	return YoastSEO()->helpers->url->home();
437	}
438
439	$pt_archive_page_id = (int) get_option( 'page_for_posts' );
440
441	// Post archive should be excluded if posts page isn't set.
442	if ( $pt_archive_page_id <= 0 ) {
443	return false;
444	}
445	}
446
447	if ( ! $this->is_post_type_archive_indexable( $post_type, $pt_archive_page_id ) ) {
448	return false;
449	}
450
451	return get_post_type_archive_link( $post_type );
452	}
453
454	/**
455	* Determines whether a post type archive is indexable.
456	*
457	* @since 11.5
458	*
459	* @param string $post_type Post type.
460	* @param int $archive_page_id The page id.
461	*
462	* @return bool True when post type archive is indexable.
463	*/
464	protected function is_post_type_archive_indexable( $post_type, $archive_page_id = -1 ) {
465
466	if ( WPSEO_Options::get( 'noindex-ptarchive-' . $post_type, false ) ) {
467	return false;
468	}
469
470	/**
471	* Filter the page which is dedicated to this post type archive.
472	*
473	* @since 9.3
474	*
475	* @param string $archive_page_id The post_id of the page.
476	* @param string $post_type The post type this archive is for.
477	*/
478	$archive_page_id = (int) apply_filters( 'wpseo_sitemap_page_for_post_type_archive', $archive_page_id, $post_type );
479
480	if ( $archive_page_id > 0 && WPSEO_Meta::get_value( 'meta-robots-noindex', $archive_page_id ) === '1' ) {
481	return false;
482	}
483
484	return true;
485	}
486
487	/**
488	* Retrieve set of posts with optimized query routine.
489	*
490	* @param string $post_type Post type to retrieve.
491	* @param int $count Count of posts to retrieve.
492	* @param int $offset Starting offset.
493	*
494	* @return object[]
495	*/
496	protected function get_posts( $post_type, $count, $offset ) {
497
498	global $wpdb;
499
500	static $filters = [];
501
502	if ( ! isset( $filters[ $post_type ] ) ) {
503	// Make sure you're wpdb->preparing everything you throw into this!!
504	$filters[ $post_type ] = [
505	/**
506	* Filter JOIN query part for the post type.
507	*
508	* @param string $join SQL part, defaults to false.
509	* @param string $post_type Post type name.
510	*/
511	'join' => apply_filters( 'wpseo_posts_join', false, $post_type ),
512
513	/**
514	* Filter WHERE query part for the post type.
515	*
516	* @param string $where SQL part, defaults to false.
517	* @param string $post_type Post type name.
518	*/
519	'where' => apply_filters( 'wpseo_posts_where', false, $post_type ),
520	];
521	}
522
523	$join_filter = $filters[ $post_type ]['join'];
524	$where_filter = $filters[ $post_type ]['where'];
525	$where = $this->get_sql_where_clause( $post_type );
526
527	/*
528	* Optimized query per this thread:
529	* {@link http://wordpress.org/support/topic/plugin-wordpress-seo-by-yoast-performance-suggestion}.
530	* Also see {@link http://explainextended.com/2009/10/23/mysql-order-by-limit-performance-late-row-lookups/}.
531	*/
532	$sql = "
533	SELECT l.ID, post_title, post_content, post_name, post_parent, post_author, post_status, post_modified_gmt, post_date, post_date_gmt
534	FROM (
535	SELECT {$wpdb->posts}.ID
536	FROM {$wpdb->posts}
537	{$join_filter}
538	{$where}
539	{$where_filter}
540	ORDER BY {$wpdb->posts}.post_modified ASC LIMIT %d OFFSET %d
541	)
542	o JOIN {$wpdb->posts} l ON l.ID = o.ID
543	";
544
545	$posts = $wpdb->get_results( $wpdb->prepare( $sql, $count, $offset ) );
546
547	$post_ids = [];
548
549	foreach ( $posts as $post_index => $post ) {
550	$post->post_type = $post_type;
551	$sanitized_post = sanitize_post( $post, 'raw' );
552	$posts[ $post_index ] = new WP_Post( $sanitized_post );
553
554	$post_ids[] = $sanitized_post->ID;
555	}
556
557	update_meta_cache( 'post', $post_ids );
558
559	return $posts;
560	}
561
562	/**
563	* Constructs an SQL where clause for a given post type.
564	*
565	* @param string $post_type Post type slug.
566	*
567	* @return string
568	*/
569	protected function get_sql_where_clause( $post_type ) {
570
571	global $wpdb;
572
573	$join = '';
574	$post_statuses = array_map( 'esc_sql', WPSEO_Sitemaps::get_post_statuses( $post_type ) );
575	$status_where = "{$wpdb->posts}.post_status IN ('" . implode( "','", $post_statuses ) . "')";
576
577	// Based on WP_Query->get_posts(). R.
578	if ( $post_type === 'attachment' ) {
579	$join = " LEFT JOIN {$wpdb->posts} AS p2 ON ({$wpdb->posts}.post_parent = p2.ID) ";
580	$parent_statuses = array_diff( $post_statuses, [ 'inherit' ] );
581	$status_where = "p2.post_status IN ('" . implode( "','", $parent_statuses ) . "') AND p2.post_password = ''";
582	}
583
584	$where_clause = "
585	{$join}
586	WHERE {$status_where}
587	AND {$wpdb->posts}.post_type = %s
588	AND {$wpdb->posts}.post_password = ''
589	AND {$wpdb->posts}.post_date != '0000-00-00 00:00:00'
590	";
591
592	return $wpdb->prepare( $where_clause, $post_type );
593	}
594
595	/**
596	* Produce array of URL parts for given post object.
597	*
598	* @param object $post Post object to get URL parts for.
599	*
600	* @return array\|bool
601	*/
602	protected function get_url( $post ) {
603
604	$url = [];
605
606	/**
607	* Filter the URL Yoast SEO uses in the XML sitemap.
608	*
609	* Note that only absolute local URLs are allowed as the check after this removes external URLs.
610	*
611	* @param string $url URL to use in the XML sitemap
612	* @param object $post Post object for the URL.
613	*/
614	$url['loc'] = apply_filters( 'wpseo_xml_sitemap_post_url', get_permalink( $post ), $post );
615	$link_type = YoastSEO()->helpers->url->get_link_type(
616	wp_parse_url( $url['loc'] ),
617	$this->get_parsed_home_url()
618	);
619
620	/*
621	* Do not include external URLs.
622	*
623	* {@link https://wordpress.org/plugins/page-links-to/} can rewrite permalinks to external URLs.
624	*/
625	if ( $link_type === SEO_Links::TYPE_EXTERNAL ) {
626	return false;
627	}
628
629	$modified = max( $post->post_modified_gmt, $post->post_date_gmt );
630
631	if ( $modified !== '0000-00-00 00:00:00' ) {
632	$url['mod'] = $modified;
633	}
634
635	$url['chf'] = 'daily'; // Deprecated, kept for backwards data compat. R.
636
637	$canonical = WPSEO_Meta::get_value( 'canonical', $post->ID );
638
639	if ( $canonical !== '' && $canonical !== $url['loc'] ) {
640	/*
641	* Let's assume that if a canonical is set for this page and it's different from
642	* the URL of this post, that page is either already in the XML sitemap OR is on
643	* an external site, either way, we shouldn't include it here.
644	*/
645	return false;
646	}
647	unset( $canonical );
648
649	$url['pri'] = 1; // Deprecated, kept for backwards data compat. R.
650
651	if ( $this->include_images ) {
652	$url['images'] = $this->get_image_parser()->get_images( $post );
653	}
654
655	return $url;
656	}
657	}
658