wordpress-seo / src / repositories / indexable-repository.php

indexable-repository.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 27.5, at src/repositories/indexable-repository.php

725 lines 21.6 KB

Raw Download Zip

1	<?php
2
3	namespace Yoast\WP\SEO\Repositories;
4
5	use Psr\Log\LoggerInterface;
6	use wpdb;
7	use Yoast\WP\Lib\Model;
8	use Yoast\WP\Lib\ORM;
9	use Yoast\WP\SEO\Builders\Indexable_Builder;
10	use Yoast\WP\SEO\Helpers\Current_Page_Helper;
11	use Yoast\WP\SEO\Helpers\Indexable_Helper;
12	use Yoast\WP\SEO\Loggers\Logger;
13	use Yoast\WP\SEO\Models\Indexable;
14	use Yoast\WP\SEO\Services\Indexables\Indexable_Version_Manager;
15
16	/**
17	* Class Indexable_Repository.
18	*/
19	class Indexable_Repository {
20
21	/**
22	* The indexable builder.
23	*
24	* @var Indexable_Builder
25	*/
26	private $builder;
27
28	/**
29	* Represents the hierarchy repository.
30	*
31	* @var Indexable_Hierarchy_Repository
32	*/
33	protected $hierarchy_repository;
34
35	/**
36	* The current page helper.
37	*
38	* @var Current_Page_Helper
39	*/
40	protected $current_page;
41
42	/**
43	* The logger object.
44	*
45	* @var LoggerInterface
46	*/
47	protected $logger;
48
49	/**
50	* The WordPress database.
51	*
52	* @var wpdb
53	*/
54	protected $wpdb;
55
56	/**
57	* Represents the indexable helper.
58	*
59	* @var Indexable_Helper
60	*/
61	protected $indexable_helper;
62
63	/**
64	* Checks if Indexables are up to date.
65	*
66	* @var Indexable_Version_Manager
67	*/
68	protected $version_manager;
69
70	/**
71	* Returns the instance of this class constructed through the ORM Wrapper.
72	*
73	* @param Indexable_Builder $builder The indexable builder.
74	* @param Current_Page_Helper $current_page The current post helper.
75	* @param Logger $logger The logger.
76	* @param Indexable_Hierarchy_Repository $hierarchy_repository The hierarchy repository.
77	* @param wpdb $wpdb The WordPress database instance.
78	* @param Indexable_Version_Manager $version_manager The indexable version manager.
79	*/
80	public function __construct(
81	Indexable_Builder $builder,
82	Current_Page_Helper $current_page,
83	Logger $logger,
84	Indexable_Hierarchy_Repository $hierarchy_repository,
85	wpdb $wpdb,
86	Indexable_Version_Manager $version_manager
87	) {
88	$this->builder = $builder;
89	$this->current_page = $current_page;
90	$this->logger = $logger;
91	$this->hierarchy_repository = $hierarchy_repository;
92	$this->wpdb = $wpdb;
93	$this->version_manager = $version_manager;
94	}
95
96	/**
97	* Starts a query for this repository.
98	*
99	* @return ORM
100	*/
101	public function query() {
102	return Model::of_type( 'Indexable' );
103	}
104
105	/**
106	* Attempts to find the indexable for the current WordPress page. Returns false if no indexable could be found.
107	* This may be the result of the indexable not existing or of being unable to determine what type of page the
108	* current page is.
109	*
110	* @return bool\|Indexable The indexable. If no indexable is found returns an empty indexable. Returns false if
111	* there is a database error.
112	*/
113	public function for_current_page() {
114	$indexable = false;
115
116	switch ( true ) {
117	case $this->current_page->is_simple_page():
118	$indexable = $this->find_by_id_and_type( $this->current_page->get_simple_page_id(), 'post' );
119	break;
120	case $this->current_page->is_home_static_page():
121	$indexable = $this->find_by_id_and_type( $this->current_page->get_front_page_id(), 'post' );
122	break;
123	case $this->current_page->is_home_posts_page():
124	$indexable = $this->find_for_home_page();
125	break;
126	case $this->current_page->is_term_archive():
127	$indexable = $this->find_by_id_and_type( $this->current_page->get_term_id(), 'term' );
128	break;
129	case $this->current_page->is_date_archive():
130	$indexable = $this->find_for_date_archive();
131	break;
132	case $this->current_page->is_search_result():
133	$indexable = $this->find_for_system_page( 'search-result' );
134	break;
135	case $this->current_page->is_post_type_archive():
136	$indexable = $this->find_for_post_type_archive( $this->current_page->get_queried_post_type() );
137	break;
138	case $this->current_page->is_author_archive():
139	$indexable = $this->find_by_id_and_type( $this->current_page->get_author_id(), 'user' );
140	break;
141	case $this->current_page->is_404():
142	$indexable = $this->find_for_system_page( '404' );
143	break;
144	}
145
146	if ( $indexable === false ) {
147	return $this->query()->create(
148	[
149	'object_type' => 'unknown',
150	'post_status' => 'unindexed',
151	'version' => 1,
152	],
153	);
154	}
155
156	return $indexable;
157	}
158
159	/**
160	* Retrieves an indexable by its permalink.
161	*
162	* @param string $permalink The indexable permalink.
163	*
164	* @return bool\|Indexable The indexable, false if none could be found.
165	*/
166	public function find_by_permalink( $permalink ) {
167	$permalink_hash = \strlen( $permalink ) . ':' . \md5( $permalink );
168
169	// Find by both permalink_hash and permalink, permalink_hash is indexed so will be used first by the DB to optimize the query.
170	return $this->query()
171	->where( 'permalink_hash', $permalink_hash )
172	->where( 'permalink', $permalink )
173	->find_one();
174	}
175
176	/**
177	* Retrieves all the indexable instances of a certain object type.
178	*
179	* @param string $object_type The object type.
180	*
181	* @return Indexable[] The array with all the indexable instances of a certain object type.
182	*/
183	public function find_all_with_type( $object_type ) {
184	/**
185	* The array with all the indexable instances of a certain object type.
186	*
187	* @var Indexable[] $indexables
188	*/
189	$indexables = $this
190	->query()
191	->where( 'object_type', $object_type )
192	->find_many();
193
194	return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
195	}
196
197	/**
198	* Retrieves all the indexable instances of a certain object subtype.
199	*
200	* @param string $object_type The object type.
201	* @param string $object_sub_type The object subtype.
202	*
203	* @return Indexable[] The array with all the indexable instances of a certain object subtype.
204	*/
205	public function find_all_with_type_and_sub_type( $object_type, $object_sub_type ) {
206	/**
207	* The array with all the indexable instances of a certain object type and subtype.
208	*
209	* @var Indexable[] $indexables
210	*/
211	$indexables = $this
212	->query()
213	->where( 'object_type', $object_type )
214	->where( 'object_sub_type', $object_sub_type )
215	->find_many();
216
217	return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
218	}
219
220	/**
221	* Retrieves a paginated set of indexable instances of public indexables.
222	*
223	* @param int $page The page number (1-based).
224	* @param int $page_size The number of items per page.
225	* @param string $post_type The post type indexables to find.
226	*
227	* @return Indexable[] The array with the paginated indexable instances which are public.
228	*/
229	public function find_all_public_paginated( int $page, int $page_size, string $post_type ): array {
230	$offset = ( ( $page - 1 ) * $page_size );
231
232	$query = $this->query()->where_raw( '( is_public IS NULL OR is_public = 1 ) AND ( is_robots_noindex IS NULL OR is_robots_noindex = 0 )' );
233	$query->where( 'object_sub_type', $post_type );
234	$query->where( 'post_status', 'publish' );
235
236	$indexables = $query->order_by_asc( 'id' )->limit( $page_size )->offset( $offset )->find_many();
237
238	return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
239	}
240
241	/**
242	* Retrieves the homepage indexable.
243	*
244	* @param bool $auto_create Optional. Create the indexable if it does not exist.
245	*
246	* @return bool\|Indexable Instance of indexable.
247	*/
248	public function find_for_home_page( $auto_create = true ) {
249	$indexable = \wp_cache_get( 'home-page', 'yoast-seo-indexables' );
250	if ( ! $indexable ) {
251	/**
252	* Indexable instance.
253	*
254	* @var Indexable $indexable
255	*/
256	$indexable = $this->query()->where( 'object_type', 'home-page' )->find_one();
257
258	if ( $auto_create && ! $indexable ) {
259	$indexable = $this->builder->build_for_home_page();
260	}
261
262	$indexable = $this->upgrade_indexable( $indexable );
263
264	\wp_cache_set( 'home-page', $indexable, 'yoast-seo-indexables', ( 5 * \MINUTE_IN_SECONDS ) );
265	}
266
267	return $indexable;
268	}
269
270	/**
271	* Retrieves the date archive indexable.
272	*
273	* @param bool $auto_create Optional. Create the indexable if it does not exist.
274	*
275	* @return bool\|Indexable Instance of indexable.
276	*/
277	public function find_for_date_archive( $auto_create = true ) {
278	/**
279	* Indexable instance.
280	*
281	* @var Indexable $indexable
282	*/
283	$indexable = $this->query()->where( 'object_type', 'date-archive' )->find_one();
284
285	if ( $auto_create && ! $indexable ) {
286	$indexable = $this->builder->build_for_date_archive();
287	}
288
289	return $this->upgrade_indexable( $indexable );
290	}
291
292	/**
293	* Retrieves an indexable for a post type archive.
294	*
295	* @param string $post_type The post type.
296	* @param bool $auto_create Optional. Create the indexable if it does not exist.
297	*
298	* @return bool\|Indexable The indexable, false if none could be found.
299	*/
300	public function find_for_post_type_archive( $post_type, $auto_create = true ) {
301	/**
302	* Indexable instance.
303	*
304	* @var Indexable $indexable
305	*/
306	$indexable = $this->query()
307	->where( 'object_type', 'post-type-archive' )
308	->where( 'object_sub_type', $post_type )
309	->find_one();
310
311	if ( $auto_create && ! $indexable ) {
312	$indexable = $this->builder->build_for_post_type_archive( $post_type );
313	}
314
315	return $this->upgrade_indexable( $indexable );
316	}
317
318	/**
319	* Retrieves the indexable for a system page.
320	*
321	* @param string $object_sub_type The type of system page.
322	* @param bool $auto_create Optional. Create the indexable if it does not exist.
323	*
324	* @return bool\|Indexable Instance of indexable.
325	*/
326	public function find_for_system_page( $object_sub_type, $auto_create = true ) {
327	/**
328	* Indexable instance.
329	*
330	* @var Indexable $indexable
331	*/
332	$indexable = $this->query()
333	->where( 'object_type', 'system-page' )
334	->where( 'object_sub_type', $object_sub_type )
335	->find_one();
336
337	if ( $auto_create && ! $indexable ) {
338	$indexable = $this->builder->build_for_system_page( $object_sub_type );
339	}
340
341	return $this->upgrade_indexable( $indexable );
342	}
343
344	/**
345	* Retrieves an indexable by its ID and type.
346	*
347	* @param int $object_id The indexable object ID.
348	* @param string $object_type The indexable object type.
349	* @param bool $auto_create Optional. Create the indexable if it does not exist.
350	*
351	* @return bool\|Indexable Instance of indexable.
352	*/
353	public function find_by_id_and_type( $object_id, $object_type, $auto_create = true ) {
354	$indexable = $this->query()
355	->where( 'object_id', $object_id )
356	->where( 'object_type', $object_type )
357	->find_one();
358
359	if ( $auto_create && ! $indexable ) {
360	$indexable = $this->builder->build_for_id_and_type( $object_id, $object_type );
361	}
362	else {
363	$indexable = $this->upgrade_indexable( $indexable );
364	}
365
366	return $indexable;
367	}
368
369	/**
370	* Retrieves multiple indexables at once by their id's and type.
371	*
372	* @param int[] $object_ids The array of indexable object id's.
373	* @param string $object_type The indexable object type.
374	* @param bool $auto_create Optional. Create the indexable if it does not exist.
375	*
376	* @return Indexable[] An array of indexables.
377	*/
378	public function find_by_multiple_ids_and_type( $object_ids, $object_type, $auto_create = true ) {
379	if ( empty( $object_ids ) ) {
380	return [];
381	}
382
383	/**
384	* Represents an array of indexable objects.
385	*
386	* @var Indexable[] $indexables
387	*/
388	$indexables = $this->query()
389	->where_in( 'object_id', $object_ids )
390	->where( 'object_type', $object_type )
391	->find_many();
392
393	if ( $auto_create ) {
394	$indexables_available = [];
395	foreach ( $indexables as $indexable ) {
396	$indexables_available[] = $indexable->object_id;
397	}
398
399	$indexables_to_create = \array_diff( $object_ids, $indexables_available );
400
401	foreach ( $indexables_to_create as $indexable_to_create ) {
402	$indexables[] = $this->builder->build_for_id_and_type( $indexable_to_create, $object_type );
403	}
404	}
405
406	return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
407	}
408
409	/**
410	* Finds the indexables by id's.
411	*
412	* @param array $indexable_ids The indexable id's.
413	*
414	* @return Indexable[] The found indexables.
415	*/
416	public function find_by_ids( array $indexable_ids ) {
417	if ( empty( $indexable_ids ) ) {
418	return [];
419	}
420
421	$indexables = $this
422	->query()
423	->where_in( 'id', $indexable_ids )
424	->find_many();
425
426	return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
427	}
428
429	/**
430	* Returns all ancestors of a given indexable.
431	*
432	* @param Indexable $indexable The indexable to find the ancestors of.
433	*
434	* @return Indexable[] All ancestors of the given indexable.
435	*/
436	public function get_ancestors( Indexable $indexable ) {
437	// If we've already set ancestors on the indexable no need to get them again.
438	if ( \is_array( $indexable->ancestors ) && ! empty( $indexable->ancestors ) ) {
439	return \array_map( [ $this, 'upgrade_indexable' ], $indexable->ancestors );
440	}
441
442	$indexable_ids = $this->hierarchy_repository->find_ancestors( $indexable );
443
444	// If we've set ancestors on the indexable because we had to build them to find them.
445	if ( \is_array( $indexable->ancestors ) && ! empty( $indexable->ancestors ) ) {
446	return \array_map( [ $this, 'upgrade_indexable' ], $indexable->ancestors );
447	}
448
449	if ( empty( $indexable_ids ) ) {
450	return [];
451	}
452
453	if ( $indexable_ids[0] === 0 && \count( $indexable_ids ) === 1 ) {
454	return [];
455	}
456
457	$indexables = $this->query()
458	->where_in( 'id', $indexable_ids )
459	->order_by_expr( 'FIELD(id,' . \implode( ',', $indexable_ids ) . ')' )
460	->find_many();
461
462	return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
463	}
464
465	/**
466	* Returns all subpages with a given post_parent.
467	*
468	* @param int $post_parent The post parent.
469	* @param array $exclude_ids The id's to exclude.
470	*
471	* @return Indexable[] array of indexables.
472	*/
473	public function get_subpages_by_post_parent( $post_parent, $exclude_ids = [] ) {
474	$query = $this->query()
475	->where( 'post_parent', $post_parent )
476	->where( 'object_type', 'post' )
477	->where( 'post_status', 'publish' );
478
479	if ( ! empty( $exclude_ids ) ) {
480	$query->where_not_in( 'object_id', $exclude_ids );
481	}
482	return $query->find_many();
483	}
484
485	/**
486	* Returns most recently modified posts of a post type.
487	*
488	* @param string $post_type The post type.
489	* @param int $limit The maximum number of posts to return.
490	* @param bool $exclude_older_than_one_year Whether to exclude posts older than one year.
491	* @param string $search_filter Optional. A search filter to apply to the breadcrumb title.
492	*
493	* @return Indexable[] array of indexables.
494	*/
495	public function get_recently_modified_posts( string $post_type, int $limit, bool $exclude_older_than_one_year, string $search_filter = '' ) {
496	$query = $this->query()
497	->where( 'object_type', 'post' )
498	->where( 'object_sub_type', $post_type )
499	->where_raw( '( is_public IS NULL OR is_public = 1 )' )
500	->order_by_desc( 'object_last_modified' )
501	->limit( $limit );
502
503	if ( $exclude_older_than_one_year === true ) {
504	$query->where_gte( 'object_published_at', \gmdate( 'Y-m-d H:i:s', \strtotime( '-1 year' ) ) );
505	}
506
507	if ( $search_filter !== '' ) {
508	$query->where_like( 'breadcrumb_title', '%' . $search_filter . '%' );
509	}
510
511	$query->order_by_desc( 'object_last_modified' )
512	->limit( $limit );
513
514	return $query->find_many();
515	}
516
517	/**
518	* Returns the most recently modified cornerstone content of a post type.
519	*
520	* @param string $post_type The post type.
521	* @param int\|null $limit The maximum number of posts to return.
522	*
523	* @return Indexable[] array of indexables.
524	*/
525	public function get_recent_cornerstone_for_post_type( string $post_type, ?int $limit ) {
526	$query = $this->query()
527	->where( 'object_type', 'post' )
528	->where( 'object_sub_type', $post_type )
529	->where_raw( '( is_public IS NULL OR is_public = 1 )' )
530	->where( 'is_cornerstone', 1 )
531	->order_by_desc( 'object_last_modified' );
532
533	if ( $limit !== null ) {
534	$query->limit( $limit );
535	}
536
537	return $query->find_many();
538	}
539
540	/**
541	* Returns the most recently modified posts with keywords of a post type.
542	*
543	* @param string $post_type The post type.
544	* @param int\|null $limit The maximum number of posts to return.
545	* @param string\|null $date_limit Only include content modified after this date.
546	*
547	* @return array<array<string, string>>\|false The array of indexable columns. False if the query failed.
548	*/
549	public function get_recent_posts_with_keywords_for_post_type( string $post_type, ?int $limit = null, ?string $date_limit = null ) {
550	$query = $this->query()
551	->select( 'object_id' )
552	->select( 'primary_focus_keyword_score' )
553	->select( 'breadcrumb_title' )
554	->where( 'object_type', 'post' )
555	->where( 'object_sub_type', $post_type )
556	->where_not_equal( 'primary_focus_keyword_score', 0 )
557	->where_not_null( 'primary_focus_keyword_score' )
558	->where_raw( "( post_status = 'publish' OR post_status IS NULL )" )
559	->where_raw( '( is_robots_noindex IS NULL OR is_robots_noindex <> 1 )' )
560	->order_by_desc( 'object_last_modified' );
561
562	if ( $limit !== null ) {
563	$query->limit( $limit );
564	}
565
566	if ( $date_limit !== null ) {
567	$query->where_gte( 'object_last_modified', $date_limit );
568	}
569
570	return $query->find_array();
571	}
572
573	/**
574	* Returns the most recently modified posts with readability scores of a post type.
575	*
576	* @param string $post_type The post type.
577	* @param int\|null $limit The maximum number of posts to return.
578	* @param string\|null $date_limit Only include content modified after this date.
579	*
580	* @return array<array<string, string\|null>>\|false The array of indexable columns. False if the query failed.
581	*/
582	public function get_recent_posts_with_readability_scores_for_post_type( string $post_type, ?int $limit = null, ?string $date_limit = null ) {
583	$query = $this->query()
584	->select( 'object_id' )
585	->select( 'readability_score' )
586	->select( 'breadcrumb_title' )
587	->where( 'object_type', 'post' )
588	->where( 'object_sub_type', $post_type )
589	->where_not_null( 'estimated_reading_time_minutes' )
590	->where_raw( "( post_status = 'publish' OR post_status IS NULL )" )
591	->order_by_desc( 'object_last_modified' );
592
593	if ( $limit !== null ) {
594	$query->limit( $limit );
595	}
596
597	if ( $date_limit !== null ) {
598	$query->where_gte( 'object_last_modified', $date_limit );
599	}
600
601	return $query->find_array();
602	}
603
604	/**
605	* Returns the most recently modified posts for a post type.
606	*
607	* @param string $post_type The post type.
608	* @param int\|null $limit The maximum number of posts to return.
609	* @param string\|null $date_limit Only include content modified after this date.
610	*
611	* @return array<array<string, string\|null>>\|false The array of indexable columns. False if the query failed.
612	*/
613	public function get_recent_posts_for_post_type( string $post_type, ?int $limit = null, ?string $date_limit = null ) {
614	$query = $this->query()
615	->select( 'object_id' )
616	->select( 'breadcrumb_title' )
617	->select( 'description' )
618	->where( 'object_type', 'post' )
619	->where( 'object_sub_type', $post_type )
620	->where_raw( "( post_status = 'publish' OR post_status IS NULL )" )
621	->where_raw( '( is_robots_noindex IS NULL OR is_robots_noindex <> 1 )' )
622	->order_by_desc( 'object_last_modified' );
623
624	if ( $limit !== null ) {
625	$query->limit( $limit );
626	}
627
628	if ( $date_limit !== null ) {
629	$query->where_gte( 'object_last_modified', $date_limit );
630	}
631
632	return $query->find_array();
633	}
634
635	/**
636	* Updates the incoming link count for an indexable without first fetching it.
637	*
638	* @param int $indexable_id The indexable id.
639	* @param int $count The incoming link count.
640	*
641	* @return bool Whether or not the update was succeful.
642	*/
643	public function update_incoming_link_count( $indexable_id, $count ) {
644	return (bool) $this->query()
645	->set( 'incoming_link_count', $count )
646	->where( 'id', $indexable_id )
647	->update_many();
648	}
649
650	/**
651	* Ensures that the given indexable has a permalink.
652	*
653	* Will be deprecated in 17.3 - Use upgrade_indexable instead.
654	*
655	* @codeCoverageIgnore
656	*
657	* @param Indexable $indexable The indexable.
658	*
659	* @return bool\|Indexable The indexable.
660	*/
661	public function ensure_permalink( $indexable ) {
662	// @phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- self::class is safe.
663	// @phpcs:ignore Squiz.PHP.CommentedOutCode.Found
664	// _deprecated_function( __METHOD__, 'Yoast SEO 17.3', self::class . '::upgrade_indexable' );
665
666	return $this->upgrade_indexable( $indexable );
667	}
668
669	/**
670	* Checks if an Indexable is outdated, and rebuilds it when necessary.
671	*
672	* @param Indexable $indexable The indexable.
673	*
674	* @return Indexable The indexable.
675	*/
676	public function upgrade_indexable( $indexable ) {
677	if ( $this->version_manager->indexable_needs_upgrade( $indexable ) ) {
678	$indexable = $this->builder->build( $indexable );
679	}
680	return $indexable;
681	}
682
683	/**
684	* Resets the permalinks of the passed object type and subtype.
685	*
686	* @param string\|null $type The type of the indexable. Can be null.
687	* @param string\|null $subtype The subtype. Can be null.
688	* @param int\|null $object_id The object ID. Can be null.
689	*
690	* @return int\|bool The number of permalinks changed if the query was succesful. False otherwise.
691	*/
692	public function reset_permalink( $type = null, $subtype = null, $object_id = null ) {
693	$query = $this->query()->set(
694	[
695	'permalink' => null,
696	'permalink_hash' => null,
697	'version' => 0,
698	],
699	);
700
701	if ( $type !== null ) {
702	$query->where( 'object_type', $type );
703	}
704
705	if ( $type !== null && $subtype !== null ) {
706	$query->where( 'object_sub_type', $subtype );
707	}
708
709	if ( $object_id !== null ) {
710	$query->where( 'object_id', $object_id );
711	}
712
713	return $query->update_many();
714	}
715
716	/**
717	* Gets the total number of stored indexables.
718	*
719	* @return int The total number of stored indexables.
720	*/
721	public function get_total_number_of_indexables() {
722	return $this->query()->count();
723	}
724	}
725