← All changes | src/repositories/indexable-repository.php +298 -9 18.0 → trunk View file →

		@@ -18,8 +18,23 @@
18	18	*/
19	19	class Indexable_Repository {
20	20
21	21	/**
	22	+ * The maximum number of comma-separated phrases honoured by a title-keyword search.
	23	+ * Any phrases beyond this are ignored so an oversized list cannot blow up the query.
	24	+ *
	25	+ * @var int
	26	+ */
	27	+ public const MAX_TITLE_KEYWORD_PHRASES = 10;
	28	+
	29	+ /**
	30	+ * The maximum page size honoured by a title-keyword search.
	31	+ *
	32	+ * @var int
	33	+ */
	34	+ public const MAX_TITLE_KEYWORD_PAGE_SIZE = 100;
	35	+
	36	+ /**
22	37	* The indexable builder.
23	38	*
24	39	* @var Indexable_Builder
25	40	*/
		@@ -106,9 +121,10 @@
106	121	* Attempts to find the indexable for the current WordPress page. Returns false if no indexable could be found.
107	122	* This may be the result of the indexable not existing or of being unable to determine what type of page the
108	123	* current page is.
109	124	*
110		- * @return bool|Indexable The indexable, false if none could be found.
	125	+ * @return bool|Indexable The indexable. If no indexable is found returns an empty indexable. Returns false if
	126	+ * there is a database error.
111	127	*/
112	128	public function for_current_page() {
113	129	$indexable = false;
114	130
		@@ -147,9 +163,9 @@
147	163	[
148	164	'object_type' => 'unknown',
149	165	'post_status' => 'unindexed',
150	166	'version' => 1,
151		- ]
	167	+ ],
152	168	);
153	169	}
154	170
155	171	return $indexable;
		@@ -216,8 +232,29 @@
216	232	return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
217	233	}
218	234
219	235	/**
	236	+ * Retrieves a paginated set of indexable instances of public indexables.
	237	+ *
	238	+ * @param int $page The page number (1-based).
	239	+ * @param int $page_size The number of items per page.
	240	+ * @param string $post_type The post type indexables to find.
	241	+ *
	242	+ * @return Indexable[] The array with the paginated indexable instances which are public.
	243	+ */
	244	+ public function find_all_public_paginated( int $page, int $page_size, string $post_type ): array {
	245	+ $offset = ( ( $page - 1 ) * $page_size );
	246	+
	247	+ $query = $this->query()->where_raw( '( is_public IS NULL OR is_public = 1 ) AND ( is_robots_noindex IS NULL OR is_robots_noindex = 0 )' );
	248	+ $query->where( 'object_sub_type', $post_type );
	249	+ $query->where( 'post_status', 'publish' );
	250	+
	251	+ $indexables = $query->order_by_asc( 'id' )->limit( $page_size )->offset( $offset )->find_many();
	252	+
	253	+ return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
	254	+ }
	255	+
	256	+ /**
220	257	* Retrieves the homepage indexable.
221	258	*
222	259	* @param bool $auto_create Optional. Create the indexable if it does not exist.
223	260	*
		@@ -375,8 +412,18 @@
375	412	}
376	413
377	414	$indexables_to_create = \array_diff( $object_ids, $indexables_available );
378	415
	416	+ if ( ! empty( $indexables_to_create ) ) {
	417	+ // Warm the object caches for the whole batch, so each build below does not trigger its own uncached queries.
	418	+ if ( $object_type === 'post' ) {
	419	+ \_prime_post_caches( $indexables_to_create );
	420	+ }
	421	+ elseif ( $object_type === 'term' ) {
	422	+ \_prime_term_caches( $indexables_to_create );
	423	+ }
	424	+ }
	425	+
379	426	foreach ( $indexables_to_create as $indexable_to_create ) {
380	427	$indexables[] = $this->builder->build_for_id_and_type( $indexable_to_create, $object_type );
381	428	}
382	429	}
		@@ -386,9 +433,9 @@
386	433
387	434	/**
388	435	* Finds the indexables by id's.
389	436	*
390		- * @param array $indexable_ids The indexable id's.
	437	+ * @param int[] $indexable_ids The indexable id's.
391	438	*
392	439	* @return Indexable[] The found indexables.
393	440	*/
394	441	public function find_by_ids( array $indexable_ids ) {
		@@ -443,9 +490,9 @@
443	490	/**
444	491	* Returns all subpages with a given post_parent.
445	492	*
446	493	* @param int $post_parent The post parent.
447		- * @param array $exclude_ids The id's to exclude.
	494	+ * @param int[] $exclude_ids The id's to exclude.
448	495	*
449	496	* @return Indexable[] array of indexables.
450	497	*/
451	498	public function get_subpages_by_post_parent( $post_parent, $exclude_ids = [] ) {
		@@ -460,8 +507,236 @@
460	507	return $query->find_many();
461	508	}
462	509
463	510	/**
	511	+ * Returns most recently modified posts of a post type.
	512	+ *
	513	+ * @param string $post_type The post type.
	514	+ * @param int $limit The maximum number of posts to return.
	515	+ * @param bool $exclude_older_than_one_year Whether to exclude posts older than one year.
	516	+ * @param string $search_filter Optional. A search filter to apply to the breadcrumb title.
	517	+ *
	518	+ * @return Indexable[] array of indexables.
	519	+ */
	520	+ public function get_recently_modified_posts( string $post_type, int $limit, bool $exclude_older_than_one_year, string $search_filter = '' ) {
	521	+ $query = $this->query()
	522	+ ->where( 'object_type', 'post' )
	523	+ ->where( 'object_sub_type', $post_type )
	524	+ ->where_raw( '( is_public IS NULL OR is_public = 1 )' )
	525	+ ->order_by_desc( 'object_last_modified' )
	526	+ ->limit( $limit );
	527	+
	528	+ if ( $exclude_older_than_one_year === true ) {
	529	+ $query->where_gte( 'object_published_at', \gmdate( 'Y-m-d H:i:s', \strtotime( '-1 year' ) ) );
	530	+ }
	531	+
	532	+ if ( $search_filter !== '' ) {
	533	+ $query->where_like( 'breadcrumb_title', '%' . $search_filter . '%' );
	534	+ }
	535	+
	536	+ $query->order_by_desc( 'object_last_modified' )
	537	+ ->limit( $limit );
	538	+
	539	+ return $query->find_many();
	540	+ }
	541	+
	542	+ /**
	543	+ * Finds posts whose breadcrumb title contains any of the given comma-separated phrases.
	544	+ *
	545	+ * The search string is a comma-separated list. Each value is matched as a whole
	546	+ * contiguous substring of the breadcrumb title, and a post is returned when it
	547	+ * contains any one of the values (a logical OR between values). For example,
	548	+ * "hiking boots, trail" returns posts whose title contains "hiking boots" or "trail".
	549	+ *
	550	+ * Results are paginated and ordered most recently modified first (with the indexable
	551	+ * id as a stable tiebreaker), so requesting a later page returns older matches.
	552	+ *
	553	+ * At most self::MAX_TITLE_KEYWORD_PHRASES phrases are honoured; any beyond that are ignored.
	554	+ * The page size is clamped to the range 1..self::MAX_TITLE_KEYWORD_PAGE_SIZE.
	555	+ *
	556	+ * @param string $keywords The comma-separated phrases to match against the breadcrumb title.
	557	+ * @param int $page The page of results to return, 1-based.
	558	+ * @param int $page_size The number of posts per page.
	559	+ * @param string $post_type The post type to restrict the search to.
	560	+ *
	561	+ * @return Indexable[] The matching indexables for the requested page, ordered by most recently modified.
	562	+ */
	563	+ public function find_posts_by_title_keywords( string $keywords, int $page = 1, int $page_size = 10, string $post_type = 'post' ) {
	564	+ $phrases = \array_map( 'trim', \explode( ',', $keywords ) );
	565	+ $phrases = \array_filter(
	566	+ $phrases,
	567	+ static function ( $phrase ) {
	568	+ return $phrase !== '';
	569	+ },
	570	+ );
	571	+ $phrases = \array_slice( $phrases, 0, self::MAX_TITLE_KEYWORD_PHRASES );
	572	+
	573	+ // An empty search must not degrade into matching every post.
	574	+ if ( empty( $phrases ) ) {
	575	+ return [];
	576	+ }
	577	+
	578	+ $likes = \array_fill( 0, \count( $phrases ), 'breadcrumb_title LIKE %s' );
	579	+ $params = \array_map(
	580	+ function ( $phrase ) {
	581	+ return '%' . $this->wpdb->esc_like( $phrase ) . '%';
	582	+ },
	583	+ $phrases,
	584	+ );
	585	+
	586	+ $page_size = \min( \max( 1, $page_size ), self::MAX_TITLE_KEYWORD_PAGE_SIZE );
	587	+ $offset = ( ( \max( 1, $page ) - 1 ) * $page_size );
	588	+
	589	+ $indexables = $this->query()
	590	+ ->where( 'object_type', 'post' )
	591	+ ->where( 'object_sub_type', $post_type )
	592	+ ->where_raw( '( ' . \implode( ' OR ', $likes ) . ' )', \array_values( $params ) )
	593	+ ->order_by_desc( 'object_last_modified' )
	594	+ ->order_by_desc( 'id' )
	595	+ ->limit( $page_size )
	596	+ ->offset( $offset )
	597	+ ->find_many();
	598	+
	599	+ return \array_map( [ $this, 'upgrade_indexable' ], $indexables );
	600	+ }
	601	+
	602	+ /**
	603	+ * Returns the most recently modified cornerstone content of a post type.
	604	+ *
	605	+ * @param string $post_type The post type.
	606	+ * @param int|null $limit The maximum number of posts to return.
	607	+ *
	608	+ * @return Indexable[] array of indexables.
	609	+ */
	610	+ public function get_recent_cornerstone_for_post_type( string $post_type, ?int $limit ) {
	611	+ $query = $this->query()
	612	+ ->where( 'object_type', 'post' )
	613	+ ->where( 'object_sub_type', $post_type )
	614	+ ->where_raw( '( is_public IS NULL OR is_public = 1 )' )
	615	+ ->where( 'is_cornerstone', 1 )
	616	+ ->order_by_desc( 'object_last_modified' );
	617	+
	618	+ if ( $limit !== null ) {
	619	+ $query->limit( $limit );
	620	+ }
	621	+
	622	+ return $query->find_many();
	623	+ }
	624	+
	625	+ /**
	626	+ * Returns the most recently modified about page based on schema_page_type.
	627	+ *
	628	+ * @param string $post_type The post type.
	629	+ *
	630	+ * @return Indexable|false The about page if its there.
	631	+ */
	632	+ public function get_most_recent_about_page( $post_type ) {
	633	+ $query = $this->query()
	634	+ ->where( 'object_type', 'post' )
	635	+ ->where( 'object_sub_type', $post_type )
	636	+ ->where( 'schema_page_type', 'AboutPage' )
	637	+ ->where_raw( '( is_public IS NULL OR is_public = 1 )' )
	638	+ ->order_by_desc( 'object_last_modified' );
	639	+
	640	+ return $query->find_one();
	641	+ }
	642	+
	643	+ /**
	644	+ * Returns the most recently modified posts with keywords of a post type.
	645	+ *
	646	+ * @param string $post_type The post type.
	647	+ * @param int|null $limit The maximum number of posts to return.
	648	+ * @param string|null $date_limit Only include content modified after this date.
	649	+ *
	650	+ * @return array<array<string, string>>|false The array of indexable columns. False if the query failed.
	651	+ */
	652	+ public function get_recent_posts_with_keywords_for_post_type( string $post_type, ?int $limit = null, ?string $date_limit = null ) {
	653	+ $query = $this->query()
	654	+ ->select( 'object_id' )
	655	+ ->select( 'primary_focus_keyword_score' )
	656	+ ->select( 'breadcrumb_title' )
	657	+ ->where( 'object_type', 'post' )
	658	+ ->where( 'object_sub_type', $post_type )
	659	+ ->where_not_equal( 'primary_focus_keyword_score', 0 )
	660	+ ->where_not_null( 'primary_focus_keyword_score' )
	661	+ ->where_raw( "( post_status = 'publish' OR post_status IS NULL )" )
	662	+ ->where_raw( '( is_robots_noindex IS NULL OR is_robots_noindex <> 1 )' )
	663	+ ->order_by_desc( 'object_last_modified' );
	664	+
	665	+ if ( $limit !== null ) {
	666	+ $query->limit( $limit );
	667	+ }
	668	+
	669	+ if ( $date_limit !== null ) {
	670	+ $query->where_gte( 'object_last_modified', $date_limit );
	671	+ }
	672	+
	673	+ return $query->find_array();
	674	+ }
	675	+
	676	+ /**
	677	+ * Returns the most recently modified posts with readability scores of a post type.
	678	+ *
	679	+ * @param string $post_type The post type.
	680	+ * @param int|null $limit The maximum number of posts to return.
	681	+ * @param string|null $date_limit Only include content modified after this date.
	682	+ *
	683	+ * @return array<array<string, string|null>>|false The array of indexable columns. False if the query failed.
	684	+ */
	685	+ public function get_recent_posts_with_readability_scores_for_post_type( string $post_type, ?int $limit = null, ?string $date_limit = null ) {
	686	+ $query = $this->query()
	687	+ ->select( 'object_id' )
	688	+ ->select( 'readability_score' )
	689	+ ->select( 'breadcrumb_title' )
	690	+ ->where( 'object_type', 'post' )
	691	+ ->where( 'object_sub_type', $post_type )
	692	+ ->where_not_null( 'estimated_reading_time_minutes' )
	693	+ ->where_raw( "( post_status = 'publish' OR post_status IS NULL )" )
	694	+ ->order_by_desc( 'object_last_modified' );
	695	+
	696	+ if ( $limit !== null ) {
	697	+ $query->limit( $limit );
	698	+ }
	699	+
	700	+ if ( $date_limit !== null ) {
	701	+ $query->where_gte( 'object_last_modified', $date_limit );
	702	+ }
	703	+
	704	+ return $query->find_array();
	705	+ }
	706	+
	707	+ /**
	708	+ * Returns the most recently modified posts for a post type.
	709	+ *
	710	+ * @param string $post_type The post type.
	711	+ * @param int|null $limit The maximum number of posts to return.
	712	+ * @param string|null $date_limit Only include content modified after this date.
	713	+ *
	714	+ * @return array<array<string, string|null>>|false The array of indexable columns. False if the query failed.
	715	+ */
	716	+ public function get_recent_posts_for_post_type( string $post_type, ?int $limit = null, ?string $date_limit = null ) {
	717	+ $query = $this->query()
	718	+ ->select( 'object_id' )
	719	+ ->select( 'breadcrumb_title' )
	720	+ ->select( 'description' )
	721	+ ->where( 'object_type', 'post' )
	722	+ ->where( 'object_sub_type', $post_type )
	723	+ ->where_raw( "( post_status = 'publish' OR post_status IS NULL )" )
	724	+ ->where_raw( '( is_robots_noindex IS NULL OR is_robots_noindex <> 1 )' )
	725	+ ->order_by_desc( 'object_last_modified' );
	726	+
	727	+ if ( $limit !== null ) {
	728	+ $query->limit( $limit );
	729	+ }
	730	+
	731	+ if ( $date_limit !== null ) {
	732	+ $query->where_gte( 'object_last_modified', $date_limit );
	733	+ }
	734	+
	735	+ return $query->find_array();
	736	+ }
	737	+
	738	+ /**
464	739	* Updates the incoming link count for an indexable without first fetching it.
465	740	*
466	741	* @param int $indexable_id The indexable id.
467	742	* @param int $count The incoming link count.
		@@ -488,9 +763,9 @@
488	763	*/
489	764	public function ensure_permalink( $indexable ) {
490	765	// @phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- self::class is safe.
491	766	// @phpcs:ignore Squiz.PHP.CommentedOutCode.Found
492		- // _deprecated_function( __METHOD__, 'WPSEO 17.3', self::class . '::upgrade_indexable' );
	767	+ // _deprecated_function( __METHOD__, 'Yoast SEO 17.3', self::class . '::upgrade_indexable' );
493	768
494	769	return $this->upgrade_indexable( $indexable );
495	770	}
496	771
		@@ -510,20 +785,21 @@
510	785
511	786	/**
512	787	* Resets the permalinks of the passed object type and subtype.
513	788	*
514		- * @param string|null $type The type of the indexable. Can be null.
515		- * @param string|null $subtype The subtype. Can be null.
	789	+ * @param string|null $type The type of the indexable. Can be null.
	790	+ * @param string|null $subtype The subtype. Can be null.
	791	+ * @param int|null $object_id The object ID. Can be null.
516	792	*
517	793	* @return int|bool The number of permalinks changed if the query was succesful. False otherwise.
518	794	*/
519		- public function reset_permalink( $type = null, $subtype = null ) {
	795	+ public function reset_permalink( $type = null, $subtype = null, $object_id = null ) {
520	796	$query = $this->query()->set(
521	797	[
522	798	'permalink' => null,
523	799	'permalink_hash' => null,
524	800	'version' => 0,
525		- ]
	801	+ ],
526	802	);
527	803
528	804	if ( $type !== null ) {
529	805	$query->where( 'object_type', $type );
		@@ -532,7 +808,20 @@
532	808	if ( $type !== null && $subtype !== null ) {
533	809	$query->where( 'object_sub_type', $subtype );
534	810	}
535	811
	812	+ if ( $object_id !== null ) {
	813	+ $query->where( 'object_id', $object_id );
	814	+ }
	815	+
536	816	return $query->update_many();
	817	+ }
	818	+
	819	+ /**
	820	+ * Gets the total number of stored indexables.
	821	+ *
	822	+ * @return int The total number of stored indexables.
	823	+ */
	824	+ public function get_total_number_of_indexables() {
	825	+ return $this->query()->count();
537	826	}
538	827	}