← All changes | src/integrations/cleanup-integration.php +130 -107 18.2 → 28.5 View file →

		@@ -2,9 +2,10 @@
2	2
3	3	namespace Yoast\WP\SEO\Integrations;
4	4
5	5	use Closure;
6		-use Yoast\WP\Lib\Model;
	6	+use Yoast\WP\SEO\Helpers\Indexable_Helper;
	7	+use Yoast\WP\SEO\Repositories\Indexable_Cleanup_Repository;
7	8
8	9	/**
9	10	* Adds cleanup hooks.
10	11	*/
		@@ -12,21 +13,49 @@
12	13
13	14	/**
14	15	* Identifier used to determine the current task.
15	16	*/
16		- const CURRENT_TASK_OPTION = 'wpseo-cleanup-current-task';
	17	+ public const CURRENT_TASK_OPTION = 'wpseo-cleanup-current-task';
17	18
18	19	/**
19	20	* Identifier for the cron job.
20	21	*/
21		- const CRON_HOOK = 'wpseo_cleanup_cron';
	22	+ public const CRON_HOOK = 'wpseo_cleanup_cron';
22	23
23	24	/**
24	25	* Identifier for starting the cleanup.
25	26	*/
26		- const START_HOOK = 'wpseo_start_cleanup_indexables';
	27	+ public const START_HOOK = 'wpseo_start_cleanup_indexables';
27	28
28	29	/**
	30	+ * The indexable helper.
	31	+ *
	32	+ * @var Indexable_Helper
	33	+ */
	34	+ private $indexable_helper;
	35	+
	36	+ /**
	37	+ * The cleanup repository.
	38	+ *
	39	+ * @var Indexable_Cleanup_Repository
	40	+ */
	41	+ private $cleanup_repository;
	42	+
	43	+ /**
	44	+ * The constructor.
	45	+ *
	46	+ * @param Indexable_Cleanup_Repository $cleanup_repository The cleanup repository.
	47	+ * @param Indexable_Helper $indexable_helper The indexable helper.
	48	+ */
	49	+ public function __construct(
	50	+ Indexable_Cleanup_Repository $cleanup_repository,
	51	+ Indexable_Helper $indexable_helper
	52	+ ) {
	53	+ $this->cleanup_repository = $cleanup_repository;
	54	+ $this->indexable_helper = $indexable_helper;
	55	+ }
	56	+
	57	+ /**
29	58	* Initializes the integration.
30	59	*
31	60	* This is the place to register hooks and filters.
32	61	*
		@@ -40,9 +69,9 @@
40	69
41	70	/**
42	71	* Returns the conditionals based on which this loadable should be active.
43	72	*
44		- * @return array The array of conditionals.
	73	+ * @return array<string> The array of conditionals.
45	74	*/
46	75	public static function get_conditionals() {
47	76	return [];
48	77	}
		@@ -54,8 +83,13 @@
54	83	*/
55	84	public function run_cleanup() {
56	85	$this->reset_cleanup();
57	86
	87	+ if ( ! $this->indexable_helper->should_index_indexables() ) {
	88	+ \wp_unschedule_hook( self::START_HOOK );
	89	+ return;
	90	+ }
	91	+
58	92	$cleanups = $this->get_cleanup_tasks();
59	93	$limit = $this->get_limit();
60	94
61	95	foreach ( $cleanups as $name => $action ) {
		@@ -70,8 +104,9 @@
70	104	}
71	105
72	106	// There are more items to delete for the current cleanup job, start a cronjob at the specified job.
73	107	$this->start_cron_job( $name );
	108	+
74	109	return;
75	110	}
76	111	}
77	112
		@@ -79,31 +114,59 @@
79	114	* Returns an array of cleanup tasks.
80	115	*
81	116	* @return Closure[] The cleanup tasks.
82	117	*/
83		- protected function get_cleanup_tasks() {
	118	+ public function get_cleanup_tasks() {
84	119	return \array_merge(
85	120	[
86		- 'clean_indexables_with_object_type_and_object_sub_type_shop_order' => function( $limit ) {
87		- return $this->clean_indexables_with_object_type_and_object_sub_type( 'post', 'shop_order', $limit );
	121	+ 'clean_indexables_with_object_type_and_object_sub_type_shop_order' => function ( $limit ) {
	122	+ return $this->cleanup_repository->clean_indexables_with_object_type_and_object_sub_type( 'post', 'shop_order', $limit );
88	123	},
89		- 'clean_indexables_by_post_status_auto-draft' => function( $limit ) {
90		- return $this->clean_indexables_with_post_status( 'auto-draft', $limit );
	124	+ 'clean_indexables_by_post_status_auto-draft' => function ( $limit ) {
	125	+ return $this->cleanup_repository->clean_indexables_with_post_status( 'auto-draft', $limit );
91	126	},
	127	+ 'clean_indexables_for_non_publicly_viewable_post' => function ( $limit ) {
	128	+ return $this->cleanup_repository->clean_indexables_for_non_publicly_viewable_post( $limit );
	129	+ },
	130	+ 'clean_indexables_for_non_publicly_viewable_taxonomies' => function ( $limit ) {
	131	+ return $this->cleanup_repository->clean_indexables_for_non_publicly_viewable_taxonomies( $limit );
	132	+ },
	133	+ 'clean_indexables_for_non_publicly_viewable_post_type_archive_pages' => function ( $limit ) {
	134	+ return $this->cleanup_repository->clean_indexables_for_non_publicly_viewable_post_type_archive_pages( $limit );
	135	+ },
	136	+ 'clean_indexables_for_authors_archive_disabled' => function ( $limit ) {
	137	+ return $this->cleanup_repository->clean_indexables_for_authors_archive_disabled( $limit );
	138	+ },
	139	+ 'clean_indexables_for_authors_without_archive' => function ( $limit ) {
	140	+ return $this->cleanup_repository->clean_indexables_for_authors_without_archive( $limit );
	141	+ },
	142	+ 'update_indexables_author_to_reassigned' => function ( $limit ) {
	143	+ return $this->cleanup_repository->update_indexables_author_to_reassigned( $limit );
	144	+ },
	145	+ 'clean_orphaned_user_indexables_without_wp_user' => function ( $limit ) {
	146	+ return $this->cleanup_repository->clean_indexables_for_orphaned_users( $limit );
	147	+ },
	148	+ 'clean_orphaned_user_indexables_without_wp_post' => function ( $limit ) {
	149	+ return $this->cleanup_repository->clean_indexables_for_object_type_and_source_table( 'posts', 'ID', 'post', $limit );
	150	+ },
	151	+ 'clean_orphaned_user_indexables_without_wp_term' => function ( $limit ) {
	152	+ return $this->cleanup_repository->clean_indexables_for_object_type_and_source_table( 'terms', 'term_id', 'term', $limit );
	153	+ },
92	154	],
93		- $this->get_additional_tasks(),
	155	+ $this->get_additional_indexable_cleanups(),
94	156	[
95	157	/* These should always be the last ones to be called. */
96		- 'clean_orphaned_content_indexable_hierarchy' => function( $limit ) {
97		- return $this->cleanup_orphaned_from_table( 'Indexable_Hierarchy', 'indexable_id', $limit );
	158	+ 'clean_orphaned_content_indexable_hierarchy' => function ( $limit ) {
	159	+ return $this->cleanup_repository->cleanup_orphaned_from_table( 'Indexable_Hierarchy', 'indexable_id', $limit );
98	160	},
99		- 'clean_orphaned_content_seo_links_indexable_id' => function( $limit ) {
100		- return $this->cleanup_orphaned_from_table( 'SEO_Links', 'indexable_id', $limit );
	161	+ 'clean_orphaned_content_seo_links_indexable_id' => function ( $limit ) {
	162	+ return $this->cleanup_repository->cleanup_orphaned_from_table( 'SEO_Links', 'indexable_id', $limit );
101	163	},
102		- 'clean_orphaned_content_seo_links_target_indexable_id' => function( $limit ) {
103		- return $this->cleanup_orphaned_from_table( 'SEO_Links', 'target_indexable_id', $limit );
	164	+ 'clean_orphaned_content_seo_links_target_indexable_id' => function ( $limit ) {
	165	+ return $this->cleanup_repository->cleanup_orphaned_from_table( 'SEO_Links', 'target_indexable_id', $limit );
104	166	},
105		- ]
	167	+ ],
	168	+ $this->get_additional_misc_cleanups(),
106	169	);
107	170	}
108	171
109	172	/**
		@@ -108,19 +171,47 @@
108	171
109	172	/**
110	173	* Gets additional tasks from the 'wpseo_cleanup_tasks' filter.
111	174	*
112		- * @return Closure[] Associative array of cleanup functions.
	175	+ * @return Closure[] Associative array of indexable cleanup functions.
113	176	*/
114		- private function get_additional_tasks() {
	177	+ private function get_additional_indexable_cleanups() {
115	178
116	179	/**
117		- * Filter: Adds the possibility to add addition cleanup functions.
	180	+ * Filter: Adds the possibility to add additional indexable cleanup functions.
118	181	*
119		- * @api array Associative array with unique keys. Value should be a cleanup function that receives a limit.
	182	+ * @param array $additional_tasks Associative array with unique keys. Value should be a cleanup function that receives a limit.
120	183	*/
121	184	$additional_tasks = \apply_filters( 'wpseo_cleanup_tasks', [] );
122	185
	186	+ return $this->validate_additional_tasks( $additional_tasks );
	187	+ }
	188	+
	189	+ /**
	190	+ * Gets additional tasks from the 'wpseo_misc_cleanup_tasks' filter.
	191	+ *
	192	+ * @return Closure[] Associative array of indexable cleanup functions.
	193	+ */
	194	+ private function get_additional_misc_cleanups() {
	195	+
	196	+ /**
	197	+ * Filter: Adds the possibility to add additional non-indexable cleanup functions.
	198	+ *
	199	+ * @param array $additional_tasks Associative array with unique keys. Value should be a cleanup function that receives a limit.
	200	+ */
	201	+ $additional_tasks = \apply_filters( 'wpseo_misc_cleanup_tasks', [] );
	202	+
	203	+ return $this->validate_additional_tasks( $additional_tasks );
	204	+ }
	205	+
	206	+ /**
	207	+ * Validates the additional tasks.
	208	+ *
	209	+ * @param Closure[] $additional_tasks The additional tasks to validate.
	210	+ *
	211	+ * @return Closure[] The validated additional tasks.
	212	+ */
	213	+ private function validate_additional_tasks( $additional_tasks ) {
123	214	if ( ! \is_array( $additional_tasks ) ) {
124	215	return [];
125	216	}
126	217
		@@ -144,9 +235,9 @@
144	235	private function get_limit() {
145	236	/**
146	237	* Filter: Adds the possibility to limit the number of items that are deleted from the database on cleanup.
147	238	*
148		- * @api int $limit Maximum number of indexables to be cleaned up per query.
	239	+ * @param int $limit Maximum number of indexables to be cleaned up per query.
149	240	*/
150	241	$limit = \apply_filters( 'wpseo_cron_query_limit_size', 1000 );
151	242
152	243	if ( ! \is_int( $limit ) ) {
		@@ -168,18 +259,19 @@
168	259
169	260	/**
170	261	* Starts the cleanup cron job.
171	262	*
172		- * @param string $task_name The task name of the next cleanup task to run.
	263	+ * @param string $task_name The task name of the next cleanup task to run.
	264	+ * @param int $schedule_time The time in seconds to wait before running the first cron job. Default is 1 hour.
173	265	*
174	266	* @return void
175	267	*/
176		- private function start_cron_job( $task_name ) {
	268	+ public function start_cron_job( $task_name, $schedule_time = 3600 ) {
177	269	\update_option( self::CURRENT_TASK_OPTION, $task_name );
178	270	\wp_schedule_event(
179		- ( \time() + \HOUR_IN_SECONDS ),
	271	+ ( \time() + $schedule_time ),
180	272	'hourly',
181		- self::CRON_HOOK
	273	+ self::CRON_HOOK,
182	274	);
183	275	}
184	276
185	277	/**
		@@ -187,12 +279,19 @@
187	279	*
188	280	* @return void
189	281	*/
190	282	public function run_cleanup_cron() {
	283	+ if ( ! $this->indexable_helper->should_index_indexables() ) {
	284	+ $this->reset_cleanup();
	285	+
	286	+ return;
	287	+ }
	288	+
191	289	$current_task_name = \get_option( self::CURRENT_TASK_OPTION );
192	290
193	291	if ( $current_task_name === false ) {
194	292	$this->reset_cleanup();
	293	+
195	294	return;
196	295	}
197	296
198	297	$limit = $this->get_limit();
		@@ -215,8 +314,9 @@
215	314	$items_cleaned = $current_task( $limit );
216	315
217	316	if ( $items_cleaned === false ) {
218	317	$this->reset_cleanup();
	318	+
219	319	return;
220	320	}
221	321
222	322	if ( $items_cleaned === 0 ) {
		@@ -222,96 +322,19 @@
222	322	if ( $items_cleaned === 0 ) {
223	323	// Check if we are finished with all tasks.
224	324	if ( \next( $tasks ) === false ) {
225	325	$this->reset_cleanup();
	326	+
226	327	return;
227	328	}
228	329
229	330	// Continue with the next task next time the cron job is run.
230	331	\update_option( self::CURRENT_TASK_OPTION, \key( $tasks ) );
	332	+
231	333	return;
232	334	}
	335	+
233	336	// There were items deleted for the current task, continue with the same task next cron call.
234	337	return;
235	338	}
236		- }
237		-
238		- /**
239		- * Deletes rows from the indexable table depending on the object_type and object_sub_type.
240		- *
241		- * @param string $object_type The object type to query.
242		- * @param string $object_sub_type The object subtype to query.
243		- * @param int $limit The limit we'll apply to the delete query.
244		- *
245		- * @return int|bool The number of rows that was deleted or false if the query failed.
246		- */
247		- protected function clean_indexables_with_object_type_and_object_sub_type( $object_type, $object_sub_type, $limit ) {
248		- global $wpdb;
249		-
250		- $indexable_table = Model::get_table_name( 'Indexable' );
251		-
252		- // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Reason: There is no unescaped user input.
253		- $sql = $wpdb->prepare( "DELETE FROM $indexable_table WHERE object_type = %s AND object_sub_type = %s ORDER BY id LIMIT %d", $object_type, $object_sub_type, $limit );
254		- // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- Reason: Already prepared.
255		- return $wpdb->query( $sql );
256		- }
257		-
258		- /**
259		- * Deletes rows from the indexable table depending on the post_status.
260		- *
261		- * @param string $post_status The post status to query.
262		- * @param int $limit The limit we'll apply to the delete query.
263		- *
264		- * @return int|bool The number of rows that was deleted or false if the query failed.
265		- */
266		- protected function clean_indexables_with_post_status( $post_status, $limit ) {
267		- global $wpdb;
268		-
269		- $indexable_table = Model::get_table_name( 'Indexable' );
270		-
271		- // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Reason: There is no unescaped user input.
272		- $sql = $wpdb->prepare( "DELETE FROM $indexable_table WHERE object_type = 'post' AND post_status = %s ORDER BY id LIMIT %d", $post_status, $limit );
273		- // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- Reason: Already prepared.
274		- return $wpdb->query( $sql );
275		- }
276		-
277		- /**
278		- * Cleans orphaned rows from a yoast table.
279		- *
280		- * @param string $table The table to clean up.
281		- * @param string $column The table column the cleanup will rely on.
282		- * @param int $limit The limit we'll apply to the queries.
283		- *
284		- * @return int|bool The number of deleted rows, false if the query fails.
285		- */
286		- protected function cleanup_orphaned_from_table( $table, $column, $limit ) {
287		- global $wpdb;
288		-
289		- $table = Model::get_table_name( $table );
290		- $indexable_table = Model::get_table_name( 'Indexable' );
291		-
292		- // Warning: If this query is changed, make sure to update the query in cleanup_orphaned_from_table in Premium as well.
293		- // phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Reason: There is no unescaped user input.
294		- $query = $wpdb->prepare(
295		- "
296		- SELECT table_to_clean.{$column}
297		- FROM {$table} table_to_clean
298		- LEFT JOIN {$indexable_table} AS indexable_table
299		- ON table_to_clean.{$column} = indexable_table.id
300		- WHERE indexable_table.id IS NULL
301		- AND table_to_clean.{$column} IS NOT NULL
302		- LIMIT %d",
303		- $limit
304		- );
305		- // phpcs:enable
306		-
307		- // phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- Reason: Already prepared.
308		- $orphans = $wpdb->get_col( $query );
309		-
310		- if ( empty( $orphans ) ) {
311		- return 0;
312		- }
313		-
314		- // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- Reason: Already prepared.
315		- return $wpdb->query( "DELETE FROM $table WHERE {$column} IN( " . \implode( ',', $orphans ) . ' )' );
316	339	}
317	340	}