wordpress-seo / src / integrations / admin / background-indexing-integration.php

background-indexing-integration.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 28.3, at src/integrations/admin/background-indexing-integration.php

340 lines 10.8 KB

Raw Download Zip

1	<?php
2
3	namespace Yoast\WP\SEO\Integrations\Admin;
4
5	use Yoast\WP\SEO\Actions\Indexing\Indexable_Indexing_Complete_Action;
6	use Yoast\WP\SEO\Actions\Indexing\Indexation_Action_Interface;
7	use Yoast\WP\SEO\Conditionals\Get_Request_Conditional;
8	use Yoast\WP\SEO\Conditionals\Migrations_Conditional;
9	use Yoast\WP\SEO\Conditionals\WP_CRON_Enabled_Conditional;
10	use Yoast\WP\SEO\Conditionals\Yoast_Admin_And_Dashboard_Conditional;
11	use Yoast\WP\SEO\Helpers\Indexable_Helper;
12	use Yoast\WP\SEO\Helpers\Indexing_Helper;
13	use Yoast\WP\SEO\Integrations\Integration_Interface;
14
15	/**
16	* Class Background_Indexing_Integration.
17	*
18	* @package Yoast\WP\SEO\Integrations\Admin
19	*/
20	class Background_Indexing_Integration implements Integration_Interface {
21
22	/**
23	* Represents the indexing completed action.
24	*
25	* @var Indexable_Indexing_Complete_Action
26	*/
27	protected $complete_indexation_action;
28
29	/**
30	* Represents the indexing helper.
31	*
32	* @var Indexing_Helper
33	*/
34	protected $indexing_helper;
35
36	/**
37	* An object that checks if we are on the Yoast admin or on the dashboard page.
38	*
39	* @var Yoast_Admin_And_Dashboard_Conditional
40	*/
41	protected $yoast_admin_and_dashboard_conditional;
42
43	/**
44	* All available indexing actions.
45	*
46	* @var Indexation_Action_Interface[]
47	*/
48	protected $indexing_actions;
49
50	/**
51	* An object that checks if we are handling a GET request.
52	*
53	* @var Get_Request_Conditional
54	*/
55	private $get_request_conditional;
56
57	/**
58	* An object that checks if WP_CRON is enabled.
59	*
60	* @var WP_CRON_Enabled_Conditional
61	*/
62	private $wp_cron_enabled_conditional;
63
64	/**
65	* The indexable helper
66	*
67	* @var Indexable_Helper
68	*/
69	private $indexable_helper;
70
71	/**
72	* Shutdown_Indexing_Integration constructor.
73	*
74	* @param Indexable_Indexing_Complete_Action $complete_indexation_action The complete indexing action.
75	* @param Indexing_Helper $indexing_helper The indexing helper.
76	* @param Indexable_Helper $indexable_helper The indexable helper.
77	* @param Yoast_Admin_And_Dashboard_Conditional $yoast_admin_and_dashboard_conditional An object that checks if we are on the Yoast admin or on the dashboard page.
78	* @param Get_Request_Conditional $get_request_conditional An object that checks if we are handling a GET request.
79	* @param WP_CRON_Enabled_Conditional $wp_cron_enabled_conditional An object that checks if WP_CRON is enabled.
80	* @param Indexation_Action_Interface ...$indexing_actions A list of all available indexing actions.
81	*/
82	public function __construct(
83	Indexable_Indexing_Complete_Action $complete_indexation_action,
84	Indexing_Helper $indexing_helper,
85	Indexable_Helper $indexable_helper,
86	Yoast_Admin_And_Dashboard_Conditional $yoast_admin_and_dashboard_conditional,
87	Get_Request_Conditional $get_request_conditional,
88	WP_CRON_Enabled_Conditional $wp_cron_enabled_conditional,
89	Indexation_Action_Interface ...$indexing_actions
90	) {
91	$this->indexing_actions = $indexing_actions;
92	$this->complete_indexation_action = $complete_indexation_action;
93	$this->indexing_helper = $indexing_helper;
94	$this->indexable_helper = $indexable_helper;
95	$this->yoast_admin_and_dashboard_conditional = $yoast_admin_and_dashboard_conditional;
96	$this->get_request_conditional = $get_request_conditional;
97	$this->wp_cron_enabled_conditional = $wp_cron_enabled_conditional;
98	}
99
100	/**
101	* Returns the conditionals based on which this integration should be active.
102	*
103	* @return array The array of conditionals.
104	*/
105	public static function get_conditionals() {
106	return [
107	Migrations_Conditional::class,
108	];
109	}
110
111	/**
112	* Register hooks.
113	*
114	* @return void
115	*/
116	public function register_hooks() {
117	\add_action( 'admin_init', [ $this, 'register_shutdown_indexing' ] );
118	\add_action( 'wpseo_indexable_index_batch', [ $this, 'index' ] );
119	// phpcs:ignore WordPress.WP.CronInterval -- The sniff doesn't understand values with parentheses. https://github.com/WordPress/WordPress-Coding-Standards/issues/2025
120	\add_filter( 'cron_schedules', [ $this, 'add_cron_schedule' ] );
121	\add_action( 'admin_init', [ $this, 'schedule_cron_indexing' ], 11 );
122
123	$this->add_limit_filters();
124	}
125
126	/**
127	* Adds the filters that change the indexing limits.
128	*
129	* @return void
130	*/
131	public function add_limit_filters() {
132	\add_filter( 'wpseo_post_indexation_limit', [ $this, 'throttle_cron_indexing' ] );
133	\add_filter( 'wpseo_post_type_archive_indexation_limit', [ $this, 'throttle_cron_indexing' ] );
134	\add_filter( 'wpseo_term_indexation_limit', [ $this, 'throttle_cron_indexing' ] );
135	\add_filter( 'wpseo_prominent_words_indexation_limit', [ $this, 'throttle_cron_indexing' ] );
136	\add_filter( 'wpseo_link_indexing_limit', [ $this, 'throttle_cron_link_indexing' ] );
137	}
138
139	/**
140	* Enqueues the required scripts.
141	*
142	* @return void
143	*/
144	public function register_shutdown_indexing() {
145	if ( $this->should_index_on_shutdown( $this->get_shutdown_limit() ) ) {
146	$this->register_shutdown_function( 'index' );
147	}
148	}
149
150	/**
151	* Run a single indexing pass of each indexing action. Intended for use as a shutdown function.
152	*
153	* @return void
154	*/
155	public function index() {
156	if ( \wp_doing_cron() && ! $this->should_index_on_cron() ) {
157	$this->unschedule_cron_indexing();
158
159	return;
160	}
161
162	foreach ( $this->indexing_actions as $indexation_action ) {
163	$indexation_action->index();
164	}
165
166	if ( $this->indexing_helper->get_limited_filtered_unindexed_count_background( 1 ) === 0 ) {
167	// We set this as complete, even though prominent words might not be complete. But that's the way we always treated that.
168	$this->complete_indexation_action->complete();
169	}
170	}
171
172	/**
173	* Adds the 'Every fifteen minutes' cron schedule to WP-Cron.
174	*
175	* @param array $schedules The existing schedules.
176	*
177	* @return array The schedules containing the fifteen_minutes schedule.
178	*/
179	public function add_cron_schedule( $schedules ) {
180	if ( ! \is_array( $schedules ) ) {
181	return $schedules;
182	}
183
184	$schedules['fifteen_minutes'] = [
185	'interval' => ( 15 * \MINUTE_IN_SECONDS ),
186	'display' => \esc_html__( 'Every fifteen minutes', 'wordpress-seo' ),
187	];
188
189	return $schedules;
190	}
191
192	/**
193	* Schedule background indexing every 15 minutes if the index isn't already up to date.
194	*
195	* @return void
196	*/
197	public function schedule_cron_indexing() {
198	/**
199	* Filter: 'wpseo_unindexed_count_queries_ran' - Informs whether the expensive unindexed count queries have been ran already.
200	*
201	* @internal
202	*
203	* @param bool $have_queries_ran
204	*/
205	$have_queries_ran = \apply_filters( 'wpseo_unindexed_count_queries_ran', false );
206
207	if ( ( ! $this->yoast_admin_and_dashboard_conditional->is_met() \|\| ! $this->get_request_conditional->is_met() ) && ! $have_queries_ran ) {
208	return;
209	}
210
211	if ( ! \wp_next_scheduled( 'wpseo_indexable_index_batch' ) && $this->should_index_on_cron() ) {
212	\wp_schedule_event( ( \time() + \HOUR_IN_SECONDS ), 'fifteen_minutes', 'wpseo_indexable_index_batch' );
213	}
214	}
215
216	/**
217	* Limit cron indexing to 15 indexables per batch instead of 25.
218	*
219	* @param int $indexation_limit The current limit (filter input).
220	*
221	* @return int The new batch limit.
222	*/
223	public function throttle_cron_indexing( $indexation_limit ) {
224	if ( \wp_doing_cron() ) {
225	/**
226	* Filter: 'wpseo_cron_indexing_limit_size' - Adds the possibility to limit the number of items that are indexed when in cron action.
227	*
228	* @param int $limit Maximum number of indexables to be indexed per indexing action.
229	*/
230	return \apply_filters( 'wpseo_cron_indexing_limit_size', 15 );
231	}
232
233	return $indexation_limit;
234	}
235
236	/**
237	* Limit cron indexing to 3 links per batch instead of 5.
238	*
239	* @param int $link_indexation_limit The current limit (filter input).
240	*
241	* @return int The new batch limit.
242	*/
243	public function throttle_cron_link_indexing( $link_indexation_limit ) {
244	if ( \wp_doing_cron() ) {
245	/**
246	* Filter: 'wpseo_cron_link_indexing_limit_size' - Adds the possibility to limit the number of links that are indexed when in cron action.
247	*
248	* @param int $limit Maximum number of link indexables to be indexed per link indexing action.
249	*/
250	return \apply_filters( 'wpseo_cron_link_indexing_limit_size', 3 );
251	}
252
253	return $link_indexation_limit;
254	}
255
256	/**
257	* Determine whether cron indexation should be performed.
258	*
259	* @return bool Should cron indexation be performed.
260	*/
261	protected function should_index_on_cron() {
262	if ( ! $this->indexable_helper->should_index_indexables() ) {
263	return false;
264	}
265
266	// The filter supersedes everything when preventing cron indexation.
267	if ( \apply_filters( 'Yoast\WP\SEO\enable_cron_indexing', true ) !== true ) {
268	return false;
269	}
270
271	return $this->indexing_helper->get_limited_filtered_unindexed_count_background( 1 ) > 0;
272	}
273
274	/**
275	* Determine whether background indexation should be performed.
276	*
277	* @param int $shutdown_limit The shutdown limit used to determine whether indexation should be run.
278	*
279	* @return bool Should background indexation be performed.
280	*/
281	protected function should_index_on_shutdown( $shutdown_limit ) {
282	if ( ! $this->yoast_admin_and_dashboard_conditional->is_met() \|\| ! $this->get_request_conditional->is_met() ) {
283	return false;
284	}
285
286	if ( ! $this->indexable_helper->should_index_indexables() ) {
287	return false;
288	}
289
290	if ( $this->wp_cron_enabled_conditional->is_met() ) {
291	return false;
292	}
293
294	$total_unindexed = $this->indexing_helper->get_limited_filtered_unindexed_count_background( $shutdown_limit );
295	if ( $total_unindexed === 0 \|\| $total_unindexed > $shutdown_limit ) {
296	return false;
297	}
298
299	return true;
300	}
301
302	/**
303	* Retrieves the shutdown limit. This limit is the amount of indexables that is generated in the background.
304	*
305	* @return int The shutdown limit.
306	*/
307	protected function get_shutdown_limit() {
308	/**
309	* Filter 'wpseo_shutdown_indexation_limit' - Allow filtering the number of objects that can be indexed during shutdown.
310	*
311	* @param int $limit The maximum number of objects indexed.
312	*/
313	return \apply_filters( 'wpseo_shutdown_indexation_limit', 25 );
314	}
315
316	/**
317	* Removes the cron indexing job from the scheduled event queue.
318	*
319	* @return void
320	*/
321	protected function unschedule_cron_indexing() {
322	$scheduled = \wp_next_scheduled( 'wpseo_indexable_index_batch' );
323	if ( $scheduled ) {
324	\wp_unschedule_event( $scheduled, 'wpseo_indexable_index_batch' );
325	}
326	}
327
328	/**
329	* Registers a method to be executed on shutdown.
330	* This wrapper mostly exists for making this class more unittestable.
331	*
332	* @param string $method_name The name of the method on the current instance to register.
333	*
334	* @return void
335	*/
336	protected function register_shutdown_function( $method_name ) {
337	\register_shutdown_function( [ $this, $method_name ] );
338	}
339	}
340