wordpress-seo / src / builders / indexable-builder.php

indexable-builder.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 28.3, at src/builders/indexable-builder.php

481 lines 15.4 KB

Raw Download Zip

1	<?php
2
3	namespace Yoast\WP\SEO\Builders;
4
5	use Throwable;
6	use Yoast\WP\SEO\Exceptions\Indexable\Indexing_Failed_Exception;
7	use Yoast\WP\SEO\Exceptions\Indexable\Not_Built_Exception;
8	use Yoast\WP\SEO\Exceptions\Indexable\Source_Exception;
9	use Yoast\WP\SEO\Helpers\Indexable_Helper;
10	use Yoast\WP\SEO\Loggers\Logger;
11	use Yoast\WP\SEO\Models\Indexable;
12	use Yoast\WP\SEO\Repositories\Indexable_Repository;
13	use Yoast\WP\SEO\Services\Indexables\Indexable_Version_Manager;
14
15	/**
16	* Builder for the indexables.
17	*
18	* Creates all the indexables.
19	*/
20	class Indexable_Builder {
21
22	/**
23	* The author builder.
24	*
25	* @var Indexable_Author_Builder
26	*/
27	private $author_builder;
28
29	/**
30	* The post builder.
31	*
32	* @var Indexable_Post_Builder
33	*/
34	private $post_builder;
35
36	/**
37	* The term builder.
38	*
39	* @var Indexable_Term_Builder
40	*/
41	private $term_builder;
42
43	/**
44	* The home page builder.
45	*
46	* @var Indexable_Home_Page_Builder
47	*/
48	private $home_page_builder;
49
50	/**
51	* The post type archive builder.
52	*
53	* @var Indexable_Post_Type_Archive_Builder
54	*/
55	private $post_type_archive_builder;
56
57	/**
58	* The data archive builder.
59	*
60	* @var Indexable_Date_Archive_Builder
61	*/
62	private $date_archive_builder;
63
64	/**
65	* The system page builder.
66	*
67	* @var Indexable_System_Page_Builder
68	*/
69	private $system_page_builder;
70
71	/**
72	* The indexable hierarchy builder.
73	*
74	* @var Indexable_Hierarchy_Builder
75	*/
76	private $hierarchy_builder;
77
78	/**
79	* The primary term builder
80	*
81	* @var Primary_Term_Builder
82	*/
83	private $primary_term_builder;
84
85	/**
86	* The link builder
87	*
88	* @var Indexable_Link_Builder
89	*/
90	private $link_builder;
91
92	/**
93	* The indexable repository.
94	*
95	* @var Indexable_Repository
96	*/
97	private $indexable_repository;
98
99	/**
100	* The indexable helper.
101	*
102	* @var Indexable_Helper
103	*/
104	protected $indexable_helper;
105
106	/**
107	* The Indexable Version Manager.
108	*
109	* @var Indexable_Version_Manager
110	*/
111	protected $version_manager;
112
113	/**
114	* The logger.
115	*
116	* @var Logger
117	*/
118	protected $logger;
119
120	/**
121	* Returns the instance of this class constructed through the ORM Wrapper.
122	*
123	* @param Indexable_Author_Builder $author_builder The author builder for creating missing indexables.
124	* @param Indexable_Post_Builder $post_builder The post builder for creating missing indexables.
125	* @param Indexable_Term_Builder $term_builder The term builder for creating missing indexables.
126	* @param Indexable_Home_Page_Builder $home_page_builder The front page builder for creating missing indexables.
127	* @param Indexable_Post_Type_Archive_Builder $post_type_archive_builder The post type archive builder for creating missing indexables.
128	* @param Indexable_Date_Archive_Builder $date_archive_builder The date archive builder for creating missing indexables.
129	* @param Indexable_System_Page_Builder $system_page_builder The search result builder for creating missing indexables.
130	* @param Indexable_Hierarchy_Builder $hierarchy_builder The hierarchy builder for creating the indexable hierarchy.
131	* @param Primary_Term_Builder $primary_term_builder The primary term builder for creating primary terms for posts.
132	* @param Indexable_Helper $indexable_helper The indexable helper.
133	* @param Indexable_Version_Manager $version_manager The indexable version manager.
134	* @param Indexable_Link_Builder $link_builder The link builder for creating missing SEO links.
135	* @param Logger $logger The logger.
136	*/
137	public function __construct(
138	Indexable_Author_Builder $author_builder,
139	Indexable_Post_Builder $post_builder,
140	Indexable_Term_Builder $term_builder,
141	Indexable_Home_Page_Builder $home_page_builder,
142	Indexable_Post_Type_Archive_Builder $post_type_archive_builder,
143	Indexable_Date_Archive_Builder $date_archive_builder,
144	Indexable_System_Page_Builder $system_page_builder,
145	Indexable_Hierarchy_Builder $hierarchy_builder,
146	Primary_Term_Builder $primary_term_builder,
147	Indexable_Helper $indexable_helper,
148	Indexable_Version_Manager $version_manager,
149	Indexable_Link_Builder $link_builder,
150	Logger $logger
151	) {
152	$this->author_builder = $author_builder;
153	$this->post_builder = $post_builder;
154	$this->term_builder = $term_builder;
155	$this->home_page_builder = $home_page_builder;
156	$this->post_type_archive_builder = $post_type_archive_builder;
157	$this->date_archive_builder = $date_archive_builder;
158	$this->system_page_builder = $system_page_builder;
159	$this->hierarchy_builder = $hierarchy_builder;
160	$this->primary_term_builder = $primary_term_builder;
161	$this->indexable_helper = $indexable_helper;
162	$this->version_manager = $version_manager;
163	$this->link_builder = $link_builder;
164	$this->logger = $logger;
165	}
166
167	/**
168	* Sets the indexable repository. Done to avoid circular dependencies.
169	*
170	* @required
171	*
172	* @param Indexable_Repository $indexable_repository The indexable repository.
173	*
174	* @return void
175	*/
176	public function set_indexable_repository( Indexable_Repository $indexable_repository ) {
177	$this->indexable_repository = $indexable_repository;
178	}
179
180	/**
181	* Creates a clean copy of an Indexable to allow for later database operations.
182	*
183	* @param Indexable $indexable The Indexable to copy.
184	*
185	* @return bool\|Indexable
186	*/
187	protected function deep_copy_indexable( $indexable ) {
188	return $this->indexable_repository
189	->query()
190	->create( $indexable->as_array() );
191	}
192
193	/**
194	* Creates an indexable by its ID and type.
195	*
196	* @param int $object_id The indexable object ID.
197	* @param string $object_type The indexable object type.
198	* @param Indexable\|bool $indexable Optional. An existing indexable to overwrite.
199	*
200	* @return bool\|Indexable Instance of indexable. False when unable to build.
201	*/
202	public function build_for_id_and_type( $object_id, $object_type, $indexable = false ) {
203	$defaults = [
204	'object_type' => $object_type,
205	'object_id' => $object_id,
206	];
207
208	$indexable = $this->build( $indexable, $defaults );
209
210	return $indexable;
211	}
212
213	/**
214	* Creates an indexable for the homepage.
215	*
216	* @param Indexable\|bool $indexable Optional. An existing indexable to overwrite.
217	*
218	* @return Indexable The home page indexable.
219	*/
220	public function build_for_home_page( $indexable = false ) {
221	return $this->build( $indexable, [ 'object_type' => 'home-page' ] );
222	}
223
224	/**
225	* Creates an indexable for the date archive.
226	*
227	* @param Indexable\|bool $indexable Optional. An existing indexable to overwrite.
228	*
229	* @return Indexable The date archive indexable.
230	*/
231	public function build_for_date_archive( $indexable = false ) {
232	return $this->build( $indexable, [ 'object_type' => 'date-archive' ] );
233	}
234
235	/**
236	* Creates an indexable for a post type archive.
237	*
238	* @param string $post_type The post type.
239	* @param Indexable\|bool $indexable Optional. An existing indexable to overwrite.
240	*
241	* @return Indexable The post type archive indexable.
242	*/
243	public function build_for_post_type_archive( $post_type, $indexable = false ) {
244	$defaults = [
245	'object_type' => 'post-type-archive',
246	'object_sub_type' => $post_type,
247	];
248	return $this->build( $indexable, $defaults );
249	}
250
251	/**
252	* Creates an indexable for a system page.
253	*
254	* @param string $page_type The type of system page.
255	* @param Indexable\|bool $indexable Optional. An existing indexable to overwrite.
256	*
257	* @return Indexable The search result indexable.
258	*/
259	public function build_for_system_page( $page_type, $indexable = false ) {
260	$defaults = [
261	'object_type' => 'system-page',
262	'object_sub_type' => $page_type,
263	];
264	return $this->build( $indexable, $defaults );
265	}
266
267	/**
268	* Ensures we have a valid indexable. Creates one if false is passed.
269	*
270	* @param Indexable\|false $indexable The indexable.
271	* @param array<string, int\|string> $defaults The initial properties of the Indexable.
272	*
273	* @return Indexable The indexable.
274	*/
275	protected function ensure_indexable( $indexable, $defaults = [] ) {
276	if ( ! $indexable ) {
277	return $this->indexable_repository->query()->create( $defaults );
278	}
279
280	return $indexable;
281	}
282
283	/**
284	* Build and author indexable from an author id if it does not exist yet, or if the author indexable needs to be upgraded.
285	*
286	* @param int $author_id The author id.
287	*
288	* @return Indexable\|false The author indexable if it has been built, `false` if it could not be built.
289	*/
290	protected function maybe_build_author_indexable( $author_id ) {
291	$author_indexable = $this->indexable_repository->find_by_id_and_type(
292	$author_id,
293	'user',
294	false,
295	);
296	if ( ! $author_indexable \|\| $this->version_manager->indexable_needs_upgrade( $author_indexable ) ) {
297	// Try to build the author.
298	$author_defaults = [
299	'object_type' => 'user',
300	'object_id' => $author_id,
301	];
302	$author_indexable = $this->build( $author_indexable, $author_defaults );
303	}
304	return $author_indexable;
305	}
306
307	/**
308	* Checks if the indexable type is one that is not supposed to have object ID for.
309	*
310	* @param string $type The type of the indexable.
311	*
312	* @return bool Whether the indexable type is one that is not supposed to have object ID for.
313	*/
314	protected function is_type_with_no_id( $type ) {
315	return \in_array( $type, [ 'home-page', 'date-archive', 'post-type-archive', 'system-page' ], true );
316	}
317
318	// phpcs:disable Squiz.Commenting.FunctionCommentThrowTag.Missing -- Most exceptions are handled in the method; the unexpected-error catch deliberately re-throws after firing the failure hook.
319
320	/**
321	* Rebuilds an Indexable from scratch.
322	*
323	* @param Indexable $indexable The Indexable to (re)build.
324	* @param array<string, int\|string>\|null $defaults The object type of the Indexable.
325	*
326	* @return Indexable\|false The resulting Indexable.
327	*/
328	public function build( $indexable, $defaults = null ) {
329	// Backup the previous Indexable, if there was one.
330	$indexable_before = ( $indexable ) ? $this->deep_copy_indexable( $indexable ) : null;
331
332	// Make sure we have an Indexable to work with.
333	$indexable = $this->ensure_indexable( $indexable, $defaults );
334
335	try {
336	if ( $indexable->object_id === 0 ) {
337	throw Not_Built_Exception::invalid_object_id( $indexable->object_id );
338	}
339	switch ( $indexable->object_type ) {
340
341	case 'post':
342	$indexable = $this->post_builder->build( $indexable->object_id, $indexable );
343
344	// Save indexable, to make sure it can be queried when building related objects like the author indexable and hierarchy.
345	$indexable = $this->indexable_helper->save_indexable( $indexable, $indexable_before );
346
347	// For attachments, we have to make sure to patch any potentially previously cleaned up SEO links.
348	if ( \is_a( $indexable, Indexable::class ) && $indexable->object_sub_type === 'attachment' ) {
349	$this->link_builder->patch_seo_links( $indexable );
350	}
351
352	// Always rebuild the primary term.
353	$this->primary_term_builder->build( $indexable->object_id );
354
355	// Always rebuild the hierarchy; this needs the primary term to run correctly.
356	$this->hierarchy_builder->build( $indexable );
357
358	$this->maybe_build_author_indexable( $indexable->author_id );
359
360	// The indexable is already saved, so return early.
361	return $indexable;
362
363	case 'user':
364	$indexable = $this->author_builder->build( $indexable->object_id, $indexable );
365	break;
366
367	case 'term':
368	$indexable = $this->term_builder->build( $indexable->object_id, $indexable );
369
370	// Save indexable, to make sure it can be queried when building hierarchy.
371	$indexable = $this->indexable_helper->save_indexable( $indexable, $indexable_before );
372
373	$this->hierarchy_builder->build( $indexable );
374
375	// The indexable is already saved, so return early.
376	return $indexable;
377
378	case 'home-page':
379	$indexable = $this->home_page_builder->build( $indexable );
380	break;
381
382	case 'date-archive':
383	$indexable = $this->date_archive_builder->build( $indexable );
384	break;
385
386	case 'post-type-archive':
387	$indexable = $this->post_type_archive_builder->build( $indexable->object_sub_type, $indexable );
388	break;
389
390	case 'system-page':
391	$indexable = $this->system_page_builder->build( $indexable->object_sub_type, $indexable );
392	break;
393	}
394
395	return $this->indexable_helper->save_indexable( $indexable, $indexable_before );
396	} catch ( Source_Exception $exception ) {
397	if ( ! $this->is_type_with_no_id( $indexable->object_type ) && ! isset( $indexable->object_id ) ) {
398	return false;
399	}
400
401	/**
402	* The current indexable could not be indexed. Create a placeholder indexable, so we can
403	* skip this indexable in future indexing runs.
404	*
405	* @var Indexable $indexable
406	*/
407	$indexable = $this->ensure_indexable(
408	$indexable,
409	[
410	'object_id' => $indexable->object_id,
411	'object_type' => $indexable->object_type,
412	'post_status' => 'unindexed',
413	'version' => 0,
414	],
415	);
416	// If we already had an existing indexable, mark it as unindexed. We cannot rely on its validity anymore.
417	$indexable->post_status = 'unindexed';
418	// Make sure that the indexing process doesn't get stuck in a loop on this broken indexable.
419	$indexable = $this->version_manager->set_latest( $indexable );
420
421	return $this->indexable_helper->save_indexable( $indexable, $indexable_before );
422	} catch ( Not_Built_Exception $exception ) {
423	$this->logger->debug(
424	$exception->getMessage(),
425	[
426	'object_id' => $indexable->object_id,
427	'object_type' => $indexable->object_type,
428	'object_sub_type' => $indexable->object_sub_type,
429	'exception' => \get_class( $exception ),
430	],
431	);
432
433	return false;
434	} catch ( Indexing_Failed_Exception $exception ) {
435	// A nested build (e.g. the author indexable built during a post build) already logged the
436	// failure, fired the action and wrapped the original error, so pass it through untouched
437	// to keep the root failing object's identity and avoid reporting the failure twice.
438	throw $exception;
439	} catch ( Throwable $exception ) {
440	$indexing_failed_exception = new Indexing_Failed_Exception(
441	$indexable->object_id,
442	$indexable->object_type,
443	$indexable->object_sub_type,
444	$exception,
445	);
446
447	$this->logger->error(
448	$indexing_failed_exception->getMessage(),
449	[
450	'object_id' => $indexable->object_id,
451	'object_type' => $indexable->object_type,
452	'object_sub_type' => $indexable->object_sub_type,
453	'exception' => \get_class( $exception ),
454	],
455	);
456
457	/**
458	* Fires when an indexable could not be built because of an unexpected error.
459	*
460	* This action lets third parties observe build failures themselves.
461	*
462	* @param int\|null $object_id The object ID of the indexable that failed to build, or null for id-less object types.
463	* @param string $object_type The object type of the indexable that failed to build.
464	* @param string\|null $object_sub_type The object sub type of the indexable that failed to build.
465	* @param Throwable $exception The error that caused the failure.
466	*/
467	\do_action(
468	'wpseo_indexable_indexing_failed',
469	$indexable->object_id,
470	$indexable->object_type,
471	$indexable->object_sub_type,
472	$exception,
473	);
474
475	throw $indexing_failed_exception;
476	}
477	}
478
479	// phpcs:enable
480	}
481