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

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

616 lines 18.2 KB

Raw Download Zip

1	<?php
2
3	namespace Yoast\WP\SEO\Builders;
4
5	use WPSEO_Image_Utils;
6	use Yoast\WP\SEO\Helpers\Image_Helper;
7	use Yoast\WP\SEO\Helpers\Indexable_Helper;
8	use Yoast\WP\SEO\Helpers\Options_Helper;
9	use Yoast\WP\SEO\Helpers\Post_Helper;
10	use Yoast\WP\SEO\Helpers\Url_Helper;
11	use Yoast\WP\SEO\Images\Application\Image_Content_Extractor;
12	use Yoast\WP\SEO\Models\Indexable;
13	use Yoast\WP\SEO\Models\SEO_Links;
14	use Yoast\WP\SEO\Repositories\Indexable_Repository;
15	use Yoast\WP\SEO\Repositories\SEO_Links_Repository;
16
17	/**
18	* Indexable link builder.
19	*/
20	class Indexable_Link_Builder {
21
22	/**
23	* The SEO links repository.
24	*
25	* @var SEO_Links_Repository
26	*/
27	protected $seo_links_repository;
28
29	/**
30	* The url helper.
31	*
32	* @var Url_Helper
33	*/
34	protected $url_helper;
35
36	/**
37	* The image helper.
38	*
39	* @var Image_Helper
40	*/
41	protected $image_helper;
42
43	/**
44	* The indexable helper.
45	*
46	* @var Indexable_Helper
47	*/
48	protected $indexable_helper;
49
50	/**
51	* The post helper.
52	*
53	* @var Post_Helper
54	*/
55	protected $post_helper;
56
57	/**
58	* The options helper.
59	*
60	* @var Options_Helper
61	*/
62	protected $options_helper;
63
64	/**
65	* The indexable repository.
66	*
67	* @var Indexable_Repository
68	*/
69	protected $indexable_repository;
70
71	/**
72	* Class that finds all images in a content string and extracts them.
73	*
74	* @var Image_Content_Extractor
75	*/
76	private $image_content_extractor;
77
78	/**
79	* Indexable_Link_Builder constructor.
80	*
81	* @param SEO_Links_Repository $seo_links_repository The SEO links repository.
82	* @param Url_Helper $url_helper The URL helper.
83	* @param Post_Helper $post_helper The post helper.
84	* @param Options_Helper $options_helper The options helper.
85	* @param Indexable_Helper $indexable_helper The indexable helper.
86	*/
87	public function __construct(
88	SEO_Links_Repository $seo_links_repository,
89	Url_Helper $url_helper,
90	Post_Helper $post_helper,
91	Options_Helper $options_helper,
92	Indexable_Helper $indexable_helper,
93	Image_Content_Extractor $image_content_extractor
94	) {
95	$this->seo_links_repository = $seo_links_repository;
96	$this->url_helper = $url_helper;
97	$this->post_helper = $post_helper;
98	$this->options_helper = $options_helper;
99	$this->indexable_helper = $indexable_helper;
100	$this->image_content_extractor = $image_content_extractor;
101	}
102
103	/**
104	* Sets the indexable repository.
105	*
106	* @required
107	*
108	* @param Indexable_Repository $indexable_repository The indexable repository.
109	* @param Image_Helper $image_helper The image helper.
110	*
111	* @return void
112	*/
113	public function set_dependencies( Indexable_Repository $indexable_repository, Image_Helper $image_helper ) {
114	$this->indexable_repository = $indexable_repository;
115	$this->image_helper = $image_helper;
116	}
117
118	/**
119	* Builds the links for a post.
120	*
121	* @param Indexable $indexable The indexable.
122	* @param string $content The content. Expected to be unfiltered.
123	*
124	* @return SEO_Links[] The created SEO links.
125	*/
126	public function build( $indexable, $content ) {
127	if ( ! $this->indexable_helper->should_index_indexable( $indexable ) ) {
128	return [];
129	}
130
131	global $post;
132	if ( $indexable->object_type === 'post' ) {
133	$post_backup = $post;
134	// phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited -- To setup the post we need to do this explicitly.
135	$post = $this->post_helper->get_post( $indexable->object_id );
136	\setup_postdata( $post );
137
138	// The below hook primes the post and meta caches for all wp-image-<ID> images.
139	// So the image loop in create_links() hits warm caches — no additional priming is needed there.
140	$content = \apply_filters( 'the_content', $content );
141	\wp_reset_postdata();
142	// phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited -- To setup the post we need to do this explicitly.
143	$post = $post_backup;
144	}
145
146	$content = \str_replace( ']]>', ']]>', $content );
147	$links = $this->gather_links( $content );
148	$images = $this->image_content_extractor->gather_images( $content );
149
150	if ( empty( $links ) && empty( $images ) ) {
151	$indexable->link_count = 0;
152	$this->update_related_indexables( $indexable, [] );
153
154	return [];
155	}
156
157	if ( ! empty( $images ) && ( $indexable->open_graph_image_source === 'first-content-image' \|\| $indexable->twitter_image_source === 'first-content-image' ) ) {
158	$this->update_first_content_image( $indexable, $images );
159	}
160
161	$links = $this->create_links( $indexable, $links, $images );
162
163	$this->update_related_indexables( $indexable, $links );
164
165	$indexable->link_count = $this->get_internal_link_count( $links );
166
167	return $links;
168	}
169
170	/**
171	* Deletes all SEO links for an indexable.
172	*
173	* @param Indexable $indexable The indexable.
174	*
175	* @return void
176	*/
177	public function delete( $indexable ) {
178	$links = ( $this->seo_links_repository->find_all_by_indexable_id( $indexable->id ) );
179	$this->seo_links_repository->delete_all_by_indexable_id( $indexable->id );
180
181	$linked_indexable_ids = [];
182	foreach ( $links as $link ) {
183	if ( $link->target_indexable_id ) {
184	$linked_indexable_ids[] = $link->target_indexable_id;
185	}
186	}
187
188	$this->update_incoming_links_for_related_indexables( $linked_indexable_ids );
189	}
190
191	/**
192	* Fixes existing SEO links that are supposed to have a target indexable but don't, because of prior indexable
193	* cleanup.
194	*
195	* @param Indexable $indexable The indexable to be the target of SEO Links.
196	*
197	* @return void
198	*/
199	public function patch_seo_links( Indexable $indexable ) {
200	if ( ! empty( $indexable->id ) && ! empty( $indexable->object_id ) ) {
201	$links = $this->seo_links_repository->find_all_by_target_post_id( $indexable->object_id );
202
203	$updated_indexable = false;
204	foreach ( $links as $link ) {
205	if ( \is_a( $link, SEO_Links::class ) && empty( $link->target_indexable_id ) ) {
206	// Since that post ID exists in an SEO link but has no target_indexable_id, it's probably because of prior indexable cleanup.
207	$this->seo_links_repository->update_target_indexable_id( $link->id, $indexable->id );
208	$updated_indexable = true;
209	}
210	}
211
212	if ( $updated_indexable ) {
213	$updated_indexable_id = [ $indexable->id ];
214	$this->update_incoming_links_for_related_indexables( $updated_indexable_id );
215	}
216	}
217	}
218
219	/**
220	* Gathers all links from content.
221	*
222	* @param string $content The content.
223	*
224	* @return string[] An array of urls.
225	*/
226	protected function gather_links( $content ) {
227	if ( \strpos( $content, 'href' ) === false ) {
228	// Nothing to do.
229	return [];
230	}
231
232	$links = [];
233	$regexp = '<a\s[^>]href=("??)([^" >]?)\1[^>]*>';
234	// Used modifiers iU to match case insensitive and make greedy quantifiers lazy.
235	if ( \preg_match_all( "/$regexp/iU", $content, $matches, \PREG_SET_ORDER ) ) {
236	foreach ( $matches as $match ) {
237	$links[] = \trim( $match[2], "'" );
238	}
239	}
240
241	return $links;
242	}
243
244	/**
245	* Creates link models from lists of URLs and image sources.
246	*
247	* @param Indexable $indexable The indexable.
248	* @param string[] $links The link URLs.
249	* @param int[] $images The image sources.
250	*
251	* @return SEO_Links[] The link models.
252	*/
253	protected function create_links( $indexable, $links, $images ) {
254	$home_url = \wp_parse_url( \home_url() );
255	$current_url = \wp_parse_url( $indexable->permalink );
256	$links = \array_map(
257	function ( $link ) use ( $home_url, $indexable ) {
258	return $this->create_internal_link( $link, $home_url, $indexable );
259	},
260	$links,
261	);
262	// Filter out links to the same page with a fragment or query.
263	$links = \array_filter(
264	$links,
265	function ( $link ) use ( $current_url ) {
266	return $this->filter_link( $link, $current_url );
267	},
268	);
269
270	$image_links = [];
271	foreach ( $images as $image_url => $image_id ) {
272	$image_links[] = $this->create_internal_link( $image_url, $home_url, $indexable, true, $image_id );
273	}
274
275	return \array_merge( $links, $image_links );
276	}
277
278	/**
279	* Get the post ID based on the link's type and its target's permalink.
280	*
281	* @param string $type The type of link (either SEO_Links::TYPE_INTERNAL or SEO_Links::TYPE_INTERNAL_IMAGE).
282	* @param string $permalink The permalink of the link's target.
283	*
284	* @return int The post ID.
285	*/
286	protected function get_post_id( $type, $permalink ) {
287	if ( $type === SEO_Links::TYPE_INTERNAL ) {
288	return \url_to_postid( $permalink );
289	}
290
291	return $this->image_helper->get_attachment_by_url( $permalink );
292	}
293
294	/**
295	* Creates an internal link.
296	*
297	* @param string $url The url of the link.
298	* @param array $home_url The home url, as parsed by wp_parse_url.
299	* @param Indexable $indexable The indexable of the post containing the link.
300	* @param bool $is_image Whether or not the link is an image.
301	* @param int $image_id The ID of the internal image.
302	*
303	* @return SEO_Links The created link.
304	*/
305	protected function create_internal_link( $url, $home_url, $indexable, $is_image = false, $image_id = 0 ) {
306	$parsed_url = \wp_parse_url( $url );
307	$link_type = $this->url_helper->get_link_type( $parsed_url, $home_url, $is_image );
308
309	/**
310	* ORM representing a link in the SEO Links table.
311	*
312	* @var SEO_Links $model
313	*/
314	$model = $this->seo_links_repository->query()->create(
315	[
316	'url' => $url,
317	'type' => $link_type,
318	'indexable_id' => $indexable->id,
319	'post_id' => $indexable->object_id,
320	],
321	);
322
323	$model->parsed_url = $parsed_url;
324
325	if ( $model->type === SEO_Links::TYPE_INTERNAL ) {
326	$permalink = $this->build_permalink( $url, $home_url );
327
328	return $this->enhance_link_from_indexable( $model, $permalink );
329	}
330
331	if ( $model->type === SEO_Links::TYPE_INTERNAL_IMAGE ) {
332	$permalink = $this->build_permalink( $url, $home_url );
333
334	/** The `wpseo_force_creating_and_using_attachment_indexables` filter is documented in indexable-link-builder.php */
335	if ( ! $this->options_helper->get( 'disable-attachment' ) \|\| \apply_filters( 'wpseo_force_creating_and_using_attachment_indexables', false ) ) {
336	$model = $this->enhance_link_from_indexable( $model, $permalink );
337	}
338	else {
339	$target_post_id = ( $image_id !== 0 ) ? $image_id : WPSEO_Image_Utils::get_attachment_by_url( $permalink );
340
341	if ( ! empty( $target_post_id ) ) {
342	$model->target_post_id = $target_post_id;
343	}
344	}
345
346	if ( $model->target_post_id ) {
347	$file = \get_attached_file( $model->target_post_id );
348
349	if ( $file ) {
350	if ( \file_exists( $file ) ) {
351	$model->size = \filesize( $file );
352	}
353	else {
354	$model->size = null;
355	}
356
357	[ , $width, $height ] = \wp_get_attachment_image_src( $model->target_post_id, 'full' );
358	$model->width = $width;
359	$model->height = $height;
360	}
361	else {
362	$model->width = 0;
363	$model->height = 0;
364	$model->size = 0;
365	}
366	}
367	}
368
369	return $model;
370	}
371
372	/**
373	* Enhances the link model with information from its indexable.
374	*
375	* @param SEO_Links $model The link's model.
376	* @param string $permalink The link's permalink.
377	*
378	* @return SEO_Links The enhanced link model.
379	*/
380	protected function enhance_link_from_indexable( $model, $permalink ) {
381	$target = $this->indexable_repository->find_by_permalink( $permalink );
382
383	if ( ! $target ) {
384	// If target indexable cannot be found, create one based on the post's post ID.
385	$post_id = $this->get_post_id( $model->type, $permalink );
386	if ( $post_id && $post_id !== 0 ) {
387	$target = $this->indexable_repository->find_by_id_and_type( $post_id, 'post' );
388	}
389	}
390
391	if ( ! $target ) {
392	return $model;
393	}
394
395	$model->target_indexable_id = $target->id;
396	if ( $target->object_type === 'post' ) {
397	$model->target_post_id = $target->object_id;
398	}
399
400	if ( $model->target_indexable_id ) {
401	$model->language = $target->language;
402	$model->region = $target->region;
403	}
404
405	return $model;
406	}
407
408	/**
409	* Builds the link's permalink.
410	*
411	* @param string $url The url of the link.
412	* @param array $home_url The home url, as parsed by wp_parse_url.
413	*
414	* @return string The link's permalink.
415	*/
416	protected function build_permalink( $url, $home_url ) {
417	$permalink = $this->get_permalink( $url, $home_url );
418
419	if ( $this->url_helper->is_relative( $permalink ) ) {
420	// Make sure we're checking against the absolute URL, and add a trailing slash if the site has a trailing slash in its permalink settings.
421	$permalink = $this->url_helper->ensure_absolute_url( \user_trailingslashit( $permalink ) );
422	}
423
424	return $permalink;
425	}
426
427	/**
428	* Filters out links that point to the same page with a fragment or query.
429	*
430	* @param SEO_Links $link The link.
431	* @param array $current_url The url of the page the link is on, as parsed by wp_parse_url.
432	*
433	* @return bool Whether or not the link should be filtered.
434	*/
435	protected function filter_link( SEO_Links $link, $current_url ) {
436	$url = $link->parsed_url;
437
438	// Always keep external links.
439	if ( $link->type === SEO_Links::TYPE_EXTERNAL ) {
440	return true;
441	}
442
443	// Always keep links with an empty path or pointing to other pages.
444	if ( isset( $url['path'] ) ) {
445	return empty( $url['path'] ) \|\| $url['path'] !== $current_url['path'];
446	}
447
448	// Only keep links to the current page without a fragment or query.
449	return ( ! isset( $url['fragment'] ) && ! isset( $url['query'] ) );
450	}
451
452	/**
453	* Updates the link counts for related indexables.
454	*
455	* @param Indexable $indexable The indexable.
456	* @param SEO_Links[] $links The link models.
457	*
458	* @return void
459	*/
460	protected function update_related_indexables( $indexable, $links ) {
461	// Old links were only stored by post id, so remove all old seo links for this post that have no indexable id.
462	// This can be removed if we ever fully clear all seo links.
463	if ( $indexable->object_type === 'post' ) {
464	$this->seo_links_repository->delete_all_by_post_id_where_indexable_id_null( $indexable->object_id );
465	}
466
467	$updated_indexable_ids = [];
468	$old_links = $this->seo_links_repository->find_all_by_indexable_id( $indexable->id );
469
470	$links_to_remove = $this->links_diff( $old_links, $links );
471	$links_to_add = $this->links_diff( $links, $old_links );
472
473	if ( ! empty( $links_to_remove ) ) {
474	$this->seo_links_repository->delete_many_by_id( \wp_list_pluck( $links_to_remove, 'id' ) );
475	}
476
477	if ( ! empty( $links_to_add ) ) {
478	$this->seo_links_repository->insert_many( $links_to_add );
479	}
480
481	foreach ( $links_to_add as $link ) {
482	if ( $link->target_indexable_id ) {
483	$updated_indexable_ids[] = $link->target_indexable_id;
484	}
485	}
486	foreach ( $links_to_remove as $link ) {
487	if ( $link->target_indexable_id ) {
488	$updated_indexable_ids[] = $link->target_indexable_id;
489	}
490	}
491
492	$this->update_incoming_links_for_related_indexables( $updated_indexable_ids );
493	}
494
495	/**
496	* Creates a diff between two arrays of SEO links, based on urls.
497	*
498	* @param SEO_Links[] $links_a The array to compare.
499	* @param SEO_Links[] $links_b The array to compare against.
500	*
501	* @return SEO_Links[] Links that are in $links_a, but not in $links_b.
502	*/
503	protected function links_diff( $links_a, $links_b ) {
504	return \array_udiff(
505	$links_a,
506	$links_b,
507	static function ( SEO_Links $link_a, SEO_Links $link_b ) {
508	return \strcmp( $link_a->url, $link_b->url );
509	},
510	);
511	}
512
513	/**
514	* Returns the number of internal links in an array of link models.
515	*
516	* @param SEO_Links[] $links The link models.
517	*
518	* @return int The number of internal links.
519	*/
520	protected function get_internal_link_count( $links ) {
521	$internal_link_count = 0;
522
523	foreach ( $links as $link ) {
524	if ( $link->type === SEO_Links::TYPE_INTERNAL ) {
525	++$internal_link_count;
526	}
527	}
528
529	return $internal_link_count;
530	}
531
532	/**
533	* Returns a cleaned permalink for a given link.
534	*
535	* @param string $link The raw URL.
536	* @param array $home_url The home URL, as parsed by wp_parse_url.
537	*
538	* @return string The cleaned permalink.
539	*/
540	protected function get_permalink( $link, $home_url ) {
541	// Get rid of the #anchor.
542	$url_split = \explode( '#', $link );
543	$link = $url_split[0];
544
545	// Get rid of URL ?query=string.
546	$url_split = \explode( '?', $link );
547	$link = $url_split[0];
548
549	// Set the correct URL scheme.
550	$link = \set_url_scheme( $link, $home_url['scheme'] );
551
552	// Add 'www.' if it is absent and should be there.
553	if ( \strpos( $home_url['host'], 'www.' ) === 0 && \strpos( $link, '://www.' ) === false ) {
554	$link = \str_replace( '://', '://www.', $link );
555	}
556
557	// Strip 'www.' if it is present and shouldn't be.
558	if ( \strpos( $home_url['host'], 'www.' ) !== 0 ) {
559	$link = \str_replace( '://www.', '://', $link );
560	}
561
562	return $link;
563	}
564
565	/**
566	* Updates incoming link counts for related indexables.
567	*
568	* @param int[] $related_indexable_ids The IDs of all related indexables.
569	*
570	* @return void
571	*/
572	protected function update_incoming_links_for_related_indexables( $related_indexable_ids ) {
573	if ( empty( $related_indexable_ids ) ) {
574	return;
575	}
576
577	$counts = $this->seo_links_repository->get_incoming_link_counts_for_indexable_ids( $related_indexable_ids );
578
579	/**
580	* Fires to signal that incoming link counts for related indexables were updated.
581	*
582	* @param int[] $related_indexable_ids The related indexable Ids to this link change.
583	*
584	* @internal
585	*/
586	\do_action( 'wpseo_related_indexables_incoming_links_updated', $related_indexable_ids );
587
588	foreach ( $counts as $count ) {
589	$this->indexable_repository->update_incoming_link_count( $count['target_indexable_id'], $count['incoming'] );
590	}
591	}
592
593	/**
594	* Updates the image ids when the indexable images are marked as first content image.
595	*
596	* @param Indexable $indexable The indexable to change.
597	* @param array<string\|int> $images The image array.
598	*
599	* @return void
600	*/
601	public function update_first_content_image( Indexable $indexable, array $images ): void {
602	$current_open_graph_image = $indexable->open_graph_image;
603	$current_twitter_image = $indexable->twitter_image;
604
605	$first_content_image_url = \key( $images );
606	$first_content_image_id = \current( $images );
607
608	if ( $indexable->open_graph_image_source === 'first-content-image' && $current_open_graph_image === $first_content_image_url && ! empty( $first_content_image_id ) ) {
609	$indexable->open_graph_image_id = $first_content_image_id;
610	}
611	if ( $indexable->twitter_image_source === 'first-content-image' && $current_twitter_image === $first_content_image_url && ! empty( $first_content_image_id ) ) {
612	$indexable->twitter_image_id = $first_content_image_id;
613	}
614	}
615	}
616