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 27.8, at src/builders/indexable-link-builder.php

613 lines 18.0 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	$content = \apply_filters( 'the_content', $content );
138	\wp_reset_postdata();
139	// phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited -- To setup the post we need to do this explicitly.
140	$post = $post_backup;
141	}
142
143	$content = \str_replace( ']]>', ']]>', $content );
144	$links = $this->gather_links( $content );
145	$images = $this->image_content_extractor->gather_images( $content );
146
147	if ( empty( $links ) && empty( $images ) ) {
148	$indexable->link_count = 0;
149	$this->update_related_indexables( $indexable, [] );
150
151	return [];
152	}
153
154	if ( ! empty( $images ) && ( $indexable->open_graph_image_source === 'first-content-image' \|\| $indexable->twitter_image_source === 'first-content-image' ) ) {
155	$this->update_first_content_image( $indexable, $images );
156	}
157
158	$links = $this->create_links( $indexable, $links, $images );
159
160	$this->update_related_indexables( $indexable, $links );
161
162	$indexable->link_count = $this->get_internal_link_count( $links );
163
164	return $links;
165	}
166
167	/**
168	* Deletes all SEO links for an indexable.
169	*
170	* @param Indexable $indexable The indexable.
171	*
172	* @return void
173	*/
174	public function delete( $indexable ) {
175	$links = ( $this->seo_links_repository->find_all_by_indexable_id( $indexable->id ) );
176	$this->seo_links_repository->delete_all_by_indexable_id( $indexable->id );
177
178	$linked_indexable_ids = [];
179	foreach ( $links as $link ) {
180	if ( $link->target_indexable_id ) {
181	$linked_indexable_ids[] = $link->target_indexable_id;
182	}
183	}
184
185	$this->update_incoming_links_for_related_indexables( $linked_indexable_ids );
186	}
187
188	/**
189	* Fixes existing SEO links that are supposed to have a target indexable but don't, because of prior indexable
190	* cleanup.
191	*
192	* @param Indexable $indexable The indexable to be the target of SEO Links.
193	*
194	* @return void
195	*/
196	public function patch_seo_links( Indexable $indexable ) {
197	if ( ! empty( $indexable->id ) && ! empty( $indexable->object_id ) ) {
198	$links = $this->seo_links_repository->find_all_by_target_post_id( $indexable->object_id );
199
200	$updated_indexable = false;
201	foreach ( $links as $link ) {
202	if ( \is_a( $link, SEO_Links::class ) && empty( $link->target_indexable_id ) ) {
203	// Since that post ID exists in an SEO link but has no target_indexable_id, it's probably because of prior indexable cleanup.
204	$this->seo_links_repository->update_target_indexable_id( $link->id, $indexable->id );
205	$updated_indexable = true;
206	}
207	}
208
209	if ( $updated_indexable ) {
210	$updated_indexable_id = [ $indexable->id ];
211	$this->update_incoming_links_for_related_indexables( $updated_indexable_id );
212	}
213	}
214	}
215
216	/**
217	* Gathers all links from content.
218	*
219	* @param string $content The content.
220	*
221	* @return string[] An array of urls.
222	*/
223	protected function gather_links( $content ) {
224	if ( \strpos( $content, 'href' ) === false ) {
225	// Nothing to do.
226	return [];
227	}
228
229	$links = [];
230	$regexp = '<a\s[^>]href=("??)([^" >]?)\1[^>]*>';
231	// Used modifiers iU to match case insensitive and make greedy quantifiers lazy.
232	if ( \preg_match_all( "/$regexp/iU", $content, $matches, \PREG_SET_ORDER ) ) {
233	foreach ( $matches as $match ) {
234	$links[] = \trim( $match[2], "'" );
235	}
236	}
237
238	return $links;
239	}
240
241	/**
242	* Creates link models from lists of URLs and image sources.
243	*
244	* @param Indexable $indexable The indexable.
245	* @param string[] $links The link URLs.
246	* @param int[] $images The image sources.
247	*
248	* @return SEO_Links[] The link models.
249	*/
250	protected function create_links( $indexable, $links, $images ) {
251	$home_url = \wp_parse_url( \home_url() );
252	$current_url = \wp_parse_url( $indexable->permalink );
253	$links = \array_map(
254	function ( $link ) use ( $home_url, $indexable ) {
255	return $this->create_internal_link( $link, $home_url, $indexable );
256	},
257	$links,
258	);
259	// Filter out links to the same page with a fragment or query.
260	$links = \array_filter(
261	$links,
262	function ( $link ) use ( $current_url ) {
263	return $this->filter_link( $link, $current_url );
264	},
265	);
266
267	$image_links = [];
268	foreach ( $images as $image_url => $image_id ) {
269	$image_links[] = $this->create_internal_link( $image_url, $home_url, $indexable, true, $image_id );
270	}
271
272	return \array_merge( $links, $image_links );
273	}
274
275	/**
276	* Get the post ID based on the link's type and its target's permalink.
277	*
278	* @param string $type The type of link (either SEO_Links::TYPE_INTERNAL or SEO_Links::TYPE_INTERNAL_IMAGE).
279	* @param string $permalink The permalink of the link's target.
280	*
281	* @return int The post ID.
282	*/
283	protected function get_post_id( $type, $permalink ) {
284	if ( $type === SEO_Links::TYPE_INTERNAL ) {
285	return \url_to_postid( $permalink );
286	}
287
288	return $this->image_helper->get_attachment_by_url( $permalink );
289	}
290
291	/**
292	* Creates an internal link.
293	*
294	* @param string $url The url of the link.
295	* @param array $home_url The home url, as parsed by wp_parse_url.
296	* @param Indexable $indexable The indexable of the post containing the link.
297	* @param bool $is_image Whether or not the link is an image.
298	* @param int $image_id The ID of the internal image.
299	*
300	* @return SEO_Links The created link.
301	*/
302	protected function create_internal_link( $url, $home_url, $indexable, $is_image = false, $image_id = 0 ) {
303	$parsed_url = \wp_parse_url( $url );
304	$link_type = $this->url_helper->get_link_type( $parsed_url, $home_url, $is_image );
305
306	/**
307	* ORM representing a link in the SEO Links table.
308	*
309	* @var SEO_Links $model
310	*/
311	$model = $this->seo_links_repository->query()->create(
312	[
313	'url' => $url,
314	'type' => $link_type,
315	'indexable_id' => $indexable->id,
316	'post_id' => $indexable->object_id,
317	],
318	);
319
320	$model->parsed_url = $parsed_url;
321
322	if ( $model->type === SEO_Links::TYPE_INTERNAL ) {
323	$permalink = $this->build_permalink( $url, $home_url );
324
325	return $this->enhance_link_from_indexable( $model, $permalink );
326	}
327
328	if ( $model->type === SEO_Links::TYPE_INTERNAL_IMAGE ) {
329	$permalink = $this->build_permalink( $url, $home_url );
330
331	/** The `wpseo_force_creating_and_using_attachment_indexables` filter is documented in indexable-link-builder.php */
332	if ( ! $this->options_helper->get( 'disable-attachment' ) \|\| \apply_filters( 'wpseo_force_creating_and_using_attachment_indexables', false ) ) {
333	$model = $this->enhance_link_from_indexable( $model, $permalink );
334	}
335	else {
336	$target_post_id = ( $image_id !== 0 ) ? $image_id : WPSEO_Image_Utils::get_attachment_by_url( $permalink );
337
338	if ( ! empty( $target_post_id ) ) {
339	$model->target_post_id = $target_post_id;
340	}
341	}
342
343	if ( $model->target_post_id ) {
344	$file = \get_attached_file( $model->target_post_id );
345
346	if ( $file ) {
347	if ( \file_exists( $file ) ) {
348	$model->size = \filesize( $file );
349	}
350	else {
351	$model->size = null;
352	}
353
354	[ , $width, $height ] = \wp_get_attachment_image_src( $model->target_post_id, 'full' );
355	$model->width = $width;
356	$model->height = $height;
357	}
358	else {
359	$model->width = 0;
360	$model->height = 0;
361	$model->size = 0;
362	}
363	}
364	}
365
366	return $model;
367	}
368
369	/**
370	* Enhances the link model with information from its indexable.
371	*
372	* @param SEO_Links $model The link's model.
373	* @param string $permalink The link's permalink.
374	*
375	* @return SEO_Links The enhanced link model.
376	*/
377	protected function enhance_link_from_indexable( $model, $permalink ) {
378	$target = $this->indexable_repository->find_by_permalink( $permalink );
379
380	if ( ! $target ) {
381	// If target indexable cannot be found, create one based on the post's post ID.
382	$post_id = $this->get_post_id( $model->type, $permalink );
383	if ( $post_id && $post_id !== 0 ) {
384	$target = $this->indexable_repository->find_by_id_and_type( $post_id, 'post' );
385	}
386	}
387
388	if ( ! $target ) {
389	return $model;
390	}
391
392	$model->target_indexable_id = $target->id;
393	if ( $target->object_type === 'post' ) {
394	$model->target_post_id = $target->object_id;
395	}
396
397	if ( $model->target_indexable_id ) {
398	$model->language = $target->language;
399	$model->region = $target->region;
400	}
401
402	return $model;
403	}
404
405	/**
406	* Builds the link's permalink.
407	*
408	* @param string $url The url of the link.
409	* @param array $home_url The home url, as parsed by wp_parse_url.
410	*
411	* @return string The link's permalink.
412	*/
413	protected function build_permalink( $url, $home_url ) {
414	$permalink = $this->get_permalink( $url, $home_url );
415
416	if ( $this->url_helper->is_relative( $permalink ) ) {
417	// 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.
418	$permalink = $this->url_helper->ensure_absolute_url( \user_trailingslashit( $permalink ) );
419	}
420
421	return $permalink;
422	}
423
424	/**
425	* Filters out links that point to the same page with a fragment or query.
426	*
427	* @param SEO_Links $link The link.
428	* @param array $current_url The url of the page the link is on, as parsed by wp_parse_url.
429	*
430	* @return bool Whether or not the link should be filtered.
431	*/
432	protected function filter_link( SEO_Links $link, $current_url ) {
433	$url = $link->parsed_url;
434
435	// Always keep external links.
436	if ( $link->type === SEO_Links::TYPE_EXTERNAL ) {
437	return true;
438	}
439
440	// Always keep links with an empty path or pointing to other pages.
441	if ( isset( $url['path'] ) ) {
442	return empty( $url['path'] ) \|\| $url['path'] !== $current_url['path'];
443	}
444
445	// Only keep links to the current page without a fragment or query.
446	return ( ! isset( $url['fragment'] ) && ! isset( $url['query'] ) );
447	}
448
449	/**
450	* Updates the link counts for related indexables.
451	*
452	* @param Indexable $indexable The indexable.
453	* @param SEO_Links[] $links The link models.
454	*
455	* @return void
456	*/
457	protected function update_related_indexables( $indexable, $links ) {
458	// Old links were only stored by post id, so remove all old seo links for this post that have no indexable id.
459	// This can be removed if we ever fully clear all seo links.
460	if ( $indexable->object_type === 'post' ) {
461	$this->seo_links_repository->delete_all_by_post_id_where_indexable_id_null( $indexable->object_id );
462	}
463
464	$updated_indexable_ids = [];
465	$old_links = $this->seo_links_repository->find_all_by_indexable_id( $indexable->id );
466
467	$links_to_remove = $this->links_diff( $old_links, $links );
468	$links_to_add = $this->links_diff( $links, $old_links );
469
470	if ( ! empty( $links_to_remove ) ) {
471	$this->seo_links_repository->delete_many_by_id( \wp_list_pluck( $links_to_remove, 'id' ) );
472	}
473
474	if ( ! empty( $links_to_add ) ) {
475	$this->seo_links_repository->insert_many( $links_to_add );
476	}
477
478	foreach ( $links_to_add as $link ) {
479	if ( $link->target_indexable_id ) {
480	$updated_indexable_ids[] = $link->target_indexable_id;
481	}
482	}
483	foreach ( $links_to_remove as $link ) {
484	if ( $link->target_indexable_id ) {
485	$updated_indexable_ids[] = $link->target_indexable_id;
486	}
487	}
488
489	$this->update_incoming_links_for_related_indexables( $updated_indexable_ids );
490	}
491
492	/**
493	* Creates a diff between two arrays of SEO links, based on urls.
494	*
495	* @param SEO_Links[] $links_a The array to compare.
496	* @param SEO_Links[] $links_b The array to compare against.
497	*
498	* @return SEO_Links[] Links that are in $links_a, but not in $links_b.
499	*/
500	protected function links_diff( $links_a, $links_b ) {
501	return \array_udiff(
502	$links_a,
503	$links_b,
504	static function ( SEO_Links $link_a, SEO_Links $link_b ) {
505	return \strcmp( $link_a->url, $link_b->url );
506	},
507	);
508	}
509
510	/**
511	* Returns the number of internal links in an array of link models.
512	*
513	* @param SEO_Links[] $links The link models.
514	*
515	* @return int The number of internal links.
516	*/
517	protected function get_internal_link_count( $links ) {
518	$internal_link_count = 0;
519
520	foreach ( $links as $link ) {
521	if ( $link->type === SEO_Links::TYPE_INTERNAL ) {
522	++$internal_link_count;
523	}
524	}
525
526	return $internal_link_count;
527	}
528
529	/**
530	* Returns a cleaned permalink for a given link.
531	*
532	* @param string $link The raw URL.
533	* @param array $home_url The home URL, as parsed by wp_parse_url.
534	*
535	* @return string The cleaned permalink.
536	*/
537	protected function get_permalink( $link, $home_url ) {
538	// Get rid of the #anchor.
539	$url_split = \explode( '#', $link );
540	$link = $url_split[0];
541
542	// Get rid of URL ?query=string.
543	$url_split = \explode( '?', $link );
544	$link = $url_split[0];
545
546	// Set the correct URL scheme.
547	$link = \set_url_scheme( $link, $home_url['scheme'] );
548
549	// Add 'www.' if it is absent and should be there.
550	if ( \strpos( $home_url['host'], 'www.' ) === 0 && \strpos( $link, '://www.' ) === false ) {
551	$link = \str_replace( '://', '://www.', $link );
552	}
553
554	// Strip 'www.' if it is present and shouldn't be.
555	if ( \strpos( $home_url['host'], 'www.' ) !== 0 ) {
556	$link = \str_replace( '://www.', '://', $link );
557	}
558
559	return $link;
560	}
561
562	/**
563	* Updates incoming link counts for related indexables.
564	*
565	* @param int[] $related_indexable_ids The IDs of all related indexables.
566	*
567	* @return void
568	*/
569	protected function update_incoming_links_for_related_indexables( $related_indexable_ids ) {
570	if ( empty( $related_indexable_ids ) ) {
571	return;
572	}
573
574	$counts = $this->seo_links_repository->get_incoming_link_counts_for_indexable_ids( $related_indexable_ids );
575
576	/**
577	* Fires to signal that incoming link counts for related indexables were updated.
578	*
579	* @param int[] $related_indexable_ids The related indexable Ids to this link change.
580	*
581	* @internal
582	*/
583	\do_action( 'wpseo_related_indexables_incoming_links_updated', $related_indexable_ids );
584
585	foreach ( $counts as $count ) {
586	$this->indexable_repository->update_incoming_link_count( $count['target_indexable_id'], $count['incoming'] );
587	}
588	}
589
590	/**
591	* Updates the image ids when the indexable images are marked as first content image.
592	*
593	* @param Indexable $indexable The indexable to change.
594	* @param array<string\|int> $images The image array.
595	*
596	* @return void
597	*/
598	public function update_first_content_image( Indexable $indexable, array $images ): void {
599	$current_open_graph_image = $indexable->open_graph_image;
600	$current_twitter_image = $indexable->twitter_image;
601
602	$first_content_image_url = \key( $images );
603	$first_content_image_id = \current( $images );
604
605	if ( $indexable->open_graph_image_source === 'first-content-image' && $current_open_graph_image === $first_content_image_url && ! empty( $first_content_image_id ) ) {
606	$indexable->open_graph_image_id = $first_content_image_id;
607	}
608	if ( $indexable->twitter_image_source === 'first-content-image' && $current_twitter_image === $first_content_image_url && ! empty( $first_content_image_id ) ) {
609	$indexable->twitter_image_id = $first_content_image_id;
610	}
611	}
612	}
613