wordpress-seo / src / ai / content-planner / infrastructure / recent-content / recent-content-collector.php

recent-content-collector.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI trunk, at src/ai/content-planner/infrastructure/recent-content/recent-content-collector.php

167 lines 4.9 KB

Raw Download Zip

1	<?php
2	// phpcs:disable Yoast.NamingConventions.NamespaceName.TooLong -- Needed in the folder structure.
3	// phpcs:disable Yoast.NamingConventions.NamespaceName.MaxExceeded
4
5	namespace Yoast\WP\SEO\AI\Content_Planner\Infrastructure\Recent_Content;
6
7	use WP_Term;
8	use WPSEO_Meta;
9	use Yoast\WP\SEO\AI\Content_Planner\Domain\Category;
10	use Yoast\WP\SEO\AI\Content_Planner\Domain\Post;
11	use Yoast\WP\SEO\AI\Content_Planner\Domain\Post_List;
12	use Yoast\WP\SEO\Models\Indexable;
13	use Yoast\WP\SEO\Repositories\Indexable_Repository;
14	use Yoast\WP\SEO\Repositories\Primary_Term_Repository;
15
16	/**
17	* Collects the most recent published posts from indexables.
18	*/
19	class Recent_Content_Collector {
20
21	/**
22	* The maximum number of posts to collect.
23	*/
24	private const LIMIT = 100;
25	/**
26	* The valid schema types that we want to send to the API. Otherwise this field is omitted.
27	*/
28	public const VALID_SCHEMA_TYPES = [
29	'SocialMediaPosting',
30	'NewsArticle',
31	'AdvertiserContentArticle',
32	'SatiricalArticle',
33	'ScholarlyArticle',
34	'TechArticle',
35	'Report',
36	];
37
38	/**
39	* The indexable repository.
40	*
41	* @var Indexable_Repository
42	*/
43	private $indexable_repository;
44
45	/**
46	* The primary term repository.
47	*
48	* @var Primary_Term_Repository
49	*/
50	private $primary_term_repository;
51
52	/**
53	* The constructor.
54	*
55	* @param Indexable_Repository $indexable_repository The indexable repository.
56	* @param Primary_Term_Repository $primary_term_repository The primary term repository.
57	*/
58	public function __construct(
59	Indexable_Repository $indexable_repository,
60	Primary_Term_Repository $primary_term_repository
61	) {
62	$this->indexable_repository = $indexable_repository;
63	$this->primary_term_repository = $primary_term_repository;
64	}
65
66	/**
67	* Collects the 100 most recent published, indexed posts.
68	*
69	* @param string $post_type The post type to collect.
70	*
71	* @return Post_List The list of recent posts.
72	*/
73	public function collect( string $post_type ): Post_List {
74	$results = $this->indexable_repository->get_recently_modified_posts( $post_type, self::LIMIT, false );
75
76	return $this->map_to_post_list( $results );
77	}
78
79	/**
80	* Collects the most recent indexed about page.
81	*
82	* @param string $post_type The post type to collect.
83	*
84	* @return array<string>\|false The about page needed information.
85	*/
86	public function collect_about_page( string $post_type ) {
87	$indexable = $this->indexable_repository->get_most_recent_about_page( $post_type );
88
89	if ( $indexable ) {
90	return [
91	'title' => ( $indexable->breadcrumb_title ?? '' ),
92	'description' => ( $indexable->description ?? '' ),
93	];
94	}
95
96	return false;
97	}
98
99	/**
100	* Maps indexable results to a Post_List.
101	*
102	* @param Indexable[] $indexables The indexable results.
103	*
104	* @return Post_List The mapped post list.
105	*/
106	private function map_to_post_list( array $indexables ): Post_List {
107	$post_list = new Post_List();
108
109	foreach ( $indexables as $indexable ) {
110	$post_list->add(
111	new Post(
112	( $indexable->breadcrumb_title ?? '' ),
113	( $indexable->description ?? '' ),
114	$this->get_primary_category( $indexable->object_id ?? 0 ),
115	( $indexable->primary_focus_keyword ?? '' ),
116	( $indexable->is_cornerstone ?? false ),
117	( $indexable->object_last_modified ?? '' ),
118	( \in_array( $indexable->schema_article_type, self::VALID_SCHEMA_TYPES, true ) ? $indexable->schema_article_type : null ),
119	),
120	);
121	}
122
123	return $post_list;
124	}
125
126	/**
127	* Resolves the post's primary category into a Category value object.
128	*
129	* Looks up Yoast's explicit primary term first, then the `_yoast_wpseo_primary_category` post meta,
130	* and finally falls back to the first category WordPress has assigned to the post.
131	*
132	* @param int $post_id The post id.
133	*
134	* @return Category\|null The primary category, or null when the post has no category at all.
135	*/
136	private function get_primary_category( int $post_id ): ?Category {
137	// We try to find the primary term using the repository first, which uses indexables.
138	$primary_term = $this->primary_term_repository->find_by_post_id_and_taxonomy( $post_id, 'category', false );
139
140	if ( $primary_term ) {
141	$term_id = $primary_term->term_id;
142	}
143	// If the repository did not return a primary term, we try to find it using the post meta, which is how older versions of Yoast SEO stored the primary category.
144	else {
145	$term_id = \get_post_meta( $post_id, WPSEO_Meta::$meta_prefix . 'primary_category', true );
146	}
147
148	// No primary term has been set in Yoast SEO, so we fall back to the first category WordPress has assigned to the post, if any.
149	if ( empty( $term_id ) ) {
150	$category_ids = \wp_get_post_categories( $post_id );
151	$term_id = ( $category_ids[0] ?? 0 );
152	}
153
154	// If there are no categories at all, we return null.
155	if ( empty( $term_id ) ) {
156	return null;
157	}
158
159	$term = \get_term( (int) $term_id, 'category' );
160	if ( ! $term instanceof WP_Term ) {
161	return null;
162	}
163
164	return new Category( $term->name, $term->term_id );
165	}
166	}
167