wordpress-seo / src / llms-txt / infrastructure / content / automatic-post-collection.php

automatic-post-collection.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 28.2, at src/llms-txt/infrastructure/content/automatic-post-collection.php

216 lines 7.0 KB

Raw Download Zip

1	<?php
2
3	// phpcs:disable Yoast.NamingConventions.NamespaceName.TooLong -- Needed in the folder structure.
4	namespace Yoast\WP\SEO\Llms_Txt\Infrastructure\Content;
5
6	use WP_Post;
7	use Yoast\WP\SEO\Helpers\Indexable_Helper;
8	use Yoast\WP\SEO\Helpers\Options_Helper;
9	use Yoast\WP\SEO\Llms_Txt\Domain\Content\Post_Collection_Interface;
10	use Yoast\WP\SEO\Llms_Txt\Domain\Content_Types\Content_Type_Entry;
11	use Yoast\WP\SEO\Repositories\Indexable_Repository;
12	use Yoast\WP\SEO\Surfaces\Meta_Surface;
13
14	/**
15	* The class that handles the automatic post collection. Based on either indexables or WP_Query.
16	*
17	* @makePublic
18	*/
19	class Automatic_Post_Collection implements Post_Collection_Interface {
20
21	/**
22	* The options helper.
23	*
24	* @var Options_Helper
25	*/
26	private $options_helper;
27
28	/**
29	* The indexable repository.
30	*
31	* @var Indexable_Repository
32	*/
33	private $indexable_repository;
34
35	/**
36	* The meta surface.
37	*
38	* @var Meta_Surface
39	*/
40	private $meta;
41
42	/**
43	* The indexable helper.
44	*
45	* @var Indexable_Helper
46	*/
47	private $indexable_helper;
48
49	/**
50	* Constructs the class.
51	*
52	* @param Options_Helper $options_helper The options helper.
53	* @param Indexable_Repository $indexable_repository The indexable repository.
54	* @param Meta_Surface $meta The meta surface.
55	* @param Indexable_Helper $indexable_helper The indexable helper.
56	*/
57	public function __construct(
58	Options_Helper $options_helper,
59	Indexable_Repository $indexable_repository,
60	Meta_Surface $meta,
61	Indexable_Helper $indexable_helper
62	) {
63	$this->options_helper = $options_helper;
64	$this->indexable_repository = $indexable_repository;
65	$this->meta = $meta;
66	$this->indexable_helper = $indexable_helper;
67	}
68
69	/**
70	* Gets the posts that are relevant for the LLMs.txt.
71	*
72	* @param string $post_type The post type.
73	* @param int $limit The maximum number of posts to return.
74	*
75	* @return array<int, array<Content_Type_Entry>> The posts that are relevant for the LLMs.txt.
76	*/
77	public function get_posts( string $post_type, int $limit ): array {
78	$posts = $this->get_recent_cornerstone_content( $post_type, $limit );
79
80	if ( \count( $posts ) >= $limit ) {
81	return $posts;
82	}
83
84	$recent_posts = $this->get_recent_posts( $post_type, $limit );
85	foreach ( $recent_posts as $recent_post ) {
86	// If the post is already in the list because it's cornerstone, don't add it again.
87	if ( isset( $posts[ $recent_post->get_id() ] ) ) {
88	continue;
89	}
90
91	$posts[ $recent_post->get_id() ] = $recent_post;
92
93	if ( \count( $posts ) >= $limit ) {
94	break;
95	}
96	}
97
98	return $posts;
99	}
100
101	/**
102	* Gets the most recently modified cornerstone content.
103	*
104	* @param string $post_type The post type.
105	* @param int $limit The maximum number of posts to return.
106	*
107	* @return array<int, array<Content_Type_Entry>> The most recently modified cornerstone content.
108	*/
109	private function get_recent_cornerstone_content( string $post_type, int $limit ): array {
110	if ( ! $this->options_helper->get( 'enable_cornerstone_content' ) ) {
111	return [];
112	}
113
114	$cornerstone_limit = ( \is_post_type_hierarchical( $post_type ) ) ? null : $limit;
115	$cornerstones = $this->indexable_repository->get_recent_cornerstone_for_post_type( $post_type, $cornerstone_limit );
116
117	$recent_cornerstone_posts = [];
118	foreach ( $cornerstones as $cornerstone ) {
119	$cornerstone_meta = $this->meta->for_indexable( $cornerstone );
120	if ( $cornerstone_meta->post instanceof WP_Post ) {
121	$recent_cornerstone_posts[ $cornerstone_meta->post->ID ] = Content_Type_Entry::from_meta( $cornerstone_meta );
122	}
123	}
124
125	return $recent_cornerstone_posts;
126	}
127
128	/**
129	* Gets the most recently modified posts.
130	*
131	* @param string $post_type The post type.
132	* @param int $limit The maximum number of posts to return.
133	* @param string $search_filter Optional. The search filter to apply to the query.
134	* @param bool $disable_excluding_old_posts Optional. Whether to disable excluding posts older than one year.
135	*
136	* @return array<Content_Type_Entry> The most recently modified posts.
137	*/
138	public function get_recent_posts( string $post_type, int $limit, string $search_filter = '', bool $disable_excluding_old_posts = false ): array {
139	$exclude_older_than_one_year = false;
140
141	if ( $post_type === 'post' && ! $disable_excluding_old_posts ) {
142	$exclude_older_than_one_year = true;
143	}
144
145	if ( $this->indexable_helper->should_index_indexables() ) {
146	return $this->get_recently_modified_posts_indexables( $post_type, $limit, $exclude_older_than_one_year, $search_filter );
147	}
148
149	return $this->get_recently_modified_posts_wp_query( $post_type, $limit, $exclude_older_than_one_year, $search_filter );
150	}
151
152	/**
153	* Returns most recently modified posts of a post type, using indexables.
154	*
155	* @param string $post_type The post type.
156	* @param int $limit The maximum number of posts to return.
157	* @param bool $exclude_older_than_one_year Whether to exclude posts older than one year.
158	* @param string $search_filter Optional. The search filter to apply to the query.
159	*
160	* @return array<Content_Type_Entry> The most recently modified posts.
161	*/
162	private function get_recently_modified_posts_indexables( string $post_type, int $limit, bool $exclude_older_than_one_year, string $search_filter = '' ): array {
163	$posts = [];
164	$recently_modified_indexables = $this->indexable_repository->get_recently_modified_posts( $post_type, $limit, $exclude_older_than_one_year, $search_filter );
165
166	foreach ( $recently_modified_indexables as $indexable ) {
167	$indexable_meta = $this->meta->for_indexable( $indexable );
168	if ( $indexable_meta->post instanceof WP_Post ) {
169	$posts[] = Content_Type_Entry::from_meta( $indexable_meta );
170	}
171	}
172
173	return $posts;
174	}
175
176	/**
177	* Returns most recently modified posts of a post type, using WP_Query.
178	*
179	* @param string $post_type The post type.
180	* @param int $limit The maximum number of posts to return.
181	* @param bool $exclude_older_than_one_year Whether to exclude posts older than one year.
182	* @param string $search_filter Optional. The search filter to apply to the query.
183	*
184	* @return array<WP_Post> The most recently modified posts.
185	*/
186	private function get_recently_modified_posts_wp_query( string $post_type, int $limit, bool $exclude_older_than_one_year, string $search_filter = '' ): array {
187	$args = [
188	'post_type' => $post_type,
189	'posts_per_page' => $limit,
190	'post_status' => 'publish',
191	'orderby' => 'modified',
192	'order' => 'DESC',
193	'has_password' => false,
194	];
195
196	if ( $exclude_older_than_one_year === true ) {
197	$args['date_query'] = [
198	[
199	'after' => '12 months ago',
200	],
201	];
202	}
203
204	if ( $search_filter !== '' ) {
205	$args['s'] = $search_filter;
206	}
207
208	$posts = [];
209	foreach ( \get_posts( $args ) as $post ) {
210	$posts[] = Content_Type_Entry::from_post( $post, \get_permalink( $post->ID ) );
211	}
212
213	return $posts;
214	}
215	}
216