thinkrank / includes / abilities / content / class-get-post-content.php

class-get-post-content.php in ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO 2.7.0, at includes/abilities/content/class-get-post-content.php

231 lines 7.5 KB

Raw Download Zip

1	<?php
2	/**
3	* Get post content ability.
4	*
5	* @package ThinkRank\Abilities\Content
6	*/
7
8	declare(strict_types=1);
9
10	namespace ThinkRank\Abilities\Content;
11
12	use ThinkRank\Abilities\Ability_Base;
13	use ThinkRank\SEO\Builder_Content;
14
15	if ( ! defined( 'ABSPATH' ) ) {
16	exit; // Exit if accessed directly.
17	}
18
19	/**
20	* Reads a post's body content.
21	*
22	* update-post-seo declares `additionalProperties: false` over an SEO-only field
23	* list, and nothing else read the body at all — so an agent was optimising a
24	* title and description for content it could not see (#676).
25	*
26	* Read-only, deliberately. There is no counterpart that writes the body, and
27	* that is the point rather than an omission: page-builder content does not live
28	* in `post_content` — Bricks, Oxygen, Breakdance and Elementor keep their trees
29	* in postmeta — so a write ability that set `post_content` would appear to
30	* succeed and change nothing on a builder page, or worse, resurrect stale
31	* blocks the builder had superseded. That is the failure mode #335 already
32	* documents. A write surface needs to be builder-aware before it is safe, which
33	* is a larger piece of work than this.
34	*
35	* `content` is therefore what the page actually renders, resolved through the
36	* same pipeline the SEO score and meta description use, and `post_content` is
37	* the raw stored column beside it so the difference is visible rather than
38	* guessed at.
39	*/
40	class Get_Post_Content extends Ability_Base {
41
42	/**
43	* Constructor.
44	*/
45	public function __construct() {
46	$this->id = 'thinkrank/get-post-content';
47	$this->label = __( 'Get ThinkRank Post Content', 'thinkrank' );
48	$this->description = __( 'Read a post or page\'s body content so you can judge what it is about before changing its SEO fields. Returns the text the page actually renders — resolved from the page builder when one owns the page — plus the raw stored post_content and which builder is in use. Read-only: use update-post-seo to change SEO fields. Use list-content-items to find ids.', 'thinkrank' );
49	}
50
51	/**
52	* {@inheritDoc}
53	*
54	* @return array<string, bool\|float\|string>
55	*/
56	public function get_annotations() {
57	return [
58	'readonly' => true,
59	'destructive' => false,
60	'idempotent' => true,
61	'priority' => 1.0,
62	'openWorldHint' => false,
63	];
64	}
65
66	/**
67	* {@inheritDoc}
68	*
69	* @return array<string, mixed>
70	*/
71	public function get_input_schema() {
72	return [
73	'type' => 'object',
74	'additionalProperties' => false,
75	'required' => [ 'post_id' ],
76	'properties' => [
77	'post_id' => [
78	'type' => 'integer',
79	'description' => __( 'Post or page ID from list-content-items.', 'thinkrank' ),
80	],
81	'max_chars' => [
82	'type' => 'integer',
83	'description' => __( 'Truncate the returned body to this many characters. Omit for the whole thing.', 'thinkrank' ),
84	],
85	],
86	];
87	}
88
89	/**
90	* {@inheritDoc}
91	*
92	* @return array<string, mixed>
93	*/
94	public function get_output_schema() {
95	return [
96	'type' => 'object',
97	'properties' => [
98	'id' => [ 'type' => 'integer' ],
99	'title' => [ 'type' => 'string' ],
100	'post_type' => [ 'type' => 'string' ],
101	'status' => [ 'type' => 'string' ],
102	'url' => [ 'type' => 'string' ],
103	'edit_url' => [ 'type' => 'string' ],
104	'excerpt' => [ 'type' => 'string' ],
105	'content' => [
106	'type' => 'string',
107	'description' => __( 'Plain-text body as the page renders it, resolved from the page builder where one owns the page.', 'thinkrank' ),
108	],
109	'post_content' => [
110	'type' => 'string',
111	'description' => __( 'The raw stored post_content column. Empty or stale on builder pages, which is why it is reported separately.', 'thinkrank' ),
112	],
113	'content_source' => [
114	'type' => 'string',
115	'enum' => [ 'post_content', 'builder' ],
116	'description' => __( 'Where the rendered body came from.', 'thinkrank' ),
117	],
118	'builder' => [
119	'type' => [ 'string', 'null' ],
120	'description' => __( 'The page builder that owns this post, or null for the block/classic editor.', 'thinkrank' ),
121	],
122	'word_count' => [ 'type' => 'integer' ],
123	'truncated' => [ 'type' => 'boolean' ],
124	],
125	];
126	}
127
128	/**
129	* Execute ability.
130	*
131	* @param array<string, mixed> $input Ability input payload.
132	* @return array<string, mixed>\|\WP_Error
133	*/
134	public function execute( $input ) {
135	$input = (array) $input;
136	$post_id = (int) ( $input['post_id'] ?? 0 );
137	$post = $post_id > 0 ? get_post( $post_id ) : null;
138
139	if ( ! $post instanceof \WP_Post ) {
140	return new \WP_Error(
141	'thinkrank_post_not_found',
142	sprintf(
143	/* translators: %d: post ID that was sent. */
144	__( 'No post with id %d. Use list-content-items to find valid ids.', 'thinkrank' ),
145	$post_id
146	),
147	[ 'status' => 404 ]
148	);
149	}
150
151	// Reading a password-protected body would hand its contents to anyone
152	// the connector answers, so it is withheld the way the front end
153	// withholds it.
154	if ( '' !== (string) $post->post_password ) {
155	return new \WP_Error(
156	'thinkrank_post_password_protected',
157	__( 'This post is password protected, so its body is not returned.', 'thinkrank' ),
158	[ 'status' => 403 ]
159	);
160	}
161
162	$resolved = trim( wp_strip_all_tags( Builder_Content::resolve( $post ) ) );
163	$builder = $this->builder_for( $post_id );
164
165	$body = $resolved;
166	$truncated = false;
167
168	if ( isset( $input['max_chars'] ) ) {
169	$limit = max( 1, (int) $input['max_chars'] );
170
171	if ( mb_strlen( $body ) > $limit ) {
172	$body = mb_substr( $body, 0, $limit );
173	$truncated = true;
174	}
175	}
176
177	return [
178	'id' => $post_id,
179	'title' => (string) get_the_title( $post ),
180	'post_type' => (string) $post->post_type,
181	'status' => (string) $post->post_status,
182	'url' => (string) get_permalink( $post ),
183	'edit_url' => (string) get_edit_post_link( $post_id, 'raw' ),
184	'excerpt' => (string) $post->post_excerpt,
185	'content' => $body,
186	'post_content' => (string) $post->post_content,
187	'content_source' => null === $builder ? 'post_content' : 'builder',
188	'builder' => $builder,
189	'word_count' => '' === $resolved ? 0 : count( preg_split( '/\s+/u', $resolved ) ?: [] ),
190	'truncated' => $truncated,
191	];
192	}
193
194	/**
195	* Which page builder owns this post, if any.
196	*
197	* Derived from the meta keys Builder_Content already resolves through, so
198	* the answer cannot drift from the resolution above.
199	*
200	* @param int $post_id Post ID.
201	* @return string\|null Builder label, or null for the block/classic editor.
202	*/
203	private function builder_for( int $post_id ): ?string {
204	if ( Builder_Content::bricks_supersedes_post_content( $post_id ) ) {
205	return 'bricks';
206	}
207
208	// Same keys Builder_Content resolves through, in the same order. Kept
209	// as a local map because this one also needs a display label per key,
210	// and Oxygen spans three of them across its generations — Oxygen 6 is
211	// Breakdance under the hood.
212	$labels = [
213	'_breakdance_data' => 'breakdance',
214	'_oxygen_data' => 'oxygen',
215	'ct_builder_shortcodes' => 'oxygen',
216	'_elementor_data' => 'elementor',
217	'_fl_builder_data' => 'beaver-builder',
218	];
219
220	foreach ( $labels as $key => $label ) {
221	$stored = get_post_meta( $post_id, $key, true );
222
223	if ( ( is_string( $stored ) && '' !== trim( $stored ) ) \|\| ( is_array( $stored ) && ! empty( $stored ) ) ) {
224	return $label;
225	}
226	}
227
228	return null;
229	}
230	}
231