thinkrank / includes / seo / class-indexability.php

class-indexability.php in ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO 2.14.1, at includes/seo/class-indexability.php

332 lines 10.2 KB

Raw Download Zip

1	<?php
2	/**
3	* Indexability
4	*
5	* @package ThinkRank
6	* @subpackage SEO
7	* @since 2.15.0
8	*/
9
10	declare(strict_types=1);
11
12	namespace ThinkRank\SEO;
13
14	// Prevent direct access.
15	if (!defined('ABSPATH')) {
16	exit;
17	}
18
19	/**
20	* Whether a post or term is a destination worth submitting to a search engine.
21	*
22	* The robots tag a page prints was the only place that decided indexability.
23	* Every URL feed re-decided it partially or not at all: the sitemap honoured a
24	* per-post noindex but not a noindexed post type or taxonomy, and listed URLs
25	* ThinkRank itself redirects; IndexNow checked nothing beyond the post type
26	* (#911). Each of those contradicts a signal the site sends elsewhere, which
27	* Search Console reports as "Submitted URL marked noindex" and "Page with
28	* redirect".
29	*
30	* So this class answers the question once, from the same inputs the robots tag
31	* reads, and every feed asks it. It is public so ThinkRank Pro's feeds (news
32	* and video sitemaps, llms-full.txt, internal-link targets) can use it too.
33	*
34	* Three things make an object a non-destination:
35	* - the effective robots decision is noindex, resolved through the cascade
36	* {@see \ThinkRank\Frontend\SEO_Manager} uses for the robots tag: site-wide
37	* directives, then the content type's row in the matrix, then the object's
38	* own override, each merged over the last;
39	* - the post has a password;
40	* - the object redirects elsewhere, either through its own "Redirect this to"
41	* field ({@see Object_Redirect}) or because a redirect provider answers the
42	* `thinkrank_url_redirects` filter for its URL.
43	*
44	* @since 2.15.0
45	*/
46	final class Indexability {
47
48	/**
49	* Site-wide robots directives (the base of the cascade).
50	*/
51	private const GLOBAL_ROBOTS_OPTION = 'thinkrank_global_robot_meta_settings';
52
53	/**
54	* Robots keys a per-object override may set, as the robots tag reads them.
55	*/
56	private const OVERRIDE_KEYS = ['index', 'noindex', 'nofollow', 'noarchive', 'noimageindex', 'nosnippet'];
57
58	/**
59	* Reasons an object is not indexable.
60	*/
61	public const REASON_NOINDEX = 'noindex';
62	public const REASON_PASSWORD = 'password';
63	public const REASON_REDIRECT = 'redirect';
64
65	/**
66	* Whether a post or term may be submitted to search engines.
67	*
68	* @since 2.15.0
69	*
70	* @param \WP_Post\|\WP_Term $item Post or term.
71	* @return bool
72	*/
73	public static function is_indexable_destination($item): bool {
74	if ($item instanceof \WP_Post) {
75	return self::is_indexable_post($item);
76	}
77
78	if ($item instanceof \WP_Term) {
79	return self::is_indexable_term($item);
80	}
81
82	return false;
83	}
84
85	/**
86	* Whether a post may be submitted to search engines.
87	*
88	* @since 2.15.0
89	*
90	* @param \WP_Post $post Post.
91	* @return bool
92	*/
93	public static function is_indexable_post(\WP_Post $post): bool {
94	return self::post_exclusion_reason($post) === null;
95	}
96
97	/**
98	* Whether a term archive may be submitted to search engines.
99	*
100	* @since 2.15.0
101	*
102	* @param \WP_Term $term Term.
103	* @return bool
104	*/
105	public static function is_indexable_term(\WP_Term $term): bool {
106	return self::term_exclusion_reason($term) === null;
107	}
108
109	/**
110	* Why a post is not indexable, or null when it is.
111	*
112	* Cheapest check first: the password is on the object, the robots cascade
113	* reads options and primed meta, and a redirect lookup may query the
114	* redirect provider's table.
115	*
116	* @since 2.15.0
117	*
118	* @param \WP_Post $post Post.
119	* @return string\|null One of the REASON_* constants, or null.
120	*/
121	public static function post_exclusion_reason(\WP_Post $post): ?string {
122	if (self::is_password_protected($post)) {
123	return self::REASON_PASSWORD;
124	}
125
126	if (self::is_post_noindexed($post)) {
127	return self::REASON_NOINDEX;
128	}
129
130	if (self::is_post_redirected($post)) {
131	return self::REASON_REDIRECT;
132	}
133
134	return null;
135	}
136
137	/**
138	* Why a term is not indexable, or null when it is.
139	*
140	* @since 2.15.0
141	*
142	* @param \WP_Term $term Term.
143	* @return string\|null One of the REASON_* constants, or null.
144	*/
145	public static function term_exclusion_reason(\WP_Term $term): ?string {
146	if (self::is_term_noindexed($term)) {
147	return self::REASON_NOINDEX;
148	}
149
150	if (self::is_term_redirected($term)) {
151	return self::REASON_REDIRECT;
152	}
153
154	return null;
155	}
156
157	/**
158	* Whether the post has a password.
159	*
160	* @since 2.15.0
161	*
162	* @param \WP_Post $post Post.
163	* @return bool
164	*/
165	public static function is_password_protected(\WP_Post $post): bool {
166	return (string) $post->post_password !== '';
167	}
168
169	/**
170	* Whether the robots tag on this post's page says noindex.
171	*
172	* Site-wide directives, then the post type's own (when its robots switch
173	* is on), then the post's override. A post-level `noindex: false` beats a
174	* noindexed type, exactly as it does in the tag.
175	*
176	* @since 2.15.0
177	*
178	* @param \WP_Post $post Post.
179	* @return bool
180	*/
181	public static function is_post_noindexed(\WP_Post $post): bool {
182	$settings = self::merge_entity_robots(self::global_robots(), (string) $post->post_type);
183
184	$settings = self::merge_override(
185	$settings,
186	(bool) get_post_meta($post->ID, '_thinkrank_robots_meta_enabled', true),
187	get_post_meta($post->ID, '_thinkrank_robots_meta', true)
188	);
189
190	return !empty($settings['noindex']);
191	}
192
193	/**
194	* Whether the robots tag on this term's archive says noindex.
195	*
196	* Site-wide directives, then the taxonomy's matrix row, then the term's
197	* override.
198	*
199	* @since 2.15.0
200	*
201	* @param \WP_Term $term Term.
202	* @return bool
203	*/
204	public static function is_term_noindexed(\WP_Term $term): bool {
205	$settings = self::merge_entity_robots(
206	self::global_robots(),
207	Content_Type_Settings::PREFIX_TAXONOMY . $term->taxonomy
208	);
209
210	$settings = self::merge_override(
211	$settings,
212	(bool) get_term_meta($term->term_id, '_thinkrank_robots_meta_enabled', true),
213	get_term_meta($term->term_id, '_thinkrank_robots_meta', true)
214	);
215
216	return !empty($settings['noindex']);
217	}
218
219	/**
220	* Whether a request for this post is redirected elsewhere.
221	*
222	* @since 2.15.0
223	*
224	* @param \WP_Post $post Post.
225	* @return bool
226	*/
227	public static function is_post_redirected(\WP_Post $post): bool {
228	if (Object_Redirect::get('post', (int) $post->ID)['url'] !== '') {
229	return true;
230	}
231
232	$url = get_permalink($post);
233
234	return is_string($url) && $url !== '' && self::url_is_redirected($url, $post);
235	}
236
237	/**
238	* Whether a request for this term's archive is redirected elsewhere.
239	*
240	* @since 2.15.0
241	*
242	* @param \WP_Term $term Term.
243	* @return bool
244	*/
245	public static function is_term_redirected(\WP_Term $term): bool {
246	if (Object_Redirect::get('term', (int) $term->term_id)['url'] !== '') {
247	return true;
248	}
249
250	$url = get_term_link($term);
251
252	return is_string($url) && $url !== '' && self::url_is_redirected($url, $term);
253	}
254
255	/**
256	* Ask the redirect provider whether a URL is redirected.
257	*
258	* @param string $url The object's URL.
259	* @param \WP_Post\|\WP_Term $item The object it belongs to.
260	* @return bool
261	*/
262	private static function url_is_redirected(string $url, $item): bool {
263	/**
264	* Filter whether a URL is redirected by a rule that is not tied to an object.
265	*
266	* Per-object redirects are already read through `thinkrank_object_redirect`.
267	* This covers the rest: a rule whose source path happens to be a post's
268	* or term's URL. Nothing answers it in the free plugin; a redirect
269	* provider such as ThinkRank Pro does, from its own rule store.
270	*
271	* @since 2.15.0
272	*
273	* @param bool $redirected Whether the URL redirects. Default false.
274	* @param string $url The post's permalink or the term's archive link.
275	* @param \WP_Post\|\WP_Term $item The object the URL belongs to.
276	*/
277	return (bool) apply_filters('thinkrank_url_redirects', false, $url, $item);
278	}
279
280	/**
281	* Site-wide robots directives with the same defaults the robots tag uses.
282	*
283	* @return array
284	*/
285	private static function global_robots(): array {
286	$stored = get_option(self::GLOBAL_ROBOTS_OPTION, []);
287
288	return array_merge(
289	['index' => true, 'noindex' => false],
290	is_array($stored) ? $stored : []
291	);
292	}
293
294	/**
295	* Merge a matrix row's robots directives when its robots switch is on.
296	*
297	* @param array $settings Directives so far.
298	* @param string $entity_key Content_Type_Settings entity key.
299	* @return array
300	*/
301	private static function merge_entity_robots(array $settings, string $entity_key): array {
302	$entity = Content_Type_Settings::get_entity_settings($entity_key);
303
304	if (!empty($entity['robots_meta_enabled']) && is_array($entity['robots_meta'] ?? null)) {
305	$settings = array_merge($settings, $entity['robots_meta']);
306	}
307
308	return $settings;
309	}
310
311	/**
312	* Merge an object's stored robots override when it is switched on.
313	*
314	* @param array $settings Directives so far.
315	* @param bool $enabled The object's `_thinkrank_robots_meta_enabled`.
316	* @param mixed $raw The object's `_thinkrank_robots_meta` JSON.
317	* @return array
318	*/
319	private static function merge_override(array $settings, bool $enabled, $raw): array {
320	if (!$enabled \|\| !is_string($raw) \|\| $raw === '') {
321	return $settings;
322	}
323
324	$override = json_decode($raw, true);
325	if (!is_array($override)) {
326	return $settings;
327	}
328
329	return array_merge($settings, array_intersect_key($override, array_flip(self::OVERRIDE_KEYS)));
330	}
331	}
332