thinkrank / includes / seo / class-url-scheme.php

class-url-scheme.php in ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO 2.7.0, at includes/seo/class-url-scheme.php

239 lines 6.4 KB

Raw Download Zip

1	<?php
2	/**
3	* The scheme ThinkRank publishes its own URLs with.
4	*
5	* @package ThinkRank
6	* @since 2.7.0
7	*/
8
9	declare(strict_types=1);
10
11	namespace ThinkRank\SEO;
12
13	if (!defined('ABSPATH')) {
14	exit;
15	}
16
17	/**
18	* Decides whether a self-referential URL goes out as http or https.
19	*
20	* Sites behind a proxy or load balancer that terminates TLS hand PHP a plain
21	* HTTP request, so `is_ssl()` is false and WordPress falls back to whatever
22	* scheme `home` was stored with. On a site installed over http and later put
23	* behind TLS without anyone updating the option, every URL WordPress builds
24	* says `http://` while every visitor is on `https://` — and the canonical tag
25	* then tells search engines the wrong address for every page on the site.
26	*
27	* The site owner frequently cannot fix that at the server level, which is why
28	* this is a setting rather than a support answer.
29	*
30	* Two rules keep it from doing damage of its own:
31	*
32	* - Only this site's own URLs are touched. A cross-domain canonical the
33	* author typed, a `sameAs` pointing at a social profile, an image on a CDN:
34	* all of them are somebody else's URLs, and forcing our scheme onto them
35	* would break them. Host has to match before anything is rewritten.
36	* - `automatic` changes nothing at all. Not "resolves to the current
37	* scheme" — nothing. A site that never opens the setting emits exactly what
38	* WordPress reports, which is what every site did before this existed.
39	*
40	* @since 2.7.0
41	*/
42	class Url_Scheme {
43
44	/**
45	* Follow whatever WordPress reports. The default.
46	*
47	* @since 2.7.0
48	* @var string
49	*/
50	public const AUTOMATIC = 'automatic';
51
52	/**
53	* Publish every self-URL as https.
54	*
55	* @since 2.7.0
56	* @var string
57	*/
58	public const HTTPS = 'https';
59
60	/**
61	* Publish every self-URL as http.
62	*
63	* @since 2.7.0
64	* @var string
65	*/
66	public const HTTP = 'http';
67
68	/**
69	* The accepted values, in the order the settings screen lists them.
70	*
71	* @since 2.7.0
72	* @var string[]
73	*/
74	public const MODES = [self::AUTOMATIC, self::HTTPS, self::HTTP];
75
76	/**
77	* Resolved preference for this request.
78	*
79	* @since 2.7.0
80	* @var string\|null
81	*/
82	private static $mode = null;
83
84	/**
85	* This site's host, lowercased.
86	*
87	* @since 2.7.0
88	* @var string\|null
89	*/
90	private static $host = null;
91
92	/**
93	* The scheme preference this site has stored.
94	*
95	* Read once per request. Every emit point calls this, including the
96	* sitemap generator inside a loop over thousands of URLs, so the settings
97	* read must not repeat.
98	*
99	* @since 2.7.0
100	* @return string One of MODES.
101	*/
102	public static function preference(): string {
103	if (null === self::$mode) {
104	$stored = '';
105
106	if (class_exists('\ThinkRank\SEO\Site_Identity_Manager')) {
107	$settings = (new Site_Identity_Manager())->get_settings('site', null);
108	$stored = (string) ($settings['canonical_scheme'] ?? '');
109	}
110
111	/**
112	* Filter the scheme ThinkRank publishes its own URLs with.
113	*
114	* @since 2.7.0
115	*
116	* @param string $mode One of Url_Scheme::MODES.
117	*/
118	$mode = (string) apply_filters('thinkrank_canonical_scheme', $stored);
119
120	self::$mode = in_array($mode, self::MODES, true) ? $mode : self::AUTOMATIC;
121	}
122
123	return self::$mode;
124	}
125
126	/**
127	* Apply the site's scheme preference to one URL.
128	*
129	* @since 2.7.0
130	* @param string $url Absolute URL, or anything else (returned untouched).
131	* @return string
132	*/
133	public static function apply(string $url): string {
134	$mode = self::preference();
135
136	if (self::AUTOMATIC === $mode \|\| '' === $url) {
137	return $url;
138	}
139
140	$scheme = wp_parse_url($url, PHP_URL_SCHEME);
141
142	// No scheme means relative or protocol-relative: not ours to decide.
143	if (!is_string($scheme) \|\| '' === $scheme) {
144	return $url;
145	}
146
147	$scheme = strtolower($scheme);
148
149	if ('http' !== $scheme && 'https' !== $scheme) {
150	return $url;
151	}
152
153	if ($scheme === $mode \|\| !self::is_own_url($url)) {
154	return $url;
155	}
156
157	return set_url_scheme($url, $mode);
158	}
159
160	/**
161	* Apply the preference to every string in a nested structure.
162	*
163	* The schema graph is the reason this exists: its `@id` and `url` values
164	* come from a dozen different producers, some deriving from WordPress and
165	* some from stored settings, and normalizing at the point of serialization
166	* is what stops one graph carrying both schemes at once.
167	*
168	* @since 2.7.0
169	* @param mixed $value Array, string, or anything else.
170	* @return mixed Same shape, schemes normalized.
171	*/
172	public static function apply_deep($value) {
173	if (self::AUTOMATIC === self::preference()) {
174	return $value;
175	}
176
177	if (is_string($value)) {
178	return self::apply($value);
179	}
180
181	if (!is_array($value)) {
182	return $value;
183	}
184
185	foreach ($value as $key => $item) {
186	$value[$key] = self::apply_deep($item);
187	}
188
189	return $value;
190	}
191
192	/**
193	* Whether a URL belongs to this site.
194	*
195	* Host comparison only. A subdomain is a different site as far as a
196	* canonical is concerned, and a port or path difference is still us.
197	*
198	* @since 2.7.0
199	* @param string $url Absolute URL.
200	* @return bool
201	*/
202	private static function is_own_url(string $url): bool {
203	$host = wp_parse_url($url, PHP_URL_HOST);
204
205	if (!is_string($host) \|\| '' === $host) {
206	return false;
207	}
208
209	return strtolower($host) === self::host();
210	}
211
212	/**
213	* This site's host, lowercased and memoised.
214	*
215	* @since 2.7.0
216	* @return string
217	*/
218	private static function host(): string {
219	if (null === self::$host) {
220	$host = wp_parse_url(home_url('/'), PHP_URL_HOST);
221
222	self::$host = is_string($host) ? strtolower($host) : '';
223	}
224
225	return self::$host;
226	}
227
228	/**
229	* Forget the memoised preference and host.
230	*
231	* @since 2.7.0
232	* @return void
233	*/
234	public static function reset(): void {
235	self::$mode = null;
236	self::$host = null;
237	}
238	}
239