wordpress-seo / src / schema-aggregator / user-interface / schemamap-xml-rewrite-integration.php

schemamap-xml-rewrite-integration.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 28.4, at src/schema-aggregator/user-interface/schemamap-xml-rewrite-integration.php

196 lines 6.6 KB

Raw Download Zip

1	<?php
2	// phpcs:disable Yoast.NamingConventions.NamespaceName.TooLong -- Needed in the folder structure.
3	namespace Yoast\WP\SEO\Schema_Aggregator\User_Interface;
4
5	use WP_Query;
6	use Yoast\WP\SEO\Conditionals\Front_End_Conditional;
7	use Yoast\WP\SEO\Helpers\Redirect_Helper;
8	use Yoast\WP\SEO\Integrations\Integration_Interface;
9	use Yoast\WP\SEO\Schema_Aggregator\Application\Schema_Map\Schema_Map_Xml_Provider;
10	use Yoast\WP\SEO\Schema_Aggregator\Infrastructure\Schema_Aggregator_Conditional;
11	use Yoast_Dynamic_Rewrites;
12
13	/**
14	* Serves the schema map at the `schemamap.xml` path of the site.
15	*
16	* @phpcs:disable Yoast.NamingConventions.ObjectNameDepth.MaxExceeded
17	*/
18	class Schemamap_Xml_Rewrite_Integration implements Integration_Interface {
19
20	/**
21	* The query variable that flags a request for the schema map.
22	*
23	* @var string
24	*/
25	public const QUERY_VAR = 'yoast_schemamap';
26
27	/**
28	* The schema map XML provider.
29	*
30	* @var Schema_Map_Xml_Provider
31	*/
32	private $schema_map_xml_provider;
33
34	/**
35	* The redirect helper.
36	*
37	* @var Redirect_Helper
38	*/
39	private $redirect_helper;
40
41	/**
42	* Returns the conditionals for this integration.
43	*
44	* The front end conditional keeps the rewrite rule out of admin requests, where the schema map is
45	* never served. It also means the rule is absent while Yoast SEO is being deactivated, so no
46	* equivalent of the `Deactivating_Yoast_Seo_Conditional` guard in `WPSEO_Sitemaps_Router` is needed.
47	*
48	* @return array<string> The conditionals that must be met to load this.
49	*/
50	public static function get_conditionals() {
51	return [ Schema_Aggregator_Conditional::class, Front_End_Conditional::class ];
52	}
53
54	/**
55	* Schemamap_Xml_Rewrite_Integration constructor.
56	*
57	* @param Schema_Map_Xml_Provider $schema_map_xml_provider The schema map XML provider.
58	* @param Redirect_Helper $redirect_helper The redirect helper.
59	*/
60	public function __construct( Schema_Map_Xml_Provider $schema_map_xml_provider, Redirect_Helper $redirect_helper ) {
61	$this->schema_map_xml_provider = $schema_map_xml_provider;
62	$this->redirect_helper = $redirect_helper;
63	}
64
65	/**
66	* Registers the hooks for the integration.
67	*
68	* The schema map is rendered on `pre_get_posts` rather than `template_redirect`, the same way
69	* `WPSEO_Sitemaps::redirect()` renders sitemaps. That skips the main posts query, which this
70	* response never uses, and leaves fewer third-party callbacks able to echo ahead of the XML —
71	* anything printed before the document would stop it parsing as a sitemap.
72	*
73	* @return void
74	*/
75	public function register_hooks() {
76	/*
77	* Integrations register their hooks on `init` priority 10, which is after
78	* `yoast_add_dynamic_rewrite_rules` has already fired on `init` priority 1. The rule is
79	* therefore added directly, the same way `WPSEO_Sitemaps::register_sitemap()` does for
80	* sitemaps registered during `init`. Rewrite rules are not read until `WP::parse_request()`,
81	* so this is still in time.
82	*/
83	$this->add_rewrite_rules( Yoast_Dynamic_Rewrites::instance() );
84
85	\add_filter( 'query_vars', [ $this, 'add_query_vars' ] );
86	\add_action( 'pre_get_posts', [ $this, 'maybe_render_schema_map' ], 1 );
87	}
88
89	/**
90	* Adds the rewrite rule for the schema map.
91	*
92	* @param Yoast_Dynamic_Rewrites $dynamic_rewrites Dynamic rewrites handler instance.
93	*
94	* @return void
95	*/
96	public function add_rewrite_rules( Yoast_Dynamic_Rewrites $dynamic_rewrites ) {
97	$dynamic_rewrites->add_rule( 'schemamap\.xml$', 'index.php?' . self::QUERY_VAR . '=1', 'top' );
98	}
99
100	/**
101	* Adds the query variable for the schema map.
102	*
103	* @param array<string> $query_vars List of query variables to filter.
104	*
105	* @return array<string> Filtered query variables.
106	*/
107	public function add_query_vars( $query_vars ) {
108	$query_vars[] = self::QUERY_VAR;
109
110	return $query_vars;
111	}
112
113	/**
114	* Outputs the schema map when the main query is a request for it.
115	*
116	* @param WP_Query $query The query that is about to run.
117	*
118	* @return void
119	*/
120	public function maybe_render_schema_map( WP_Query $query ) {
121	if ( ! $query->is_main_query() \|\| ! $query->get( self::QUERY_VAR ) ) {
122	return;
123	}
124
125	/*
126	* Without control over the headers the response cannot be typed as XML, and echoing the
127	* document into a half-written HTML body would produce neither a page nor a schema map.
128	*/
129	if ( $this->headers_already_sent() ) {
130	return;
131	}
132
133	/*
134	* Build the document before committing any headers. Generating it hits the database and can
135	* throw, and a failure after the 200 and the cache headers were sent would let a proxy store
136	* an HTML error page at this path for the full max-age.
137	*/
138	$xml = $this->schema_map_xml_provider->get_xml();
139
140	$this->send_headers();
141
142	// DOMDocument::createTextNode() already escaped every value; escaping again would corrupt the XML.
143	echo $xml; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- Raw XML.
144
145	$this->finish_request();
146	}
147
148	/**
149	* Sends the response headers for the schema map.
150	*
151	* These replace what `WP::send_headers()` already sent: the HTML content type, and the `Expires`
152	* and `Cache-Control` pair that `wp_get_nocache_headers()` adds for logged-in users.
153	*
154	* @return void
155	*/
156	protected function send_headers() {
157	\status_header( 200 );
158	// Prevent the search engines from indexing the schema map.
159	$this->redirect_helper->set_header( 'X-Robots-Tag: noindex, follow' );
160	$this->redirect_helper->set_header( 'Content-Type: application/xml; charset=UTF-8' );
161	$this->redirect_helper->set_header( 'Cache-Control: public, max-age=300' );
162	// The response is public, so the no-cache Expires date from wp_get_nocache_headers() must go.
163	$this->redirect_helper->remove_header( 'Expires' );
164	}
165
166	/**
167	* Tells whether the response headers have already been sent.
168	*
169	* Wrapped in an overridable method so the unit tests can drive both branches of
170	* maybe_render_schema_map(); the test process cannot control what headers_sent() reports.
171	* Protected rather than private because Mockery cannot stub private methods.
172	*
173	* @codeCoverageIgnore It only wraps a PHP function.
174	*
175	* @return bool Whether the headers have already been sent.
176	*/
177	protected function headers_already_sent(): bool {
178	return \headers_sent();
179	}
180
181	/**
182	* Ends the request after the schema map has been output.
183	*
184	* The exit() is wrapped in an overridable method so the unit tests can stub it out; executing it
185	* would terminate the test runner. Protected rather than private because Mockery cannot stub
186	* private methods.
187	*
188	* @codeCoverageIgnore It only wraps a language construct that terminates the request.
189	*
190	* @return void
191	*/
192	protected function finish_request() {
193	exit();
194	}
195	}
196