wordpress-seo / src / helpers / crawl-cleanup-helper.php

crawl-cleanup-helper.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 27.5, at src/helpers/crawl-cleanup-helper.php

315 lines 8.1 KB

Raw Download Zip

1	<?php
2
3	namespace Yoast\WP\SEO\Helpers;
4
5	/**
6	* Class Crawl_Cleanup_Helper.
7	*
8	* Used by the Crawl_Cleanup_Permalinks class.
9	*/
10	class Crawl_Cleanup_Helper {
11
12	/**
13	* The current page helper
14	*
15	* @var Current_Page_Helper
16	*/
17	private $current_page_helper;
18
19	/**
20	* The options helper.
21	*
22	* @var Options_Helper
23	*/
24	private $options_helper;
25
26	/**
27	* The URL helper.
28	*
29	* @var Url_Helper
30	*/
31	private $url_helper;
32
33	/**
34	* The Redirect Helper.
35	*
36	* @var Redirect_Helper
37	*/
38	private $redirect_helper;
39
40	/**
41	* Crawl Cleanup Basic integration constructor.
42	*
43	* @param Current_Page_Helper $current_page_helper The current page helper.
44	* @param Options_Helper $options_helper The option helper.
45	* @param Url_Helper $url_helper The URL helper.
46	* @param Redirect_Helper $redirect_helper The Redirect Helper.
47	*/
48	public function __construct(
49	Current_Page_Helper $current_page_helper,
50	Options_Helper $options_helper,
51	Url_Helper $url_helper,
52	Redirect_Helper $redirect_helper
53	) {
54	$this->current_page_helper = $current_page_helper;
55	$this->options_helper = $options_helper;
56	$this->url_helper = $url_helper;
57	$this->redirect_helper = $redirect_helper;
58	}
59
60	/**
61	* Checks if the current URL is not robots, sitemap, empty or user is logged in.
62	*
63	* @return bool True if the current URL is a valid URL.
64	*/
65	public function should_avoid_redirect() {
66	// phpcs:ignore WordPress.Security.NonceVerification.Recommended -- We're not processing anything yet...
67	if ( \is_robots() \|\| \get_query_var( 'sitemap' ) \|\| empty( $_GET ) \|\| \is_user_logged_in() ) {
68	return true;
69	}
70	return false;
71	}
72
73	/**
74	* Returns the list of the allowed extra vars.
75	*
76	* @return array The list of the allowed extra vars.
77	*/
78	public function get_allowed_extravars() {
79	$default_allowed_extravars = [
80	'utm_source',
81	'utm_medium',
82	'utm_campaign',
83	'utm_term',
84	'utm_content',
85	'gclid',
86	'gtm_debug',
87	];
88
89	/**
90	* Filter: 'Yoast\WP\SEO\allowlist_permalink_vars' - Allows plugins to register their own variables not to clean.
91	*
92	* @since 19.2.0
93	*
94	* @param array $allowed_extravars The list of the allowed vars (empty by default).
95	*/
96	$allowed_extravars = \apply_filters( 'Yoast\WP\SEO\allowlist_permalink_vars', $default_allowed_extravars );
97
98	$clean_permalinks_extra_variables = $this->options_helper->get( 'clean_permalinks_extra_variables' );
99
100	if ( $clean_permalinks_extra_variables !== '' ) {
101	$allowed_extravars = \array_merge( $allowed_extravars, \explode( ',', $clean_permalinks_extra_variables ) );
102	}
103	return $allowed_extravars;
104	}
105
106	/**
107	* Gets the allowed query vars from the current URL.
108	*
109	* @param string $current_url The current URL.
110	* @return array is_allowed and allowed_query.
111	*/
112	public function allowed_params( $current_url ) {
113	// This is a Premium plugin-only function: Allows plugins to register their own variables not to clean.
114	$allowed_extravars = $this->get_allowed_extravars();
115
116	$allowed_query = [];
117
118	$parsed_url = \wp_parse_url( $current_url, \PHP_URL_QUERY );
119
120	$query = $this->url_helper->parse_str_params( $parsed_url );
121
122	if ( ! empty( $allowed_extravars ) ) {
123	foreach ( $allowed_extravars as $get ) {
124	$get = \trim( $get );
125	if ( isset( $query[ $get ] ) ) {
126	$allowed_query[ $get ] = \rawurlencode_deep( $query[ $get ] );
127	unset( $query[ $get ] );
128	}
129	}
130	}
131	return [
132	'query' => $query,
133	'allowed_query' => $allowed_query,
134	];
135	}
136
137	/**
138	* Returns the proper URL for singular pages.
139	*
140	* @return string The proper URL.
141	*/
142	public function singular_url() {
143
144	global $post;
145	$proper_url = \get_permalink( $post->ID );
146	$page = \get_query_var( 'page' );
147
148	if ( $page && $page !== 1 ) {
149	$the_post = \get_post( $post->ID );
150	$page_count = \substr_count( $the_post->post_content, '<!--nextpage-->' );
151	$proper_url = \user_trailingslashit( \trailingslashit( $proper_url ) . $page );
152	if ( $page > ( $page_count + 1 ) ) {
153	$proper_url = \user_trailingslashit( \trailingslashit( $proper_url ) . ( $page_count + 1 ) );
154	}
155	}
156
157	// Fix reply to comment links, whoever decided this should be a GET variable?
158	// phpcs:ignore WordPress.Security -- We know this is scary.
159	if ( isset( $_SERVER['REQUEST_URI'] ) && \preg_match( '`(\?replytocom=[^&]+)`', \sanitize_text_field( $_SERVER['REQUEST_URI'] ), $matches ) ) {
160	$proper_url .= \str_replace( '?replytocom=', '#comment-', $matches[0] );
161	}
162	unset( $matches );
163
164	return $proper_url;
165	}
166
167	/**
168	* Returns the proper URL for front page.
169	*
170	* @return string The proper URL.
171	*/
172	public function front_page_url() {
173	if ( $this->current_page_helper->is_home_posts_page() ) {
174	return \home_url( '/' );
175	}
176	if ( $this->current_page_helper->is_home_static_page() ) {
177	return \get_permalink( $GLOBALS['post']->ID );
178	}
179	return '';
180	}
181
182	/**
183	* Returns the proper URL for 404 page.
184	*
185	* @param string $current_url The current URL.
186	* @return string The proper URL.
187	*/
188	public function page_not_found_url( $current_url ) {
189	if ( ! \is_multisite() \|\| \is_subdomain_install() \|\| ! \is_main_site() ) {
190	return '';
191	}
192
193	if ( $current_url !== \home_url() . '/blog/' && $current_url !== \home_url() . '/blog' ) {
194	return '';
195	}
196
197	if ( $this->current_page_helper->is_home_static_page() ) {
198	return \get_permalink( \get_option( 'page_for_posts' ) );
199	}
200
201	return \home_url();
202	}
203
204	/**
205	* Returns the proper URL for taxonomy page.
206	*
207	* @return string The proper URL.
208	*/
209	public function taxonomy_url() {
210	global $wp_query;
211	$term = $wp_query->get_queried_object();
212
213	if ( \is_feed() ) {
214	return \get_term_feed_link( $term->term_id, $term->taxonomy );
215	}
216	return \get_term_link( $term, $term->taxonomy );
217	}
218
219	/**
220	* Returns the proper URL for search page.
221	*
222	* @return string The proper URL.
223	*/
224	public function search_url() {
225	$s = \get_search_query();
226	return \home_url() . '/?s=' . \rawurlencode( $s );
227	}
228
229	/**
230	* Returns the proper URL for url with page param.
231	*
232	* @param string $proper_url The proper URL.
233	* @return string The proper URL.
234	*/
235	public function query_var_page_url( $proper_url ) {
236	global $wp_query;
237	if ( \is_search( $proper_url ) ) {
238	return \home_url() . '/page/' . $wp_query->query_vars['paged'] . '/?s=' . \rawurlencode( \get_search_query() );
239	}
240	return \user_trailingslashit( \trailingslashit( $proper_url ) . 'page/' . $wp_query->query_vars['paged'] );
241	}
242
243	/**
244	* Returns true if query is with page param.
245	*
246	* @param string $proper_url The proper URL.
247	* @return bool is query with page param.
248	*/
249	public function is_query_var_page( $proper_url ) {
250	global $wp_query;
251	if ( empty( $proper_url ) \|\| $wp_query->query_vars['paged'] === 0 \|\| $wp_query->post_count === 0 ) {
252	return false;
253	}
254	return true;
255	}
256
257	/**
258	* Redirects clean permalink.
259	*
260	* @param string $proper_url The proper URL.
261	* @return void
262	*/
263	public function do_clean_redirect( $proper_url ) {
264	$this->redirect_helper->set_header( 'Content-Type: redirect', true );
265	$this->redirect_helper->remove_header( 'Content-Type' );
266	$this->redirect_helper->remove_header( 'Last-Modified' );
267	$this->redirect_helper->remove_header( 'X-Pingback' );
268
269	$message = \sprintf(
270	/* translators: %1$s: Yoast SEO */
271	\__( '%1$s: unregistered URL parameter removed. See %2$s', 'wordpress-seo' ),
272	'Yoast SEO',
273	'https://yoa.st/advanced-crawl-settings',
274	);
275
276	$this->redirect_helper->do_safe_redirect( $proper_url, 301, $message );
277	}
278
279	/**
280	* Gets the type of URL.
281	*
282	* @return string The type of URL.
283	*/
284	public function get_url_type() {
285	if ( \is_singular() ) {
286	return 'singular_url';
287	}
288	if ( \is_front_page() ) {
289	return 'front_page_url';
290	}
291	if ( $this->current_page_helper->is_posts_page() ) {
292	return 'page_for_posts_url';
293	}
294	if ( \is_category() \|\| \is_tag() \|\| \is_tax() ) {
295	return 'taxonomy_url';
296	}
297	if ( \is_search() ) {
298	return 'search_url';
299	}
300	if ( \is_404() ) {
301	return 'page_not_found_url';
302	}
303	return '';
304	}
305
306	/**
307	* Returns the proper URL for posts page.
308	*
309	* @return string The proper URL.
310	*/
311	public function page_for_posts_url() {
312	return \get_permalink( \get_option( 'page_for_posts' ) );
313	}
314	}
315