wordpress-seo / src / integrations / front-end / crawl-cleanup-searches.php

crawl-cleanup-searches.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI trunk, at src/integrations/front-end/crawl-cleanup-searches.php

207 lines 5.5 KB

Raw Download Zip

1	<?php
2
3	namespace Yoast\WP\SEO\Integrations\Front_End;
4
5	use WP_Query;
6	use Yoast\WP\SEO\Conditionals\Front_End_Conditional;
7	use Yoast\WP\SEO\Helpers\Options_Helper;
8	use Yoast\WP\SEO\Helpers\Redirect_Helper;
9	use Yoast\WP\SEO\Integrations\Integration_Interface;
10
11	/**
12	* Class Crawl_Cleanup_Searches.
13	*/
14	class Crawl_Cleanup_Searches implements Integration_Interface {
15
16	/**
17	* Patterns to match against to find spam.
18	*
19	* @var array
20	*/
21	private $patterns = [
22	'/[：（）【】［］]+/u',
23	'/(TALK\|QQ)\:/iu',
24	];
25
26	/**
27	* The options helper.
28	*
29	* @var Options_Helper
30	*/
31	private $options_helper;
32
33	/**
34	* The redirect helper.
35	*
36	* @var Redirect_Helper
37	*/
38	private $redirect_helper;
39
40	/**
41	* Crawl_Cleanup_Searches integration constructor.
42	*
43	* @param Options_Helper $options_helper The option helper.
44	* @param Redirect_Helper $redirect_helper The redirect helper.
45	*/
46	public function __construct( Options_Helper $options_helper, Redirect_Helper $redirect_helper ) {
47	$this->options_helper = $options_helper;
48	$this->redirect_helper = $redirect_helper;
49	}
50
51	/**
52	* Initializes the integration.
53	*
54	* This is the place to register hooks and filters.
55	*
56	* @return void
57	*/
58	public function register_hooks() {
59	if ( $this->options_helper->get( 'search_cleanup' ) ) {
60	\add_filter( 'pre_get_posts', [ $this, 'validate_search' ] );
61	}
62	if ( $this->options_helper->get( 'redirect_search_pretty_urls' ) && ! empty( \get_option( 'permalink_structure' ) ) ) {
63	\add_action( 'template_redirect', [ $this, 'maybe_redirect_searches' ], 2 );
64	}
65	}
66
67	/**
68	* Returns the conditionals based in which this loadable should be active.
69	*
70	* @return array The array of conditionals.
71	*/
72	public static function get_conditionals() {
73	return [ Front_End_Conditional::class ];
74	}
75
76	/**
77	* Check if we want to allow this search to happen.
78	*
79	* @param WP_Query $query The main query.
80	*
81	* @return WP_Query
82	*/
83	public function validate_search( WP_Query $query ) {
84	if ( ! $query->is_search() ) {
85	return $query;
86	}
87	// First check against emoji and patterns we might not want.
88	$this->check_unwanted_patterns( $query );
89
90	// Then limit characters if still needed.
91	$this->limit_characters();
92
93	return $query;
94	}
95
96	/**
97	* Redirect pretty search URLs to the "raw" equivalent
98	*
99	* @return void
100	*/
101	public function maybe_redirect_searches() {
102	if ( ! \is_search() ) {
103	return;
104	}
105
106	// phpcs:ignore WordPress.Security.ValidatedSanitizedInput
107	if ( isset( $_SERVER['REQUEST_URI'] ) && \stripos( $_SERVER['REQUEST_URI'], '/search/' ) === 0 ) {
108	$args = [];
109
110	// phpcs:ignore WordPress.Security.ValidatedSanitizedInput
111	$parsed = \wp_parse_url( $_SERVER['REQUEST_URI'] );
112
113	if ( ! empty( $parsed['query'] ) ) {
114	\wp_parse_str( $parsed['query'], $args );
115	}
116
117	$args['s'] = \get_search_query();
118
119	$proper_url = \home_url( '/' );
120
121	if ( (int) \get_query_var( 'paged' ) > 1 ) {
122	$proper_url .= \sprintf( 'page/%s/', \get_query_var( 'paged' ) );
123	unset( $args['paged'] );
124	}
125
126	$proper_url = \add_query_arg( \array_map( 'rawurlencode_deep', $args ), $proper_url );
127
128	if ( ! empty( $parsed['fragment'] ) ) {
129	$proper_url .= '#' . \rawurlencode( $parsed['fragment'] );
130	}
131
132	$this->redirect_away( 'We redirect pretty URLs to the raw format.', $proper_url );
133	}
134	}
135
136	/**
137	* Check query against unwanted search patterns.
138	*
139	* @param WP_Query $query The main WordPress query.
140	*
141	* @return void
142	*/
143	private function check_unwanted_patterns( WP_Query $query ) {
144	$s = \rawurldecode( $query->query_vars['s'] );
145	if ( $this->options_helper->get( 'search_cleanup_emoji' ) && $this->has_emoji( $s ) ) {
146	$this->redirect_away( 'We don\'t allow searches with emojis and other special characters.' );
147	}
148
149	if ( ! $this->options_helper->get( 'search_cleanup_patterns' ) ) {
150	return;
151	}
152	foreach ( $this->patterns as $pattern ) {
153	$outcome = \preg_match( $pattern, $s, $matches );
154	if ( $outcome && $matches !== [] ) {
155	$this->redirect_away( 'Your search matched a common spam pattern.' );
156	}
157	}
158	}
159
160	/**
161	* Redirect to the homepage for invalid searches.
162	*
163	* @param string $reason The reason for redirecting away.
164	* @param string $to_url The URL to redirect to.
165	*
166	* @return void
167	*/
168	private function redirect_away( $reason, $to_url = '' ) {
169	if ( empty( $to_url ) ) {
170	$to_url = \get_home_url();
171	}
172
173	$this->redirect_helper->do_safe_redirect( $to_url, 301, 'Yoast Search Filtering: ' . $reason );
174	}
175
176	/**
177	* Limits the number of characters in the search query.
178	*
179	* @return void
180	*/
181	private function limit_characters() {
182	// We retrieve the search term unescaped because we want to count the characters properly. We make sure to escape it afterwards, if we do something with it.
183	$unescaped_s = \get_search_query( false );
184
185	// We then unslash the search term, again because we want to count the characters properly. We make sure to slash it afterwards, if we do something with it.
186	$raw_s = \wp_unslash( $unescaped_s );
187	if ( \mb_strlen( $raw_s, 'UTF-8' ) > $this->options_helper->get( 'search_character_limit' ) ) {
188	$new_s = \mb_substr( $raw_s, 0, $this->options_helper->get( 'search_character_limit' ), 'UTF-8' );
189	\set_query_var( 's', \wp_slash( \esc_attr( $new_s ) ) );
190	}
191	}
192
193	/**
194	* Determines if a text string contains an emoji or not.
195	*
196	* @param string $text The text string to detect emoji in.
197	*
198	* @return bool
199	*/
200	private function has_emoji( $text ) {
201	$emojis_regex = '/([^-\p{L}\x00-\x7F]+)/u';
202	\preg_match( $emojis_regex, $text, $matches );
203
204	return ! empty( $matches );
205	}
206	}
207