flywp / includes / StaticSite / Engine / Discovery / RedirectSourceUrls.php

RedirectSourceUrls.php in FlyWP Helper – Page Cache, Page Optimization, Emails for FlyWP Server Control Panel 1.8.0, at includes/StaticSite/Engine/Discovery/RedirectSourceUrls.php

285 lines 8.5 KB

Raw Download Zip

1	<?php
2
3	namespace FlyWP\StaticSite\Engine\Discovery;
4
5	/**
6	* Source URLs of redirect plugins. The crawler fetches them, the origin answers with a 3xx,
7	* and the engine writes the rule.
8	*
9	* - Plugins: Redirection, Rank Math, Yoast SEO Premium, Safe Redirect Manager and Simple 301 Redirects.
10	* - One class that returns absolute URLs on the origin.
11	* - Absolute sources on a site host (with or without "www.", http or https) move to the origin scheme and host.
12	* - No Rank Math row limit. The _redirects limits apply later.
13	* - Simple 301 wildcard sources become dynamic rules only when its wildcard option is on.
14	* - `wp flywp static site` writes the result to site.json.
15	*
16	* - A plugin is read only when its classes or functions exist (the plugin is active).
17	* - Regex rules and sources with "*" are not crawled.
18	*
19	* @since 1.8.0
20	*/
21	class RedirectSourceUrls {
22
23	const RANK_MATH_PAGE_SIZE = 500;
24
25	const CODES = [ 301, 302, 307, 308 ];
26
27	/**
28	* Absolute source URLs on the origin.
29	*
30	* @param string $origin_url Origin URL.
31	* @param string[] $site_urls Other URLs of the site (home URL, site URL).
32	*
33	* @return string[]
34	*/
35	public static function all( $origin_url, array $site_urls = [] ) {
36	$sources = array_merge(
37	self::redirection(),
38	self::rank_math(),
39	self::yoast_premium(),
40	self::safe_redirect_manager(),
41	self::simple_301_redirects()
42	);
43
44	$urls = [];
45
46	foreach ( $sources as $source ) {
47	$url = self::to_origin( (string) $source, $origin_url, $site_urls );
48
49	if ( $url !== null && strpos( $url, '*' ) === false ) {
50	$urls[ $url ] = true;
51	}
52	}
53
54	return array_keys( $urls );
55	}
56
57	/**
58	* Simple 301 Redirects wildcard rules, when its wildcard option is on.
59	*
60	* @param string $origin_url Origin URL.
61	* @param string[] $site_urls Other URLs of the site.
62	*
63	* @return array<int, array{from: string, to: string, status: int}>
64	*/
65	public static function wildcard_rules( $origin_url, array $site_urls = [] ) {
66	if ( ! class_exists( 'Simple301Redirects' ) \|\| get_option( '301_redirects_wildcard' ) !== 'true' ) {
67	return [];
68	}
69
70	$rules = [];
71
72	foreach ( (array) get_option( '301_redirects' ) as $from => $to ) {
73	$from_url = self::to_origin( (string) $from, $origin_url, $site_urls );
74
75	if ( $from_url === null \|\| strpos( $from_url, '*' ) === false ) {
76	continue;
77	}
78
79	$to_url = self::to_origin( (string) $to, $origin_url, $site_urls );
80	$rules[] = [
81	'from' => $from_url,
82	'to' => $to_url !== null ? $to_url : (string) $to,
83	'status' => 301,
84	];
85	}
86
87	return $rules;
88	}
89
90	/**
91	* Move a source onto the origin URL.
92	*
93	* @param string $source Relative path or absolute URL.
94	* @param string $origin_url Origin URL.
95	* @param string[] $site_urls Other URLs of the site.
96	*
97	* @return string\|null Null for an empty source or a source on another host.
98	*/
99	public static function to_origin( $source, $origin_url, array $site_urls ) {
100	$source = trim( $source );
101
102	if ( $source === '' ) {
103	return null;
104	}
105
106	// Plugins store site-root paths ("/blog/old-page" on a site in "/blog"), so resolve them like a browser.
107	if ( ! Url::is_http( $source ) ) {
108	return Url::resolve( rtrim( $origin_url, '/' ) . '/', $source );
109	}
110
111	$hosts = [];
112
113	foreach ( array_merge( [ $origin_url ], $site_urls ) as $url ) {
114	$hosts[ self::bare_host( $url ) ] = true;
115	}
116
117	if ( ! isset( $hosts[ self::bare_host( $source ) ] ) ) {
118	return null;
119	}
120
121	$parts = Url::parts( $source );
122	$origin = Url::parts( $origin_url );
123
124	$parts['scheme'] = $origin['scheme'];
125	$parts['host'] = $origin['host'];
126	$parts['port'] = $origin['port'];
127	$parts['user'] = '';
128	$parts['pass'] = '';
129
130	return Url::build( $parts );
131	}
132
133	/**
134	* Host without "www.".
135	*
136	* @param string $url URL.
137	*
138	* @return string
139	*/
140	private static function bare_host( $url ) {
141	return (string) preg_replace( '/^www\./', '', Url::parts( $url )['host'] );
142	}
143
144	/**
145	* Redirection plugin.
146	*
147	* @return string[]
148	*/
149	private static function redirection() {
150	if ( ! class_exists( 'Red_Item' ) \|\| ! method_exists( 'Red_Item', 'get_all' ) ) {
151	return [];
152	}
153
154	$urls = [];
155
156	foreach ( \Red_Item::get_all() as $item ) {
157	if ( ! $item->is_enabled() \|\| $item->is_dynamic() \|\| $item->is_regex() ) {
158	continue;
159	}
160
161	if ( $item->get_match_type() !== 'url' \|\| $item->get_action_type() !== 'url' ) {
162	continue;
163	}
164
165	$urls[] = $item->get_url();
166	}
167
168	return $urls;
169	}
170
171	/**
172	* Rank Math, all active rules, page by page.
173	*
174	* @return string[]
175	*/
176	private static function rank_math() {
177	if ( ! class_exists( '\RankMath\Redirections\DB' ) ) {
178	return [];
179	}
180
181	$urls = [];
182	$first_id = null;
183
184	for ( $page = 1; ; $page++ ) {
185	$result = \RankMath\Redirections\DB::get_redirections(
186	[
187	'status' => 'active',
188	'limit' => self::RANK_MATH_PAGE_SIZE,
189	'paged' => $page,
190	]
191	);
192	$redirects = isset( $result['redirections'] ) ? array_values( (array) $result['redirections'] ) : [];
193	$page_id = isset( $redirects[0]['id'] ) ? $redirects[0]['id'] : null;
194
195	// Stop when the plugin ignores "paged" and gives the first page again.
196	if ( ! $redirects \|\| ( $page > 1 && $page_id === $first_id ) ) {
197	break;
198	}
199
200	$first_id = $page === 1 ? $page_id : $first_id;
201
202	foreach ( $redirects as $redirect ) {
203	if ( ! in_array( (int) $redirect['header_code'], self::CODES, true ) ) {
204	continue;
205	}
206
207	foreach ( (array) maybe_unserialize( $redirect['sources'] ) as $source ) {
208	if ( isset( $source['comparison'], $source['pattern'] ) && $source['comparison'] === 'exact' ) {
209	$urls[] = $source['pattern'];
210	}
211	}
212	}
213
214	if ( count( $redirects ) < self::RANK_MATH_PAGE_SIZE ) {
215	break;
216	}
217	}
218
219	return $urls;
220	}
221
222	/**
223	* Yoast SEO Premium.
224	*
225	* @return string[]
226	*/
227	private static function yoast_premium() {
228	if ( ! class_exists( 'WPSEO_Redirect_Manager' ) ) {
229	return [];
230	}
231
232	$urls = [];
233
234	foreach ( (array) get_option( 'wpseo-premium-redirects-base' ) as $item ) {
235	if ( is_array( $item ) && isset( $item['format'], $item['type'], $item['origin'] ) && $item['format'] === 'plain' && in_array( (int) $item['type'], self::CODES, true ) ) {
236	$urls[] = $item['origin'];
237	}
238	}
239
240	return $urls;
241	}
242
243	/**
244	* Safe Redirect Manager.
245	*
246	* @return string[]
247	*/
248	private static function safe_redirect_manager() {
249	if ( ! class_exists( 'SRM_Redirect' ) \|\| ! function_exists( 'srm_get_redirects' ) \|\| ! function_exists( 'srm_get_max_redirects' ) ) {
250	return [];
251	}
252
253	$redirects = srm_get_redirects(
254	[
255	'posts_per_page' => srm_get_max_redirects(), // phpcs:ignore WordPress.WP.PostsPerPage -- the limit of Safe Redirect Manager itself.
256	'post_status' => 'publish',
257	]
258	);
259	$urls = [];
260
261	foreach ( (array) $redirects as $item ) {
262	if ( empty( $item['enable_regex'] ) && in_array( (int) $item['status_code'], self::CODES, true ) ) {
263	$urls[] = $item['redirect_from'];
264	}
265	}
266
267	return $urls;
268	}
269
270	/**
271	* Simple 301 Redirects. Sources with "*" are removed in all().
272	*
273	* @return string[]
274	*/
275	private static function simple_301_redirects() {
276	if ( ! class_exists( 'Simple301Redirects' ) ) {
277	return [];
278	}
279
280	$redirects = get_option( '301_redirects' );
281
282	return is_array( $redirects ) ? array_map( 'strval', array_keys( $redirects ) ) : [];
283	}
284	}
285