flywp / includes / StaticSite / Engine / Cli / SiteCommand.php

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

265 lines 10.2 KB

Raw Download Zip

1	<?php
2	// phpcs:disable WordPress.WP.AlternativeFunctions -- WP-CLI command: STDERR and one local output file, no WP_Filesystem.
3
4	namespace FlyWP\StaticSite\Engine\Cli;
5
6	use FlyWP\StaticSite\Bootstrap;
7	use FlyWP\StaticSite\Engine\Discovery\RedirectSourceUrls;
8	use FlyWP\StaticSite\Engine\Discovery\Url;
9	use FlyWP\StaticSite\Engine\Discovery\UrlDiscovery;
10	use FlyWP\StaticSite\Engine\Discovery\WordPressGateway;
11	use FlyWP\StaticSite\Engine\Discovery\WpGateway;
12	use FlyWP\StaticSite\Engine\Incremental\ChangeSetBuilder;
13	use FlyWP\StaticSite\Engine\Incremental\ChangeStore;
14	use FlyWP\StaticSite\Engine\Incremental\WpChangeStore;
15	use Throwable;
16
17	/**
18	* WP-CLI command `wp flywp static site`: write the site data for the FlyWP static engine (site.json, contract N1).
19	*
20	* - The engine runs outside WordPress and reads this file.
21	* - The URLs are on the origin URL that the engine fetches (--origin-url, default the home URL), as in the old engine.
22	* The change log keeps retry URLs in this form, so the origin must be the same for each build.
23	* - A URL that is not valid UTF-8 cannot be written to JSON. The command drops it and writes a warning to STDERR.
24	* - max_change_id is the highest change row that the file includes. `wp flywp static changes clear` deletes
25	* the rows up to it after the build.
26	*
27	* @since 1.8.0
28	*/
29	class SiteCommand extends JsonCommand {
30
31	/**
32	* Version of the site.json format.
33	*/
34	const VERSION = 1;
35
36	const MODES = [ 'full', 'incremental' ];
37
38	/**
39	* @var callable function( string $mode, string\|null $origin_url ): array. The site.json data.
40	*/
41	private $snapshot;
42
43	/**
44	* @var callable function( string $message ). Writes one warning.
45	*/
46	private $warning;
47
48	/**
49	* @param callable\|null $snapshot Data factory. self::wordpress_snapshot() when null.
50	* @param callable\|null $output Line writer. STDOUT when null.
51	* @param callable\|null $warning Warning writer. STDERR when null.
52	*/
53	public function __construct( $snapshot = null, $output = null, $warning = null ) {
54	parent::__construct( $output );
55
56	$this->snapshot = $snapshot ? $snapshot : [ self::class, 'wordpress_snapshot' ];
57	$this->warning = $warning ? $warning : function ( $message ) {
58	fwrite( STDERR, 'Warning: ' . $message . "\n" );
59	};
60	}
61
62	/**
63	* Write the site data for a static build.
64	*
65	* ## OPTIONS
66	*
67	* [--output=<file>]
68	* : The JSON file to write. Necessary.
69	*
70	* [--mode=<mode>]
71	* : full or incremental. An incremental file also has the change set.
72	* ---
73	* default: full
74	* ---
75	*
76	* [--origin-url=<url>]
77	* : The URL that the engine fetches. The URLs in the file are on this URL. Default: the home URL.
78	*
79	* ## EXAMPLES
80	*
81	* wp flywp static site --output=/tmp/flywp-static-12.json --mode=incremental --origin-url=https://example.com
82	*
83	* @param array $args Positional arguments.
84	* @param array $assoc_args Named arguments.
85	*
86	* @return void
87	*/
88	public function __invoke( $args, $assoc_args ) {
89	$this->execute( $assoc_args );
90	}
91
92	/**
93	* {@inheritdoc}
94	*/
95	public function run( array $assoc_args ) {
96	$file = isset( $assoc_args['output'] ) ? (string) $assoc_args['output'] : '';
97	$mode = isset( $assoc_args['mode'] ) ? (string) $assoc_args['mode'] : 'full';
98	$origin = isset( $assoc_args['origin-url'] ) ? rtrim( (string) $assoc_args['origin-url'], '/' ) : null;
99	$site = null;
100
101	if ( $file === '' ) {
102	return $this->result( 'Set --output.' );
103	}
104
105	if ( ! in_array( $mode, self::MODES, true ) ) {
106	return $this->result( sprintf( 'Unknown mode "%s". Use full or incremental.', $mode ) );
107	}
108
109	if ( $origin !== null && ! Url::is_http( $origin ) ) {
110	return $this->result( 'The --origin-url value must be an absolute http or https URL.' );
111	}
112
113	try {
114	// Plugin hooks and filters run during discovery. Their output must not go to stdout.
115	Bootstrap::quietly(
116	function () use ( $mode, $origin, &$site ) {
117	$site = call_user_func( $this->snapshot, $mode, $origin );
118	}
119	);
120	} catch ( Throwable $e ) {
121	return $this->result( $e->getMessage() );
122	}
123
124	$json = json_encode( $this->without_invalid_urls( $site ), JSON_UNESCAPED_SLASHES \| JSON_UNESCAPED_UNICODE );
125
126	if ( $json === false ) {
127	return $this->result( sprintf( 'Unable to encode the site data: %s.', json_last_error_msg() ) );
128	}
129
130	if ( file_put_contents( $file, $json ) === false ) {
131	return $this->result( sprintf( 'Unable to write %s.', $file ) );
132	}
133
134	return $this->result();
135	}
136
137	/**
138	* The site data of this WordPress site.
139	*
140	* @param string $mode full or incremental.
141	* @param string\|null $origin_url Origin URL that the engine fetches, without trailing slash. Null for the home URL.
142	*
143	* @return array
144	*/
145	public static function wordpress_snapshot( $mode, $origin_url = null ) {
146	$origin_url = $origin_url ? $origin_url : rtrim( home_url(), '/' );
147	$site_urls = [ home_url(), site_url() ];
148
149	return self::snapshot(
150	$mode,
151	$origin_url,
152	[
153	'plugin_version' => FLYWP_VERSION,
154	'home_url' => home_url(),
155	'site_url' => site_url(),
156	'respect_noindex' => (string) get_option( 'blog_public' ) !== '0',
157	'pretty_permalinks' => (string) get_option( 'permalink_structure' ) !== '',
158	'pagination_base' => isset( $GLOBALS['wp_rewrite']->pagination_base ) ? (string) $GLOBALS['wp_rewrite']->pagination_base : 'page',
159	'redirect_sources' => RedirectSourceUrls::all( $origin_url, $site_urls ),
160	'redirect_wildcards' => RedirectSourceUrls::wildcard_rules( $origin_url, $site_urls ),
161	],
162	new WpGateway(),
163	new WpChangeStore()
164	);
165	}
166
167	/**
168	* The site.json data.
169	*
170	* - The URLs are on the origin URL. The discovery and the change set also accept URLs on the home URL.
171	* - The change rows are read before the URL discovery. A change during the discovery stays for the next build.
172	* - Incremental: max_change_id is the highest row of the change set, so the clear does not delete a newer row.
173	*
174	* @param string $mode full or incremental.
175	* @param string $origin_url Origin URL that the engine fetches.
176	* @param array $site Site values: plugin_version, home_url, site_url, respect_noindex,
177	* pretty_permalinks, pagination_base, redirect_sources, redirect_wildcards.
178	* @param WordPressGateway $wp Gateway.
179	* @param ChangeStore $changes Change rows.
180	*
181	* @return array
182	*/
183	public static function snapshot( $mode, $origin_url, array $site, WordPressGateway $wp, ChangeStore $changes ) {
184	$discovery = new UrlDiscovery( $wp, $origin_url, UrlDiscovery::default_sources( $wp, $origin_url ) );
185	$change_set = null;
186
187	if ( $mode === 'incremental' ) {
188	$set = ( new ChangeSetBuilder( $changes, $wp, $origin_url ) )->build( 0 );
189	$max_change_id = $set->max_change_id;
190	$change_set = [
191	'full_reason' => $set->full_reason,
192	'urls' => $set->urls,
193	'deleted_urls' => $set->deleted_urls,
194	'lists' => $set->lists,
195	];
196	} else {
197	$max_change_id = $changes->max_id();
198	}
199
200	return array_merge(
201	[ 'version' => self::VERSION ],
202	$site,
203	[
204	'urls' => iterator_to_array( $discovery->urls(), false ),
205	'excluded_urls' => iterator_to_array( $discovery->excluded_urls(), false ),
206	'max_change_id' => (int) $max_change_id,
207	'change_set' => $change_set,
208	]
209	);
210	}
211
212	/**
213	* Remove the URLs that are not valid UTF-8, because json_encode() cannot write them.
214	*
215	* @param array $site The site.json data.
216	*
217	* @return array
218	*/
219	private function without_invalid_urls( array $site ) {
220	$site['urls'] = $this->valid_items( $site['urls'], 'url' );
221	$site['excluded_urls'] = $this->valid_items( $site['excluded_urls'] );
222	$site['redirect_sources'] = $this->valid_items( $site['redirect_sources'] );
223	$site['redirect_wildcards'] = $this->valid_items( $site['redirect_wildcards'], 'from', 'to' );
224
225	if ( is_array( $site['change_set'] ) ) {
226	$site['change_set']['urls'] = $this->valid_items( $site['change_set']['urls'] );
227	$site['change_set']['deleted_urls'] = $this->valid_items( $site['change_set']['deleted_urls'] );
228	$site['change_set']['lists'] = $this->valid_items( $site['change_set']['lists'], 'url' );
229	}
230
231	return $site;
232	}
233
234	/**
235	* The items whose URLs are valid UTF-8. Writes one warning for each item that it removes.
236	*
237	* @param array $items URLs, or arrays or objects with URLs.
238	* @param string ...$keys The URL keys of an item. None when the item is the URL.
239	*
240	* @return array
241	*/
242	private function valid_items( array $items, ...$keys ) {
243	$valid = [];
244
245	foreach ( $items as $item ) {
246	$urls = $keys ? array_intersect_key( (array) $item, array_flip( $keys ) ) : [ $item ];
247	$invalid = array_filter(
248	$urls,
249	function ( $url ) {
250	return is_string( $url ) && preg_match( '//u', $url ) !== 1;
251	}
252	);
253
254	if ( $invalid ) {
255	call_user_func( $this->warning, sprintf( 'Skipped a URL that is not valid UTF-8: %s', json_encode( reset( $invalid ), JSON_UNESCAPED_SLASHES \| JSON_INVALID_UTF8_SUBSTITUTE ) ) );
256	continue;
257	}
258
259	$valid[] = $item;
260	}
261
262	return $valid;
263	}
264	}
265