solid-performance / src / Performance / Preload / Preloader.php

Preloader.php in Solid Performance – Your No-Code Caching, Performance, & Page Speed Solution 2.0.0, at src/Performance/Preload/Preloader.php

317 lines 8.5 KB

Raw Download Zip

1	<?php
2	/**
3	* The preloader that concurrently processes crawled sitemap URLs.
4	*
5	* @package SolidWP\Performance
6	*/
7
8	declare( strict_types=1 );
9
10	namespace SolidWP\Performance\Preload;
11
12	use SolidWP\Performance\Preload\Contracts\Preloadable;
13	use SolidWP\Performance\Preload\Limiter\Batch_Limiter;
14	use SolidWP\Performance\Preload\Sitemap\Repositories\Contracts\Sitemap_Repository;
15	use SolidWP\Performance\Preload\Traits\With_Request_Params;
16	use SolidWP\Performance\Preload\Traits\With_Timeout;
17	use SolidWP\Performance\Psr\Log\LoggerInterface;
18	use SolidWP\Performance\StellarWP\SuperGlobals\SuperGlobals;
19	use SolidWP\Performance\Symfony\Contracts\HttpClient\Exception\ClientExceptionInterface;
20	use SolidWP\Performance\Symfony\Contracts\HttpClient\Exception\ServerExceptionInterface;
21	use SolidWP\Performance\Symfony\Contracts\HttpClient\Exception\TimeoutExceptionInterface;
22	use SolidWP\Performance\Symfony\Contracts\HttpClient\Exception\TransportExceptionInterface;
23	use SolidWP\Performance\Symfony\Contracts\HttpClient\ResponseInterface;
24	use Throwable;
25
26	/**
27	* The preloader that concurrently processes crawled sitemap URLs.
28	*
29	* @package SolidWP\Performance
30	*/
31	final class Preloader implements Preloadable {
32
33	use With_Timeout;
34	use With_Request_Params;
35
36	/**
37	* @var Sitemap_Repository
38	*/
39	private Sitemap_Repository $repository;
40
41	/**
42	* @var Preload_Scheduler
43	*/
44	private Preload_Scheduler $scheduler;
45
46	/**
47	* @var Client
48	*/
49	private Client $client;
50
51	/**
52	* @var Batch_Limiter
53	*/
54	private Batch_Limiter $batch_limiter;
55
56	/**
57	* @var LoggerInterface
58	*/
59	private LoggerInterface $logger;
60
61	/**
62	* @var Preload_Mode_Manager
63	*/
64	private Preload_Mode_Manager $preload_mode_manager;
65
66	/**
67	* @var Trigger_Url
68	*/
69	private Trigger_Url $trigger_url;
70
71	/**
72	* How many seconds to sleep between batches.
73	*
74	* @var int
75	*/
76	private int $sleep;
77
78	/**
79	* The maximum number of errors before high performance mode is disabled.
80	*
81	* @var int
82	*/
83	private int $max_errors_per_batch;
84
85	/**
86	* @var int[]
87	*/
88	private array $error_codes;
89
90	/**
91	* @param Sitemap_Repository $repository The Sitemap Repository.
92	* @param Preload_Scheduler $scheduler The Preload Scheduler.
93	* @param Client $client The Preload Client.
94	* @param Batch_Limiter $batch_limiter The batch rate limiter.
95	* @param LoggerInterface $logger The logger.
96	* @param Preload_Mode_Manager $preload_mode_manager The preload mode manager.
97	* @param Trigger_Url $trigger_url The home URL to use to trigger a preloader batch.
98	* @param int $sleep How many seconds to sleep between batches.
99	* @param int $max_errors_per_batch The maximum number of errors before high performance mode is disabled.
100	* @param int[] $error_codes The http response codes considered errors.
101	*/
102	public function __construct(
103	Sitemap_Repository $repository,
104	Preload_Scheduler $scheduler,
105	Client $client,
106	Batch_Limiter $batch_limiter,
107	LoggerInterface $logger,
108	Preload_Mode_Manager $preload_mode_manager,
109	Trigger_Url $trigger_url,
110	int $sleep = 1,
111	int $max_errors_per_batch = 3,
112	array $error_codes = [
113	429,
114	503,
115	508,
116	]
117	) {
118	$this->repository = $repository;
119	$this->scheduler = $scheduler;
120	$this->client = $client;
121	$this->batch_limiter = $batch_limiter;
122	$this->logger = $logger;
123	$this->preload_mode_manager = $preload_mode_manager;
124	$this->trigger_url = $trigger_url;
125	$this->sleep = $sleep;
126	$this->max_errors_per_batch = $max_errors_per_batch;
127	$this->error_codes = $error_codes;
128	}
129
130	/**
131	* Send a preload request.
132	*
133	* @action solidwp/performance/cache/purge/url/after
134	*
135	* @param string $url The full or relative URL to preload.
136	* @param array<string, mixed> $params The query parameters to attach to the request.
137	*
138	* @throws TransportExceptionInterface When an error happens at the transport level.
139	*
140	* @return ResponseInterface
141	*/
142	public function preload_url( string $url, array $params = [] ): ResponseInterface {
143	/**
144	* Filters the Accept-Encoding HTTP header to decide the order of which cache file type to create when preloading.
145	*
146	* @link https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Accept-Encoding
147	*
148	* @param string $accept_encoding The valid Accept-Encoding string.
149	*/
150	$encoding = apply_filters( 'solidwp/performance/preload/accept_encoding', 'gzip, deflate, br, zstd' );
151
152	return $this->client->get(
153	$url,
154	$params,
155	[
156	'Cache-Control' => 'no-cache',
157	'Accept-Encoding' => $encoding,
158	],
159	[
160	'timeout' => $this->get_request_timeout(),
161	]
162	);
163	}
164
165	/**
166	* Preload a batch of crawled sitemap URLs.
167	*
168	* @action template_redirect
169	*
170	* @return void
171	*/
172	public function preload(): void {
173	// The preloader query string parameter isn't included.
174	if ( ! SuperGlobals::get_get_var( self::PRELOAD_TRIGGER_PARAM ) ) {
175	return;
176	}
177
178	// This should only run on the frontend.
179	if ( is_admin() \|\| wp_doing_ajax() \|\| wp_is_serving_rest_request() ) {
180	return;
181	}
182
183	// Check if the preloader is actually running.
184	if ( ! $this->scheduler->is_running() ) {
185	return;
186	}
187
188	// We're out of URls to crawl, complete the preloading.
189	if ( $this->repository->count() <= 0 ) {
190	$this->logger->debug( 'Completing Preloader: No more URLs' );
191	$this->scheduler->complete();
192
193	return;
194	}
195
196	$batch_size = $this->batch_limiter->get_batch_size();
197
198	$this->logger->debug(
199	'URL batch size: {batch_size}',
200	[
201	'batch_size' => $batch_size,
202	]
203	);
204
205	// Only sleep if the batch limiter is at the minimum.
206	if ( $batch_size === Batch_Limiter::MIN_BATCH_SIZE ) {
207	$this->logger->debug(
208	'Minimum batch size detected. Sleeping for {sleep_seconds}s',
209	[
210	'sleep_seconds' => $this->sleep,
211	]
212	);
213
214	sleep( $this->sleep );
215	}
216
217	$urls = $this->repository->pull( $batch_size );
218
219	$responses = [];
220
221	foreach ( $urls as $url ) {
222	$params = [];
223
224	if ( $this->scheduler->is_forced() ) {
225	// We're doing an entire site preload, cache bust every request.
226	$params = $this->get_preload_cache_bust_params();
227	}
228
229	try {
230	$responses[] = $this->preload_url( $url, $params );
231	} catch ( TimeoutExceptionInterface $e ) {
232	// We expected timeouts.
233	$this->logger->debug( $e->getMessage() );
234	}
235	}
236
237	if ( $responses ) {
238	// Trigger the next batch, use the homepage because it has to exist, right?
239	try {
240	$responses[] = $this->preload_url(
241	$this->trigger_url->get(),
242	$this->get_preload_trigger_params()
243	);
244	} catch ( TimeoutExceptionInterface $e ) {
245	// We expected timeouts.
246	$this->logger->debug( $e->getMessage() );
247	}
248	}
249
250	$errors = 0;
251
252	// Process responses for the batch.
253	foreach ( $responses as $response ) {
254	try {
255	$response->getHeaders();
256	} catch ( ServerExceptionInterface \|ClientExceptionInterface $e ) {
257	$this->logger->debug(
258	$e->getMessage(),
259	[
260	'exception' => $e,
261	]
262	);
263
264	try {
265	$status_code = $e->getResponse()->getStatusCode();
266
267	// Check if we're hitting any arbitrary resource limits or the server is rate limiting us.
268	if ( ! in_array( $status_code, $this->error_codes, true ) ) {
269	continue;
270	}
271
272	++$errors;
273
274	if ( $errors < $this->max_errors_per_batch ) {
275	continue;
276	}
277
278	if ( $this->preload_mode_manager->is_high_performance_mode() ) {
279	$this->logger->error(
280	'Max response errors "{errors}" reached. Disabling high performance mode.',
281	[
282	'errors' => $errors,
283	'last_status_code' => $status_code,
284	'response_info' => $e->getResponse()->getInfo(),
285	]
286	);
287
288	// This is currently permanent, but can be reversed with `wp solid perf preload perf-mode on`.
289	$this->preload_mode_manager->disable_high_performance_mode();
290	} else {
291	$this->logger->warning(
292	'Max response errors "{errors}" reached. High performance mode already disabled!',
293	[
294	'errors' => $errors,
295	'last_status_code' => $status_code,
296	'response_info' => $e->getResponse()->getInfo(),
297	]
298	);
299	}
300	} catch ( TransportExceptionInterface $e ) {
301	$this->logger->warning(
302	'Response threw exception when checking status code',
303	[
304	'exception' => $e,
305	]
306	);
307	}
308	} catch ( Throwable $e ) {
309	// Ignore any requests that failed for any other reason.
310	$this->logger->debug( $e->getMessage() );
311	} finally {
312	$response->cancel();
313	}
314	}
315	}
316	}
317