| 1 |
<?php |
| 2 |
|
| 3 |
namespace Upress\EzCache; |
| 4 |
|
| 5 |
use Upress\EzCache\Utilities\Logger; |
| 6 |
|
| 7 |
/** |
| 8 |
* Cache Preloader. |
| 9 |
* |
| 10 |
* Inspired by WP Rocket's Preload module, ezCache's Preload feature warms up |
| 11 |
* the cache for all the public URLs of the website by crawling them in |
| 12 |
* background batches. Sources of URLs include: |
| 13 |
* - The homepage and links found on it |
| 14 |
* - The XML sitemap (auto-detected or user provided) |
| 15 |
* - URLs added programmatically via the `ezcache_preload_queue_url` filter |
| 16 |
* |
| 17 |
* The queue is stored as a WordPress option to avoid the need for a database |
| 18 |
* table. URLs are processed in small batches by a recurring cron event so we |
| 19 |
* don't overload the host. Each URL is fetched twice (desktop + mobile UA) so |
| 20 |
* both cache variants are generated. |
| 21 |
*/ |
| 22 |
class Preload { |
| 23 |
|
| 24 |
const QUEUE_OPTION = 'ezcache_preload_queue'; |
| 25 |
const STATE_OPTION = 'ezcache_preload_state'; |
| 26 |
const CRON_HOOK = 'ezcache_preload_process_queue'; |
| 27 |
const CRON_INTERVAL = 'every_minute_ezcache'; |
| 28 |
const PROCESSED_OPTION = 'ezcache_preload_processed'; |
| 29 |
|
| 30 |
/** @var Preload */ |
| 31 |
protected static $instance; |
| 32 |
|
| 33 |
/** @var Cache */ |
| 34 |
protected $cache; |
| 35 |
|
| 36 |
/** @var object */ |
| 37 |
protected $settings; |
| 38 |
|
| 39 |
public static function instance() { |
| 40 |
if ( ! self::$instance ) { |
| 41 |
self::$instance = new self(); |
| 42 |
} |
| 43 |
|
| 44 |
return self::$instance; |
| 45 |
} |
| 46 |
|
| 47 |
public function __construct() { |
| 48 |
$this->cache = Cache::instance(); |
| 49 |
$this->settings = Settings::get_settings(); |
| 50 |
|
| 51 |
add_filter( 'cron_schedules', [ $this, 'register_cron_schedule' ] ); |
| 52 |
add_action( self::CRON_HOOK, [ $this, 'process_queue' ] ); |
| 53 |
|
| 54 |
// Trigger preload after cache is cleared (full or partial). |
| 55 |
add_action( 'ezcache_after_clear_cache', [ $this, 'maybe_start_preload' ], 10, 0 ); |
| 56 |
add_action( 'ezcache_after_clear_cache_single', [ $this, 'enqueue_post_url' ], 10, 1 ); |
| 57 |
} |
| 58 |
|
| 59 |
/** |
| 60 |
* Register a 1-minute cron schedule that we use to process the queue |
| 61 |
* |
| 62 |
* @param array $schedules |
| 63 |
* @return array |
| 64 |
*/ |
| 65 |
public function register_cron_schedule( $schedules ) { |
| 66 |
if ( ! isset( $schedules[ self::CRON_INTERVAL ] ) ) { |
| 67 |
$schedules[ self::CRON_INTERVAL ] = [ |
| 68 |
'interval' => 60, |
| 69 |
'display' => __( 'Every Minute (ezCache Preload)', 'ezcache' ), |
| 70 |
]; |
| 71 |
} |
| 72 |
|
| 73 |
return $schedules; |
| 74 |
} |
| 75 |
|
| 76 |
/** |
| 77 |
* Schedule (or unschedule) the recurring queue processor based on settings. |
| 78 |
*/ |
| 79 |
public function maybe_schedule_cron() { |
| 80 |
if ( ! empty( $this->settings->enable_preload ) ) { |
| 81 |
if ( ! wp_next_scheduled( self::CRON_HOOK ) ) { |
| 82 |
wp_schedule_event( time() + 30, self::CRON_INTERVAL, self::CRON_HOOK ); |
| 83 |
} |
| 84 |
} else { |
| 85 |
$timestamp = wp_next_scheduled( self::CRON_HOOK ); |
| 86 |
if ( $timestamp ) { |
| 87 |
wp_unschedule_event( $timestamp, self::CRON_HOOK ); |
| 88 |
} |
| 89 |
} |
| 90 |
} |
| 91 |
|
| 92 |
/** |
| 93 |
* Start a preload run if the feature is enabled. |
| 94 |
*/ |
| 95 |
public function maybe_start_preload() { |
| 96 |
if ( empty( $this->settings->enable_preload ) ) { |
| 97 |
return; |
| 98 |
} |
| 99 |
|
| 100 |
if ( empty( $this->settings->preload_on_cache_clear ) ) { |
| 101 |
return; |
| 102 |
} |
| 103 |
|
| 104 |
$this->start(); |
| 105 |
} |
| 106 |
|
| 107 |
/** |
| 108 |
* Build the initial queue and kick off the cron processor. |
| 109 |
* |
| 110 |
* @return array Status data. |
| 111 |
*/ |
| 112 |
public function start() { |
| 113 |
$urls = $this->collect_initial_urls(); |
| 114 |
$urls = apply_filters( 'ezcache_preload_initial_urls', $urls ); |
| 115 |
|
| 116 |
$urls = $this->normalize_urls( $urls ); |
| 117 |
|
| 118 |
update_option( self::QUEUE_OPTION, array_values( $urls ), false ); |
| 119 |
update_option( self::PROCESSED_OPTION, [], false ); |
| 120 |
update_option( self::STATE_OPTION, [ |
| 121 |
'status' => 'running', |
| 122 |
'total' => count( $urls ), |
| 123 |
'processed' => 0, |
| 124 |
'started' => time(), |
| 125 |
'finished' => 0, |
| 126 |
], false ); |
| 127 |
|
| 128 |
$this->maybe_schedule_cron(); |
| 129 |
|
| 130 |
// Run an immediate first batch so the user sees movement quickly. |
| 131 |
if ( ! defined( 'EZCACHE_PRELOAD_NO_IMMEDIATE' ) || ! EZCACHE_PRELOAD_NO_IMMEDIATE ) { |
| 132 |
$this->process_queue(); |
| 133 |
} |
| 134 |
|
| 135 |
return $this->get_status(); |
| 136 |
} |
| 137 |
|
| 138 |
/** |
| 139 |
* Cancel any in-progress preload. |
| 140 |
*/ |
| 141 |
public function stop() { |
| 142 |
delete_option( self::QUEUE_OPTION ); |
| 143 |
delete_option( self::PROCESSED_OPTION ); |
| 144 |
update_option( self::STATE_OPTION, [ |
| 145 |
'status' => 'cancelled', |
| 146 |
'total' => 0, |
| 147 |
'processed' => 0, |
| 148 |
'started' => 0, |
| 149 |
'finished' => time(), |
| 150 |
], false ); |
| 151 |
|
| 152 |
$timestamp = wp_next_scheduled( self::CRON_HOOK ); |
| 153 |
if ( $timestamp ) { |
| 154 |
wp_unschedule_event( $timestamp, self::CRON_HOOK ); |
| 155 |
} |
| 156 |
|
| 157 |
return $this->get_status(); |
| 158 |
} |
| 159 |
|
| 160 |
/** |
| 161 |
* Return the status of the running/last preload run. |
| 162 |
* |
| 163 |
* @return array |
| 164 |
*/ |
| 165 |
public function get_status() { |
| 166 |
$state = get_option( self::STATE_OPTION, [] ); |
| 167 |
$queue = get_option( self::QUEUE_OPTION, [] ); |
| 168 |
|
| 169 |
$defaults = [ |
| 170 |
'status' => 'idle', |
| 171 |
'total' => 0, |
| 172 |
'processed' => 0, |
| 173 |
'started' => 0, |
| 174 |
'finished' => 0, |
| 175 |
]; |
| 176 |
$state = array_merge( $defaults, is_array( $state ) ? $state : [] ); |
| 177 |
|
| 178 |
$state['remaining'] = is_array( $queue ) ? count( $queue ) : 0; |
| 179 |
$state['enabled'] = ! empty( $this->settings->enable_preload ); |
| 180 |
|
| 181 |
return $state; |
| 182 |
} |
| 183 |
|
| 184 |
/** |
| 185 |
* Add a URL to the queue if it's not yet been processed. |
| 186 |
* |
| 187 |
* @param string $url |
| 188 |
*/ |
| 189 |
public function enqueue_url( $url ) { |
| 190 |
$url = $this->normalize_url( $url ); |
| 191 |
if ( ! $url ) { |
| 192 |
return; |
| 193 |
} |
| 194 |
|
| 195 |
$queue = get_option( self::QUEUE_OPTION, [] ); |
| 196 |
if ( ! is_array( $queue ) ) { |
| 197 |
$queue = []; |
| 198 |
} |
| 199 |
$processed = get_option( self::PROCESSED_OPTION, [] ); |
| 200 |
if ( ! is_array( $processed ) ) { |
| 201 |
$processed = []; |
| 202 |
} |
| 203 |
|
| 204 |
if ( in_array( $url, $queue, true ) || in_array( $url, $processed, true ) ) { |
| 205 |
return; |
| 206 |
} |
| 207 |
|
| 208 |
$queue[] = $url; |
| 209 |
update_option( self::QUEUE_OPTION, array_values( $queue ), false ); |
| 210 |
|
| 211 |
$state = get_option( self::STATE_OPTION, [] ); |
| 212 |
if ( ! is_array( $state ) ) { |
| 213 |
$state = []; |
| 214 |
} |
| 215 |
$state['total'] = ( isset( $state['total'] ) ? (int) $state['total'] : 0 ) + 1; |
| 216 |
$state['status'] = 'running'; |
| 217 |
update_option( self::STATE_OPTION, $state, false ); |
| 218 |
|
| 219 |
$this->maybe_schedule_cron(); |
| 220 |
} |
| 221 |
|
| 222 |
/** |
| 223 |
* Convenience method to enqueue a single post permalink. |
| 224 |
* |
| 225 |
* @param int $post_id |
| 226 |
*/ |
| 227 |
public function enqueue_post_url( $post_id ) { |
| 228 |
if ( empty( $this->settings->enable_preload ) ) { |
| 229 |
return; |
| 230 |
} |
| 231 |
|
| 232 |
$url = get_permalink( $post_id ); |
| 233 |
if ( $url ) { |
| 234 |
$this->enqueue_url( $url ); |
| 235 |
} |
| 236 |
} |
| 237 |
|
| 238 |
/** |
| 239 |
* Process the next batch of URLs from the queue. |
| 240 |
*/ |
| 241 |
public function process_queue() { |
| 242 |
$queue = get_option( self::QUEUE_OPTION, [] ); |
| 243 |
if ( ! is_array( $queue ) || empty( $queue ) ) { |
| 244 |
$state = get_option( self::STATE_OPTION, [] ); |
| 245 |
if ( is_array( $state ) && isset( $state['status'] ) && 'running' === $state['status'] ) { |
| 246 |
$state['status'] = 'completed'; |
| 247 |
$state['finished'] = time(); |
| 248 |
update_option( self::STATE_OPTION, $state, false ); |
| 249 |
} |
| 250 |
return; |
| 251 |
} |
| 252 |
|
| 253 |
$batch_size = isset( $this->settings->preload_batch_size ) ? max( 1, (int) $this->settings->preload_batch_size ) : 5; |
| 254 |
$batch = array_splice( $queue, 0, $batch_size ); |
| 255 |
|
| 256 |
// Save updated queue first so a long fetch doesn't reprocess the same URL. |
| 257 |
update_option( self::QUEUE_OPTION, array_values( $queue ), false ); |
| 258 |
|
| 259 |
$processed = get_option( self::PROCESSED_OPTION, [] ); |
| 260 |
if ( ! is_array( $processed ) ) { |
| 261 |
$processed = []; |
| 262 |
} |
| 263 |
|
| 264 |
foreach ( $batch as $url ) { |
| 265 |
$this->fetch_url( $url ); |
| 266 |
$processed[] = $url; |
| 267 |
} |
| 268 |
|
| 269 |
// Cap processed list to prevent it from growing without bound on huge sites. |
| 270 |
$cap = (int) apply_filters( 'ezcache_preload_processed_cap', 50000 ); |
| 271 |
if ( count( $processed ) > $cap ) { |
| 272 |
$processed = array_slice( $processed, -1 * $cap ); |
| 273 |
} |
| 274 |
|
| 275 |
update_option( self::PROCESSED_OPTION, $processed, false ); |
| 276 |
|
| 277 |
$state = get_option( self::STATE_OPTION, [] ); |
| 278 |
if ( ! is_array( $state ) ) { |
| 279 |
$state = []; |
| 280 |
} |
| 281 |
$state['processed'] = ( isset( $state['processed'] ) ? (int) $state['processed'] : 0 ) + count( $batch ); |
| 282 |
$state['status'] = empty( $queue ) ? 'completed' : 'running'; |
| 283 |
if ( 'completed' === $state['status'] ) { |
| 284 |
$state['finished'] = time(); |
| 285 |
} |
| 286 |
update_option( self::STATE_OPTION, $state, false ); |
| 287 |
} |
| 288 |
|
| 289 |
/** |
| 290 |
* Fetch a single URL with both desktop and mobile user agents to populate |
| 291 |
* both cache variants. |
| 292 |
* |
| 293 |
* @param string $url |
| 294 |
*/ |
| 295 |
protected function fetch_url( $url ) { |
| 296 |
$desktop_ua = apply_filters( |
| 297 |
'ezcache_desktop_useragent', |
| 298 |
'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/108.0.0.0 Safari/537.36 (ezCache Preload)' |
| 299 |
); |
| 300 |
$mobile_ua = apply_filters( |
| 301 |
'ezcache_mobile_useragent', |
| 302 |
'Mozilla/5.0 (iPhone; CPU iPhone OS 15_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/15.0 Mobile/15E148 Safari/604.1 (ezCache Preload)' |
| 303 |
); |
| 304 |
|
| 305 |
$args = [ |
| 306 |
'timeout' => 30, |
| 307 |
'redirection' => 3, |
| 308 |
'blocking' => true, |
| 309 |
'sslverify' => apply_filters( 'https_local_ssl_verify', false ), |
| 310 |
'user-agent' => $desktop_ua, |
| 311 |
'headers' => [ |
| 312 |
'X-Ezcache-Preload' => '1', |
| 313 |
'Cache-Control' => 'no-cache', |
| 314 |
], |
| 315 |
]; |
| 316 |
|
| 317 |
wp_safe_remote_get( $url, $args ); |
| 318 |
|
| 319 |
if ( ! empty( $this->settings->separate_mobile_cache ) ) { |
| 320 |
$args['user-agent'] = $mobile_ua; |
| 321 |
wp_safe_remote_get( $url, $args ); |
| 322 |
} |
| 323 |
|
| 324 |
// Optionally extract internal links from the homepage to seed the queue. |
| 325 |
if ( $this->is_homepage( $url ) && ! empty( $this->settings->preload_crawl_homepage_links ) ) { |
| 326 |
$this->seed_from_homepage( $url ); |
| 327 |
} |
| 328 |
|
| 329 |
Logger::log( 'ezCache Preload fetched ' . $url ); |
| 330 |
} |
| 331 |
|
| 332 |
/** |
| 333 |
* Returns true when the URL points to the site homepage. |
| 334 |
* |
| 335 |
* @param string $url |
| 336 |
* @return bool |
| 337 |
*/ |
| 338 |
protected function is_homepage( $url ) { |
| 339 |
return untrailingslashit( $url ) === untrailingslashit( home_url() ); |
| 340 |
} |
| 341 |
|
| 342 |
/** |
| 343 |
* Crawl the homepage HTML and seed the queue with internal links found there. |
| 344 |
* |
| 345 |
* @param string $url |
| 346 |
*/ |
| 347 |
protected function seed_from_homepage( $url ) { |
| 348 |
$response = wp_safe_remote_get( $url, [ |
| 349 |
'timeout' => 30, |
| 350 |
'sslverify' => false, |
| 351 |
'user-agent' => 'ezCache Preload Crawler', |
| 352 |
] ); |
| 353 |
|
| 354 |
if ( is_wp_error( $response ) ) { |
| 355 |
return; |
| 356 |
} |
| 357 |
|
| 358 |
$body = wp_remote_retrieve_body( $response ); |
| 359 |
if ( empty( $body ) ) { |
| 360 |
return; |
| 361 |
} |
| 362 |
|
| 363 |
if ( ! preg_match_all( '#<a\s[^>]*href=["\']([^"\']+)["\']#i', $body, $matches ) ) { |
| 364 |
return; |
| 365 |
} |
| 366 |
|
| 367 |
$home_host = wp_parse_url( home_url(), PHP_URL_HOST ); |
| 368 |
foreach ( $matches[1] as $href ) { |
| 369 |
$href = trim( $href ); |
| 370 |
$href = strtok( $href, '#' ); |
| 371 |
$first = substr( $href, 0, 1 ); |
| 372 |
|
| 373 |
if ( '' === $href || '#' === $first || 0 === strpos( $href, 'mailto:' ) || 0 === strpos( $href, 'tel:' ) || 0 === strpos( $href, 'javascript:' ) ) { |
| 374 |
continue; |
| 375 |
} |
| 376 |
|
| 377 |
// Resolve protocol relative & relative URLs. |
| 378 |
if ( 0 === strpos( $href, '//' ) ) { |
| 379 |
$href = ( is_ssl() ? 'https:' : 'http:' ) . $href; |
| 380 |
} elseif ( '/' === $first ) { |
| 381 |
$href = home_url( $href ); |
| 382 |
} elseif ( ! preg_match( '#^https?://#i', $href ) ) { |
| 383 |
continue; |
| 384 |
} |
| 385 |
|
| 386 |
$host = wp_parse_url( $href, PHP_URL_HOST ); |
| 387 |
if ( $host !== $home_host ) { |
| 388 |
continue; |
| 389 |
} |
| 390 |
|
| 391 |
$this->enqueue_url( $href ); |
| 392 |
} |
| 393 |
} |
| 394 |
|
| 395 |
/** |
| 396 |
* Build the initial list of URLs from sitemaps and the homepage. |
| 397 |
* |
| 398 |
* @return array |
| 399 |
*/ |
| 400 |
protected function collect_initial_urls() { |
| 401 |
$urls = [ home_url( '/' ) ]; |
| 402 |
|
| 403 |
// Try the user provided sitemap first. |
| 404 |
$sitemap_urls = []; |
| 405 |
if ( ! empty( $this->settings->preload_sitemap_url ) ) { |
| 406 |
$sitemap_urls[] = trim( $this->settings->preload_sitemap_url ); |
| 407 |
} |
| 408 |
|
| 409 |
// Auto-detect common sitemap locations. |
| 410 |
$sitemap_urls = array_merge( $sitemap_urls, [ |
| 411 |
home_url( '/wp-sitemap.xml' ), |
| 412 |
home_url( '/sitemap_index.xml' ), |
| 413 |
home_url( '/sitemap.xml' ), |
| 414 |
] ); |
| 415 |
|
| 416 |
$found = false; |
| 417 |
foreach ( $sitemap_urls as $sitemap_url ) { |
| 418 |
$found_urls = $this->parse_sitemap( $sitemap_url ); |
| 419 |
if ( ! empty( $found_urls ) ) { |
| 420 |
$urls = array_merge( $urls, $found_urls ); |
| 421 |
$found = true; |
| 422 |
break; |
| 423 |
} |
| 424 |
} |
| 425 |
|
| 426 |
if ( ! $found ) { |
| 427 |
// Fallback: use recently published posts and pages. |
| 428 |
$query = new \WP_Query( [ |
| 429 |
'post_type' => [ 'post', 'page' ], |
| 430 |
'post_status' => 'publish', |
| 431 |
'posts_per_page' => 200, |
| 432 |
'orderby' => 'modified', |
| 433 |
'order' => 'DESC', |
| 434 |
'no_found_rows' => true, |
| 435 |
'fields' => 'ids', |
| 436 |
] ); |
| 437 |
|
| 438 |
foreach ( $query->posts as $post_id ) { |
| 439 |
$urls[] = get_permalink( $post_id ); |
| 440 |
} |
| 441 |
} |
| 442 |
|
| 443 |
return $urls; |
| 444 |
} |
| 445 |
|
| 446 |
/** |
| 447 |
* Parse an XML sitemap (recursively) and return all `<loc>` URLs. |
| 448 |
* |
| 449 |
* @param string $sitemap_url |
| 450 |
* @param int $depth |
| 451 |
* @return array |
| 452 |
*/ |
| 453 |
protected function parse_sitemap( $sitemap_url, $depth = 0 ) { |
| 454 |
if ( $depth > 3 ) { |
| 455 |
return []; |
| 456 |
} |
| 457 |
|
| 458 |
$response = wp_safe_remote_get( $sitemap_url, [ |
| 459 |
'timeout' => 30, |
| 460 |
'sslverify' => false, |
| 461 |
'user-agent' => 'ezCache Preload', |
| 462 |
] ); |
| 463 |
|
| 464 |
if ( is_wp_error( $response ) || 200 !== (int) wp_remote_retrieve_response_code( $response ) ) { |
| 465 |
return []; |
| 466 |
} |
| 467 |
|
| 468 |
$body = wp_remote_retrieve_body( $response ); |
| 469 |
if ( empty( $body ) ) { |
| 470 |
return []; |
| 471 |
} |
| 472 |
|
| 473 |
libxml_use_internal_errors( true ); |
| 474 |
$xml = simplexml_load_string( $body ); |
| 475 |
if ( false === $xml ) { |
| 476 |
return []; |
| 477 |
} |
| 478 |
|
| 479 |
$urls = []; |
| 480 |
|
| 481 |
// Sitemap index: contains <sitemap><loc>...</loc></sitemap> entries. |
| 482 |
if ( isset( $xml->sitemap ) ) { |
| 483 |
foreach ( $xml->sitemap as $entry ) { |
| 484 |
$loc = (string) $entry->loc; |
| 485 |
if ( $loc ) { |
| 486 |
$urls = array_merge( $urls, $this->parse_sitemap( $loc, $depth + 1 ) ); |
| 487 |
} |
| 488 |
} |
| 489 |
} |
| 490 |
|
| 491 |
// Regular sitemap: contains <url><loc>...</loc></url> entries. |
| 492 |
if ( isset( $xml->url ) ) { |
| 493 |
foreach ( $xml->url as $entry ) { |
| 494 |
$loc = (string) $entry->loc; |
| 495 |
if ( $loc ) { |
| 496 |
$urls[] = $loc; |
| 497 |
} |
| 498 |
} |
| 499 |
} |
| 500 |
|
| 501 |
return $urls; |
| 502 |
} |
| 503 |
|
| 504 |
/** |
| 505 |
* Normalize a URL: trim, ensure same host, drop fragments, dedupe. |
| 506 |
* |
| 507 |
* @param string $url |
| 508 |
* @return string|false |
| 509 |
*/ |
| 510 |
protected function normalize_url( $url ) { |
| 511 |
$url = trim( (string) $url ); |
| 512 |
if ( empty( $url ) ) { |
| 513 |
return false; |
| 514 |
} |
| 515 |
|
| 516 |
$url = strtok( $url, '#' ); |
| 517 |
|
| 518 |
if ( ! preg_match( '#^https?://#i', $url ) ) { |
| 519 |
return false; |
| 520 |
} |
| 521 |
|
| 522 |
$home_host = wp_parse_url( home_url(), PHP_URL_HOST ); |
| 523 |
$url_host = wp_parse_url( $url, PHP_URL_HOST ); |
| 524 |
if ( ! $url_host || $url_host !== $home_host ) { |
| 525 |
return false; |
| 526 |
} |
| 527 |
|
| 528 |
// Excluded URIs from settings: don't waste cycles preloading them. |
| 529 |
$settings = $this->settings; |
| 530 |
if ( ! empty( $settings->rejected_uri ) ) { |
| 531 |
$rejected = preg_split( "/\\r\\n|\\r|\\n/u", trim( $settings->rejected_uri ), -1, PREG_SPLIT_NO_EMPTY ); |
| 532 |
$path = wp_parse_url( $url, PHP_URL_PATH ); |
| 533 |
foreach ( $rejected as $pattern ) { |
| 534 |
$pattern = trim( $pattern ); |
| 535 |
if ( '' === $pattern ) { |
| 536 |
continue; |
| 537 |
} |
| 538 |
$regex = str_replace( '\*', '.*', preg_quote( $pattern, '#' ) ); |
| 539 |
if ( @preg_match( '#^' . $regex . '/?$#u', $path ) ) { |
| 540 |
return false; |
| 541 |
} |
| 542 |
} |
| 543 |
} |
| 544 |
|
| 545 |
return $url; |
| 546 |
} |
| 547 |
|
| 548 |
/** |
| 549 |
* Normalize and dedupe an array of URLs. |
| 550 |
* |
| 551 |
* @param array $urls |
| 552 |
* @return array |
| 553 |
*/ |
| 554 |
protected function normalize_urls( $urls ) { |
| 555 |
$out = []; |
| 556 |
foreach ( (array) $urls as $url ) { |
| 557 |
$url = $this->normalize_url( $url ); |
| 558 |
if ( $url && ! isset( $out[ $url ] ) ) { |
| 559 |
$out[ $url ] = true; |
| 560 |
} |
| 561 |
} |
| 562 |
|
| 563 |
return array_keys( $out ); |
| 564 |
} |
| 565 |
} |
| 566 |
|
