argus-wp-defence/includes/class-argus-cache-warmer.php

<?php

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

class Argus_Cache_Warmer {

	const BACKOFF_OPTION      = 'argus_wpd_cache_warmer_backoff';
	const SLOW_ORIGIN_MS      = 3000;
	const MAX_ATTEMPTS        = 3;
	const REFILL_LIMIT        = 200;

	public static function process_queue() {
		$backoff = get_option( self::BACKOFF_OPTION, array( 'consecutive_slow' => 0, 'next_allowed_ts' => 0 ) );
		if ( time() < (int) ( $backoff['next_allowed_ts'] ?? 0 ) ) {
			return array( 'skipped' => true, 'reason' => 'Backing off -- origin was slow on a recent warming pass.' );
		}

		if ( self::origin_looks_unhealthy() ) {
			self::apply_backoff( $backoff );
			return array( 'skipped' => true, 'reason' => 'Origin average response time is elevated; pausing warming this cycle.' );
		}

		self::refill_queue();

		$batch_size = max( 1, (int) Argus_Settings::get( 'cache_warm_batch_size', 20 ) );
		$batch      = self::claim_batch( $batch_size );
		if ( empty( $batch ) ) {
			return array( 'processed' => 0 );
		}

		$concurrency = max( 1, min( 10, (int) Argus_Settings::get( 'cache_warm_concurrency', 4 ) ) );
		$min_interval = max( 0, (int) Argus_Settings::get( 'cache_warm_min_interval_secs', 1 ) );

		$results = array( 'ok' => 0, 'failed' => 0 );
		foreach ( array_chunk( $batch, $concurrency ) as $chunk ) {
			$chunk_results = self::fetch_concurrently( wp_list_pluck( $chunk, 'url' ) );
			foreach ( $chunk as $item ) {
				$ok = ! empty( $chunk_results[ $item->url ] );
				self::record_result( $item, $ok );
				$ok ? ++$results['ok'] : ++$results['failed'];
			}
			if ( $min_interval > 0 ) {
				sleep( $min_interval );
			}
		}

		if ( $results['failed'] > $results['ok'] ) {
			$backoff['consecutive_slow'] = (int) ( $backoff['consecutive_slow'] ?? 0 ) + 1;
			self::apply_backoff( $backoff );
		} else {
			update_option( self::BACKOFF_OPTION, array( 'consecutive_slow' => 0, 'next_allowed_ts' => 0 ), false );
		}

		return array( 'processed' => count( $batch ) ) + $results;
	}

	protected static function origin_looks_unhealthy() {
		$summary = Argus_Cache_Log::summary( 1 );
		$miss_ms = $summary[ Argus_Static_Cache::STATUS_MISS ]['avg_ms'] ?? null;
		return null !== $miss_ms && $miss_ms > self::SLOW_ORIGIN_MS;
	}

	protected static function apply_backoff( array $backoff ) {
		$consecutive = max( 1, (int) ( $backoff['consecutive_slow'] ?? 1 ) );
		$delay = min( HOUR_IN_SECONDS, 60 * ( 2 ** $consecutive ) );
		update_option(
			self::BACKOFF_OPTION,
			array( 'consecutive_slow' => $consecutive, 'next_allowed_ts' => time() + $delay ),
			false
		);
	}

	protected static function refill_queue() {
		global $wpdb;
		$discovered = Argus_DB::table( 'cache_discovered_urls' );
		$queue      = Argus_DB::table( 'cache_warm_queue' );
		$home       = home_url( '/' );
		$now        = current_time( 'mysql', true );

		$candidates = $wpdb->get_results(
			$wpdb->prepare(
				"SELECT url, lastmod FROM {$discovered}
				WHERE eligibility = %s
				AND ( last_warmed_at IS NULL OR last_warmed_at < %s )
				AND NOT EXISTS ( SELECT 1 FROM {$queue} q WHERE q.url = {$discovered}.url AND q.status = 'queued' )
				ORDER BY (url = %s) DESC, lastmod DESC
				LIMIT %d", // phpcs:ignore
				Argus_Cache_Eligibility::SAFE_TO_CACHE,
				gmdate( 'Y-m-d H:i:s', strtotime( '-1 hour' ) ),
				$home,
				self::REFILL_LIMIT
			)
		);

		foreach ( $candidates as $i => $row ) {
			$wpdb->query(
				$wpdb->prepare(
					"INSERT INTO {$queue} (url, priority, status, queued_at) VALUES (%s, %d, 'queued', %s)", // phpcs:ignore
					$row->url,
					$row->url === $home ? 0 : ( 100 + $i ),
					$now
				)
			);
		}
	}

	protected static function claim_batch( $limit ) {
		global $wpdb;
		$table = Argus_DB::table( 'cache_warm_queue' );

		$rows = $wpdb->get_results(
			$wpdb->prepare( "SELECT * FROM {$table} WHERE status = 'queued' ORDER BY priority ASC, id ASC LIMIT %d", $limit ) // phpcs:ignore
		);
		if ( empty( $rows ) ) {
			return array();
		}

		$ids = wp_list_pluck( $rows, 'id' );
		$wpdb->query( "UPDATE {$table} SET status = 'processing' WHERE id IN (" . implode( ',', array_map( 'absint', $ids ) ) . ')' ); // phpcs:ignore

		return $rows;
	}

	protected static function record_result( $item, $ok ) {
		global $wpdb;
		$queue      = Argus_DB::table( 'cache_warm_queue' );
		$discovered = Argus_DB::table( 'cache_discovered_urls' );
		$now        = current_time( 'mysql', true );

		if ( $ok ) {
			$wpdb->update( $queue, array( 'status' => 'done', 'last_attempt_at' => $now, 'last_result' => 'warmed' ), array( 'id' => $item->id ) );
			$wpdb->update( $discovered, array( 'last_warmed_at' => $now ), array( 'url' => $item->url ) );
			Argus_Cache_Log::record( Argus_Static_Cache::STATUS_WARMED, $item->url, 'Preloaded by cache warmer.' );
			return;
		}

		$attempts = (int) $item->attempts + 1;
		if ( $attempts >= self::MAX_ATTEMPTS ) {
			$wpdb->update( $queue, array( 'status' => 'failed', 'attempts' => $attempts, 'last_attempt_at' => $now, 'last_result' => 'gave up after ' . $attempts . ' attempts' ), array( 'id' => $item->id ) );
		} else {
			$wpdb->update( $queue, array( 'status' => 'queued', 'attempts' => $attempts, 'last_attempt_at' => $now, 'last_result' => 'retrying' ), array( 'id' => $item->id ) );
		}
	}

	protected static function fetch_concurrently( array $urls ) {
		if ( ! function_exists( 'curl_multi_init' ) ) {
			$out = array();
			foreach ( $urls as $url ) {
				$response = wp_remote_get( $url, array( 'timeout' => 20, 'user-agent' => self::user_agent() ) );
				$out[ $url ] = ! is_wp_error( $response ) && wp_remote_retrieve_response_code( $response ) < 500;
			}
			return $out;
		}

		$mh      = curl_multi_init(); // phpcs:ignore WordPress.WP.AlternativeFunctions
		$handles = array();

		foreach ( $urls as $url ) {
			$ch = curl_init( $url ); // phpcs:ignore WordPress.WP.AlternativeFunctions
			curl_setopt_array( // phpcs:ignore WordPress.WP.AlternativeFunctions
				$ch,
				array(
					CURLOPT_RETURNTRANSFER => true,
					CURLOPT_TIMEOUT        => 20,
					CURLOPT_USERAGENT      => self::user_agent(),
					CURLOPT_NOBODY         => false,
					CURLOPT_SSL_VERIFYPEER => true,
				)
			);
			curl_multi_add_handle( $mh, $ch ); // phpcs:ignore WordPress.WP.AlternativeFunctions
			$handles[ $url ] = $ch;
		}

		$running = null;
		do {
			curl_multi_exec( $mh, $running ); // phpcs:ignore WordPress.WP.AlternativeFunctions
			curl_multi_select( $mh ); // phpcs:ignore WordPress.WP.AlternativeFunctions
		} while ( $running > 0 );

		$out = array();
		foreach ( $handles as $url => $ch ) {
			$code = curl_getinfo( $ch, CURLINFO_HTTP_CODE ); // phpcs:ignore WordPress.WP.AlternativeFunctions
			$out[ $url ] = $code > 0 && $code < 500;
			curl_multi_remove_handle( $mh, $ch ); // phpcs:ignore WordPress.WP.AlternativeFunctions
			curl_close( $ch ); // phpcs:ignore WordPress.WP.AlternativeFunctions
		}
		curl_multi_close( $mh ); // phpcs:ignore WordPress.WP.AlternativeFunctions

		return $out;
	}

	protected static function user_agent() {
		return 'ARGUS-Defence-CacheWarmer/1.0 (+' . home_url( '/' ) . ')';
	}

	public static function status() {
		global $wpdb;
		$table = Argus_DB::table( 'cache_warm_queue' );
		$rows  = $wpdb->get_results( "SELECT status, COUNT(*) AS cnt FROM {$table} GROUP BY status" ); // phpcs:ignore

		$out = array( 'queued' => 0, 'processing' => 0, 'done' => 0, 'failed' => 0 );
		foreach ( $rows as $row ) {
			if ( isset( $out[ $row->status ] ) ) {
				$out[ $row->status ] = (int) $row->cnt;
			}
		}
		$backoff = get_option( self::BACKOFF_OPTION, array() );
		$out['paused_until'] = ( ! empty( $backoff['next_allowed_ts'] ) && $backoff['next_allowed_ts'] > time() ) ? $backoff['next_allowed_ts'] : null;
		return $out;
	}
}