sirius-press/plugins/sirius-press-sia-export/includes/class-spe-runner.php

<?php
/**
 * Draining the queue.
 *
 * One row at a time: fetch the source URL, rewrite its links, compare the
 * result against what is already in the bucket, and upload only if it
 * differs. The hash comparison is not an optimisation detail — a WordPress
 * page re-renders byte-identically far more often than it changes, and
 * without it every post edit would push the home page, every archive and
 * every asset they reference back over the wire.
 *
 * Batches are small and bounded by wall-clock time rather than count, because
 * the thing that kills a cron-driven exporter is a run that takes longer than
 * PHP's `max_execution_time` and dies halfway, leaving rows claimed and
 * nothing to show. Finishing early and picking up next time is always
 * correct here; the queue is the state.
 *
 * @package SiriusPress
 */

defined( 'ABSPATH' ) || exit;

final class SPE_Runner {

	const CRON_HOOK = 'sirius_press_export_tick';

	/** Stop a batch after this long, whatever is left in the queue. */
	const TIME_BUDGET = 20;

	public static function hooks() {
		add_action( self::CRON_HOOK, array( __CLASS__, 'run_batch' ) );
		add_filter( 'cron_schedules', array( __CLASS__, 'add_schedule' ) );
	}

	public static function add_schedule( $schedules ) {
		$schedules['sirius_minute'] = array(
			'interval' => 60,
			'display'  => __( 'Every minute (Sirius Press export)', 'sirius-press' ),
		);
		return $schedules;
	}

	public static function schedule() {
		if ( ! wp_next_scheduled( self::CRON_HOOK ) ) {
			wp_schedule_event( time() + 60, 'sirius_minute', self::CRON_HOOK );
		}
	}

	public static function unschedule() {
		$timestamp = wp_next_scheduled( self::CRON_HOOK );
		if ( $timestamp ) {
			wp_unschedule_event( $timestamp, self::CRON_HOOK );
		}
	}

	/**
	 * Process a batch.
	 *
	 * @param int $limit Maximum rows to attempt.
	 * @return array{done:int,failed:int,skipped:int,left:int}
	 */
	public static function run_batch( $limit = 10 ) {
		$result = array(
			'done'    => 0,
			'failed'  => 0,
			'skipped' => 0,
			'left'    => 0,
		);

		if ( SP_Settings::MODE_SERVER !== SP_Settings::mode() ) {
			// Manual mode: the server has no key, so there is nothing for cron
			// to do. Rows wait for a browser to sign them.
			$result['left'] = SPE_Queue::counts()[ SPE_Queue::STATUS_PENDING ];
			return $result;
		}
		if ( ! SP_Settings::is_configured() ) {
			return $result;
		}

		$deadline = microtime( true ) + self::TIME_BUDGET;
		foreach ( SPE_Queue::claim( $limit ) as $row ) {
			if ( microtime( true ) > $deadline ) {
				break;
			}
			$outcome = self::process( $row );
			if ( 'done' === $outcome ) {
				$result['done']++;
			} elseif ( 'skipped' === $outcome ) {
				$result['skipped']++;
			} else {
				$result['failed']++;
			}
		}

		$result['left'] = SPE_Queue::counts()[ SPE_Queue::STATUS_PENDING ];
		return $result;
	}

	/**
	 * One row.
	 *
	 * @return string 'done', 'skipped' or 'failed'.
	 */
	public static function process( $row ) {
		$body = self::build( $row );
		if ( is_wp_error( $body ) ) {
			SPE_Queue::mark_failed( $row->id, $body->get_error_message() );
			return 'failed';
		}

		$hash = hash( 'sha256', $body );
		if ( '' !== $row->content_hash && hash_equals( $row->content_hash, $hash ) ) {
			// Byte-identical to what is already published.
			SPE_Queue::mark_done( $row->id, $hash );
			return 'skipped';
		}

		$upload = SP_Gateway::put( $row->path, $body );
		if ( empty( $upload['ok'] ) ) {
			SPE_Queue::mark_failed( $row->id, $upload['error'] );
			self::report_failure( $row, $upload['error'] );
			return 'failed';
		}

		SPE_Queue::mark_done( $row->id, $hash );
		/**
		 * Fires after a path is published to the name's bucket.
		 *
		 * @param string $path
		 * @param int    $bytes
		 */
		do_action( 'sirius_press_exported', $row->path, strlen( $body ) );
		return 'done';
	}

	/**
	 * Produce the bytes for a row.
	 *
	 * Assets are fetched and passed through untouched. Pages are fetched,
	 * rewritten, and the assets they mention are queued behind them — which
	 * is how a bulk export discovers the theme's stylesheet without anybody
	 * having to list it.
	 *
	 * @return string|WP_Error
	 */
	public static function build( $row ) {
		$fetched = SPE_Renderer::fetch( $row->source_url );
		if ( empty( $fetched['ok'] ) ) {
			return new WP_Error( 'sirius_export_fetch', $fetched['error'] );
		}

		if ( 'asset' === $row->kind || SPE_Mapper::is_asset( $row->path ) ) {
			return $fetched['body'];
		}

		$rewritten = SPE_Renderer::rewrite( $fetched['body'], $row->path );
		foreach ( $rewritten['assets'] as $path => $url ) {
			// Only queue an asset the first time it is seen; re-queueing it on
			// every page that links the stylesheet would reset its row to
			// pending forever and the queue would never empty.
			if ( '' === SPE_Queue::hash_at( $path ) ) {
				SPE_Queue::add( $path, $url, 'asset' );
			}
		}
		return $rewritten['html'];
	}

	/**
	 * Tell somebody when an export stops working.
	 *
	 * Only on the attempt that exhausts the retries, and only once per hour:
	 * a misconfigured key fails on every row in the queue, and an inbox with
	 * four hundred copies of the same message is indistinguishable from no
	 * message at all.
	 */
	private static function report_failure( $row, $error ) {
		if ( (int) $row->attempts + 1 < SPE_Queue::MAX_ATTEMPTS ) {
			return;
		}
		if ( get_transient( 'sirius_export_failure_notified' ) ) {
			return;
		}
		set_transient( 'sirius_export_failure_notified', 1, HOUR_IN_SECONDS );

		SP_Inbox::add(
			0,
			__( 'A page could not be published to your name', 'sirius-press' ),
			sprintf(
				/* translators: 1: bucket path, 2: error message, 3: URL of the export screen. */
				wp_kses_post( __( '<code>%1$s</code> failed to upload: %2$s<br><br>Other pages may be failing for the same reason. <a href="%3$s">Open the export screen</a> to see the queue.', 'sirius-press' ) ),
				esc_html( $row->path ),
				esc_html( $error ),
				esc_url( admin_url( 'admin.php?page=sirius-press-export' ) )
			),
			'export'
		);
	}

	/** Queue every exportable page on the site. */
	public static function queue_full_site() {
		$queued = 0;
		foreach ( SPE_Mapper::full_site() as $path => $target ) {
			if ( SPE_Queue::add( $path, $target['url'], $target['kind'] ) ) {
				$queued++;
			}
		}
		return $queued;
	}
}