trove/app/Enums/CrawlOutcomeEnum.php

<?php

declare(strict_types=1);

namespace App\Enums;

enum CrawlOutcomeEnum: string
{
    case Success = 'success';
    case Failed = 'failed';
    case Timeout = 'timeout';
    case BlockedRobots = 'blocked_robots';
    case Blocked4xx = 'blocked_4xx';
    case Blocked5xx = 'blocked_5xx';

    /**
     * The HTTP fetch succeeded (2xx) but the response is unindexable in v0.1
     * (non-HTML Content-Type). Worker MUST also write `pages.status = Rejected`
     * on this outcome — do NOT treat as Failed. Page row STAYS in the DB to
     * prevent re-discovery loops as fediverse re-shares the URL.
     */
    case Rejected = 'rejected';

    /**
     * The PageStatusEnum value the parent `pages` row should land on for this outcome.
     */
    public function toPageStatus(): PageStatusEnum
    {
        return match ($this) {
            self::Success => PageStatusEnum::Fetched,
            self::Rejected => PageStatusEnum::Rejected,
            self::Failed,
            self::Timeout,
            self::BlockedRobots,
            self::Blocked4xx,
            self::Blocked5xx => PageStatusEnum::Failed,
        };
    }

    /**
     * True if the worker should retry this outcome (transient failures only).
     * Permanent failures (4xx, robots block, rejected content type) and successes do not retry.
     */
    public function isRetryable(): bool
    {
        return match ($this) {
            self::Failed, self::Timeout, self::Blocked5xx => true,
            self::Success, self::Rejected, self::BlockedRobots, self::Blocked4xx => false,
        };
    }

    /**
     * True if the worker should register the outbound links discovered during the fetch.
     * Only Success outcomes have meaningful links; everything else either failed or returned no usable HTML.
     */
    public function shouldRegisterOutboundLinks(): bool
    {
        return $this === self::Success;
    }
}