2026-04-26 13:06:22 +02:00
|
|
|
<?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';
|
2026-04-26 19:49:08 +02:00
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* 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.
|
|
|
|
|
*/
|
2026-04-26 16:35:46 +02:00
|
|
|
case Rejected = 'rejected';
|
2026-04-26 13:06:22 +02:00
|
|
|
}
|