trove/tests/Unit/Services/PolitenessServiceTest.php

<?php

declare(strict_types=1);

namespace Tests\Unit\Services;

use App\Services\PolitenessService;
use Illuminate\Support\Facades\Http;
use Tests\TestCase;

class PolitenessServiceTest extends TestCase
{
    public function test_min_delay_for_returns_config_default(): void
    {
        $this->assertSame(10, (new PolitenessService)->minDelayFor('example.com'));
    }

    public function test_min_delay_for_respects_config_override(): void
    {
        config()->set('crawler.min_domain_delay_seconds', 30);

        $this->assertSame(30, (new PolitenessService)->minDelayFor('example.com'));
    }

    public function test_min_delay_for_uses_robots_crawl_delay_when_higher(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response(
                // Spatie does exact-token matching (lowercased), so the fixture UA
                // must match the full string the service passes to crawlDelayFor().
                "User-agent: TroveBot/0.1 (+https://trove.lvl0.xyz/bot)\nCrawl-delay: 30",
                200,
            ),
        ]);

        config()->set('crawler.min_domain_delay_seconds', 10);
        config()->set('crawler.user_agent', 'TroveBot/0.1 (+https://trove.lvl0.xyz/bot)');

        $this->assertSame(30, app(PolitenessService::class)->minDelayFor('example.com'));
    }

    public function test_min_delay_for_uses_config_when_higher_than_robots(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response(
                "User-agent: TroveBot/0.1 (+https://trove.lvl0.xyz/bot)\nCrawl-delay: 10",
                200,
            ),
        ]);

        config()->set('crawler.min_domain_delay_seconds', 60);
        config()->set('crawler.user_agent', 'TroveBot/0.1 (+https://trove.lvl0.xyz/bot)');

        $this->assertSame(60, app(PolitenessService::class)->minDelayFor('example.com'));
    }
}