trove/tests/Unit/Services/RobotsServiceTest.php

<?php

declare(strict_types=1);

namespace Tests\Unit\Services;

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

class RobotsServiceTest extends TestCase
{
    public function test_is_allowed_returns_true_when_robots_txt_allows_path(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response(
                "User-agent: *\nAllow: /",
                200,
            ),
        ]);

        $service = app(RobotsService::class);

        $this->assertTrue($service->isAllowed('https://example.com/article', 'TroveBot/0.1'));
    }

    public function test_is_allowed_returns_false_when_robots_txt_disallows_path(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response(
                "User-agent: *\nDisallow: /",
                200,
            ),
        ]);

        $service = app(RobotsService::class);

        $this->assertFalse($service->isAllowed('https://example.com/article', 'TroveBot/0.1'));
    }

    public function test_is_allowed_returns_true_when_robots_txt_fetch_fails(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response('', 500),
        ]);

        $service = app(RobotsService::class);

        $this->assertTrue($service->isAllowed('https://example.com/article', 'TroveBot/0.1'));
    }

    public function test_is_allowed_caches_robots_txt_body_per_host(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response(
                "User-agent: *\nAllow: /",
                200,
            ),
        ]);

        $service = app(RobotsService::class);

        $service->isAllowed('https://example.com/article', 'TroveBot/0.1');
        $service->isAllowed('https://example.com/another-article', 'TroveBot/0.1');

        Http::assertSentCount(1);
    }

    public function test_crawl_delay_for_returns_parsed_value(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response(
                "User-agent: TroveBot/0.1\nCrawl-delay: 30",
                200,
            ),
        ]);

        $service = app(RobotsService::class);

        $this->assertSame(30, $service->crawlDelayFor('example.com', 'TroveBot/0.1'));
    }

    public function test_crawl_delay_for_returns_null_when_absent(): void
    {
        Http::fake([
            'https://example.com/robots.txt' => Http::response(
                "User-agent: *\nDisallow: /private",
                200,
            ),
        ]);

        $service = app(RobotsService::class);

        $this->assertNull($service->crawlDelayFor('example.com', 'TroveBot/0.1'));
    }
}