trove/tests/Unit/Services/UrlServiceTest.php

<?php

declare(strict_types=1);

namespace Tests\Unit\Services;

use App\Services\UrlService;
use PHPUnit\Framework\Attributes\DataProvider;
use Tests\TestCase;

class UrlServiceTest extends TestCase
{
    private UrlService $service;

    protected function setUp(): void
    {
        parent::setUp();

        $this->service = new UrlService;
    }

    // -------------------------------------------------------------------------
    // Happy path — simple URL
    // -------------------------------------------------------------------------

    public function test_extracts_host_from_simple_url(): void
    {
        $this->assertSame('example.com', $this->service->host('https://example.com'));
    }

    // -------------------------------------------------------------------------
    // Path, query string, and fragment are ignored
    // -------------------------------------------------------------------------

    #[DataProvider('urlsWithNoise')]
    public function test_extracts_host_ignoring_path_query_and_fragment(string $url, string $expectedHost): void
    {
        $this->assertSame($expectedHost, $this->service->host($url));
    }

    public static function urlsWithNoise(): array
    {
        return [
            'path only' => ['https://example.com/some/path', 'example.com'],
            'path and query' => ['https://example.com/page?q=hello&lang=en', 'example.com'],
            'path, query, fragment' => ['https://example.com/page?q=1#section', 'example.com'],
            'http scheme with path' => ['http://news.ycombinator.com/item?id=42', 'news.ycombinator.com'],
        ];
    }

    // -------------------------------------------------------------------------
    // Port number is stripped from the host
    // -------------------------------------------------------------------------

    public function test_strips_port_from_host(): void
    {
        $this->assertSame('example.com', $this->service->host('https://example.com:8080/path'));
    }

    // -------------------------------------------------------------------------
    // Host is always returned as lowercase
    // -------------------------------------------------------------------------

    public function test_lowercases_host(): void
    {
        $this->assertSame('example.com', $this->service->host('https://EXAMPLE.COM/path'));
    }

    // -------------------------------------------------------------------------
    // Throws on malformed, disallowed, or IP-literal input
    // -------------------------------------------------------------------------

    #[DataProvider('invalidInputs')]
    public function test_throws_on_invalid_input(string $url): void
    {
        $this->expectException(\InvalidArgumentException::class);

        $this->service->host($url);
    }

    public static function invalidInputs(): array
    {
        return [
            // malformed / missing structure
            'empty string' => [''],
            'no scheme' => ['example.com/path'],
            'scheme only' => ['https://'],
            'bare string' => ['not a url at all'],

            // disallowed schemes
            'javascript scheme' => ['javascript:alert(1)'],
            'ftp scheme' => ['ftp://example.com'],
            'data scheme' => ['data:text/html,<h1>hi</h1>'],

            // IP literals — not valid page-URL hosts for Trove's purposes
            'ipv4 literal' => ['https://192.168.1.1/path'],
            'ipv6 literal' => ['https://[::1]/path'],
            'ipv4 without path' => ['http://10.0.0.1'],
        ];
    }
}