toolvana/tests/feature/SeoPagesTest.php

<?php

declare(strict_types=1);

namespace Tests\Feature;

use CodeIgniter\Test\CIUnitTestCase;
use CodeIgniter\Test\FeatureTestTrait;

/**
 * SEO contract tests: every indexable page must render complete,
 * crawlable HTML on the FIRST response — no JS required.
 *
 * These run against the test database (migrated + seeded by setUp).
 */
final class SeoPagesTest extends CIUnitTestCase
{
    use FeatureTestTrait;

    public static function setUpBeforeClass(): void
    {
        parent::setUpBeforeClass();

        // migrate + seed the test database once (idempotent)
        try {
            command('migrate --all');
            command('db:seed InitialSeeder');
        } catch (\Throwable $e) {
            fwrite(STDERR, 'Test DB bootstrap: ' . $e->getMessage() . "\n");
        }
    }

    private function get(string $page): \CodeIgniter\Test\TestResponse
    {
        return $this->call('get', $page);
    }

    public function testToolPageRendersFullSeoContract(): void
    {
        $response = $this->get('/youtube-to-mp3');
        $response->assertStatus(200);

        $html = (string) $response->getBody();

        // canonical, title, description, OG
        $this->assertMatchesRegularExpression('#<link rel="canonical" href="https?://[^"]+/youtube-to-mp3">#', $html);
        $this->assertMatchesRegularExpression('/<title>[^<]+<\/title>/', $html);
        $this->assertStringContainsString('name="description"', $html);
        $this->assertStringContainsString('property="og:title"', $html);
        $this->assertStringContainsString('name="twitter:card" content="summary_large_image"', $html);

        // structured data: FAQPage + BreadcrumbList + WebApplication
        $jsonLd = '';
        if (preg_match('/<script type="application\/ld\+json">(.*?)<\/script>/s', $html, $m)) {
            $jsonLd = $m[1];
        }
        $this->assertNotSame('', $jsonLd, 'JSON-LD block missing');
        foreach (['BreadcrumbList', 'WebApplication', 'FAQPage'] as $type) {
            $this->assertStringContainsString('"' . $type . '"', $jsonLd, "{$type} missing from JSON-LD");
        }

        // content in first response: H1, how-to steps, FAQ questions visible
        $this->assertStringContainsString('<h1>', $html);
        $this->assertStringContainsString('Frequently asked questions', $html);
        $this->assertStringContainsString('Related tools', $html);
        $this->assertStringContainsString('breadcrumbs', $html);

        // robots meta allows indexing
        $this->assertStringContainsString('content="index, follow', $html);
    }

    public function testCategoryAndHomepageAreIndexable(): void
    {
        foreach (['/', '/youtube-tools', '/tools'] as $page) {
            $html = (string) $this->get($page)->getBody();
            $this->assertStringContainsString('rel="canonical"', $html, "canonical missing on {$page}");
            $this->assertStringContainsString('index, follow', $html, "robots meta wrong on {$page}");
        }
    }

    public function testSearchIsNoindexed(): void
    {
        $html = (string) $this->get('/search?q=mp3')->getBody();

        $this->assertStringContainsString('noindex', $html);
    }

    public function testSitemapsContainOnlyCanonicalUrls(): void
    {
        $index  = (string) $this->get('/sitemap.xml')->getBody();
        $this->assertStringContainsString('sitemapindex', $index);
        $this->assertStringContainsString('sitemap-tools-1.xml', $index);
        $this->assertStringContainsString('sitemap-guides.xml', $index);

        $tools = (string) $this->get('/sitemap-tools-1.xml')->getBody();
        $this->assertStringContainsString('/youtube-to-mp3</loc>', $tools);
        $this->assertDoesNotMatchRegularExpression('#<loc>[^<]*/(api/|admin|download/|search\?)#', $tools, 'Non-indexable URL leaked into sitemap');

        $categories = (string) $this->get('/sitemap-categories.xml')->getBody();
        $this->assertStringContainsString('/youtube-tools</loc>', $categories);
    }

    public function testRobotsTxtBlocksProcessingEndpoints(): void
    {
        $body = (string) $this->get('/robots.txt')->getBody();

        $this->assertStringContainsString('Disallow: /api/', $body);
        $this->assertStringContainsString('Disallow: /download/', $body);
        $this->assertStringContainsString('Disallow: /admin', $body);
        $this->assertStringContainsString('Sitemap:', $body);
    }

    public function testUppercaseUrl301RedirectsToCanonical(): void
    {
        $this->get('/YouTube-To-MP3')->assertStatus(301);
    }

    public function testUnknownSlugReturnsUseful404(): void
    {
        $result   = $this->get('/this-tool-does-not-exist');
        $result->assertStatus(404);
        $html = (string) $result->getBody();
        $this->assertStringContainsString('Tool not found', $html);
        $this->assertStringContainsString('Popular tools', $html, '404 page must link popular tools');
    }
}