SeoCheckerController.php

<?php

namespace App\Http\Controllers;

use Illuminate\Http\Request;
use Inertia\Inertia;
use Symfony\Component\DomCrawler\Crawler;
use Illuminate\Support\Facades\Http;

class SeoCheckerController extends Controller
{

    public function index(Request $request)
    {
        return Inertia::render('seo');
    }

    /**
     * Analiza una URL y devuelve un puntaje SEO
     *
     * @param Request $request
     */
    public function analyzeUrl(Request $request)
    {
        $validated =  $request->validate([
            'url' => ['required', 'url'],
        ]);

        $baseUrl = parse_url($validated['url'], PHP_URL_SCHEME) . '://' . parse_url($validated['url'], PHP_URL_HOST);
        $visited = [];
        $results = [];

        $this->crawl($validated['url'], $baseUrl, $visited, $results);

        $returnData = $this->calculateSeoScore($results);

        return Inertia::render('seo', [
            'item' => $returnData
        ]);
    }

    /**
     * Realiza un rastreo de la página web
     *
     * @param string $url URL de la página a rastrear
     * @param string $baseUrl URL base de la página
     * @param array $visited URLs visitadas
     * @param array $results Resultados del rastreo
     * @param int $depth Profundidad actual
     * @param int $maxDepth Profundidad máxima
     * @return void
     */
    private function crawl($url, $baseUrl, &$visited, &$results, $depth = 0, $maxDepth = 10): void
    {
        $normalizedUrl = $this->normalizeUrl($url);

        if (isset($visited[$normalizedUrl]) || $depth > $maxDepth) {
            return;
        }

        $visited[$normalizedUrl] = true;

        $response = Http::get($url);
        if (!$response->successful()) {
            return;
        }

        $crawler = new Crawler($response->body(), $url);

        $has_title = $crawler->filter('title')->count() ? true :  false;
        $has_description = $crawler->filter('meta[name="description"]')->count()
            ? true
            : false;

        $title = $crawler->filter('title')->count() ? $crawler->filter('title')->text() : '';
        $metaDescription = $crawler->filter('meta[name="description"]')->count()
            ? $crawler->filter('meta[name="description"]')->attr('content')
            : '';
        $h1Text = $crawler->filter('h1')->count() ? $crawler->filter('h1')->text() : 'No H1 found';
        $has_h1 = $crawler->filter('h1')->count() ? true : false;

        $results[] = [
            'url' => $normalizedUrl,
            'title' => $title,
            'meta_description' => $metaDescription,
            'has_description' =>  $has_description,
            'has_title' =>  $has_title,
            'h1' => $h1Text,
            'has_h1' => $has_h1
        ];

        $links = $crawler->filter('a')->links();
        foreach ($links as $link) {
            $href = $link->getUri();

            if (
                strpos($href, $baseUrl) === 0 &&
                strpos($href, '#') === false &&
                !preg_match('/\.(pdf|docx?|xlsx?|pptx?|zip|rar|png|jpg|jpeg|webp)$/i', $href)
            ) {
                $this->crawl($href, $baseUrl, $visited, $results, $depth + 1, $maxDepth);
            }
        }
    }

    /**
     * Normaliza una URL para evitar duplicados
     *
     * @param string $url URL a normalizar
     * @return string URL normalizada
     *
     */
    private function normalizeUrl($url): string
    {
        $parsedUrl = parse_url($url);

        $scheme = $parsedUrl['scheme'] ?? 'http';
        $host = $parsedUrl['host'] ?? '';
        $path = $parsedUrl['path'] ?? '/';

        $path = ($path === '/') ? '/' : rtrim($path, '/');

        $normalizedUrl = "{$scheme}://{$host}{$path}";

        if (isset($parsedUrl['query'])) {
            $normalizedUrl .= "?{$parsedUrl['query']}";
        }

        return $normalizedUrl;
    }


    /**
     * Calcula el puntaje SEO de las páginas
     *
     * @param array $data Datos de las páginas
     * @return array Datos de las páginas con puntaje SEO
     *
     */
    private function calculateSeoScore(array $data): array
    {
        $totalScore = 0;
        $pageCount = count($data);

        $data = array_map(function ($page) use (&$totalScore) {
            $score = 100;

            if (!$page['has_title']) {
                $score -= 40;
            }

            if (!$page['has_description']) {
                $score -= 20;
            }

            if (!$page['has_h1']) {
                $score -= 30;
            }

            $score = max(0, $score);

            $totalScore += $score;

            return array_merge($page, ['score' => $score]);
        }, $data);

        $averageScore = $pageCount > 0 ? $totalScore / $pageCount : 0;

        return ['pages' => $data, 'average_score' => $averageScore];
    }
}