KantSites/src/Controller/SeoController.php

<?php

namespace App\Controller;

use App\Content\Languages;
use App\Content\PostRepository;
use App\Content\SiteSettings;
use Doctrine\DBAL\Connection;
use Symfony\Bundle\FrameworkBundle\Controller\AbstractController;
use Symfony\Component\HttpFoundation\Response;
use Symfony\Component\Routing\Attribute\Route;

/**
 * The machine readable surface of a site: sitemap, robots and the optional llms.txt. Everything
 * here is derived from published content, never maintained separately, so it cannot drift apart
 * from what a visitor actually sees.
 */
final class SeoController extends AbstractController
{
    public function __construct(
        private Connection $db,
        private SiteSettings $settings,
        private PostRepository $posts,
        private string $siteOrigin,
        private string $adminPath,
        private Languages $languages,
    ) {
    }

    #[Route('/sitemap.xml', name: 'sitemap', methods: ['GET'])]
    public function sitemap(): Response
    {
        $origin = rtrim($this->siteOrigin, '/');
        $home = $this->settings->homePath();
        $entries = [];
        // Only published pages that are allowed to be indexed. A draft or a noindex page in the
        // sitemap would ask a search engine to fetch something it must not list.
        foreach ($this->db->fetchAllAssociative(
            "SELECT p.path, p.type, p.published_at, r.created_at FROM page p JOIN revision r ON r.id = p.published_revision_id AND r.page_id = p.id WHERE p.website_id = ? AND p.robots = 'index' AND p.deleted_at IS NULL AND p.id <> ? ORDER BY p.path",
            ['default', $this->settings->load()['notFoundPageId']],
        ) as $row) {
            $entries[] = [
                'loc' => $origin.$this->settings->publicPath($row['path'], $home),
                // Das juengere der beiden Daten: eine spaeter bearbeitete und neu veroeffentlichte
                // Seite behielte sonst das Datum ihrer ersten Veroeffentlichung.
                'lastmod' => substr(max((string) ($row['published_at'] ?? ''), (string) ($row['created_at'] ?? '')), 0, 10),
            ];
        }
        // Jede weitere Sprache mit ihren veröffentlichten Seiten (ADR 111), unter denselben
        // Bedingungen wie die Hauptsprache. Die Zuordnung der Sprachen zueinander steht als
        // hreflang in den Seiten selbst.
        $extra = $this->languages->extra();
        if ([] !== $extra) {
            foreach ($this->db->fetchAllAssociative(
                "SELECT t.locale, t.path AS translated, p.path, p.published_at, r.created_at FROM page_translation t JOIN page p ON p.id = t.page_id JOIN revision r ON r.id = p.published_revision_id AND r.page_id = p.id WHERE t.website_id = ? AND p.robots = 'index' AND p.deleted_at IS NULL AND p.id <> ? ORDER BY t.locale, t.path",
                ['default', $this->settings->load()['notFoundPageId']],
            ) as $row) {
                if (!in_array($row['locale'], $extra, true)) {
                    continue;
                }
                $entries[] = [
                    'loc' => $origin.'/'.$row['locale'].($row['path'] === $home ? '' : '/'.$row['translated']),
                    'lastmod' => substr(max((string) ($row['published_at'] ?? ''), (string) ($row['created_at'] ?? '')), 0, 10),
                ];
            }
        }
        // Archives exist only while they list something.
        if ([] !== $this->posts->published(1, null, 1)['posts']) {
            $entries[] = ['loc' => $origin.'/beitraege', 'lastmod' => null];
        }
        foreach ($this->db->fetchAllAssociative(
            'SELECT DISTINCT t.kind, t.slug FROM term t JOIN page_term pt ON pt.term_id = t.id JOIN page p ON p.id = pt.page_id AND p.published_revision_id IS NOT NULL AND p.deleted_at IS NULL ORDER BY t.kind, t.slug',
        ) as $row) {
            $entries[] = ['loc' => $origin.'/'.PostRepository::ARCHIVE_PREFIX[$row['kind']].'/'.$row['slug'], 'lastmod' => null];
        }

        return $this->xml($this->renderView('seo/sitemap.xml.twig', ['entries' => $entries]));
    }

    #[Route('/robots.txt', name: 'robots', methods: ['GET'])]
    public function robots(): Response
    {
        $settings = $this->settings->load();
        // Die Verwaltung wird hier nur genannt, solange sie „admin" heisst - diese Adresse kennt
        // ohnehin jeder Scanner. Wer sie umbenannt hat, will sie nicht in einer oeffentlichen
        // Datei wiederfinden: robots.txt ist das erste, was ein Angreifer liest. Vom Indexieren
        // haelt die Verwaltung ihr eigener Kopf `X-Robots-Tag: noindex` ab.
        $lines = ['User-agent: *'];
        if ('admin' === $this->adminPath) {
            $lines[] = 'Disallow: /admin';
        }
        $lines[] = 'Disallow: /formular/';
        if ('' !== $settings['robotsExtra']) {
            $lines[] = $settings['robotsExtra'];
        }
        $lines[] = '';
        $lines[] = 'Sitemap: '.rtrim($this->siteOrigin, '/').'/sitemap.xml';

        return $this->text(implode("\n", $lines)."\n");
    }

    /**
     * A machine readable summary of public content. Optional and switched off by default. It is
     * no ranking factor and no requirement for any search product; it simply saves an agent the
     * crawl. Only published, indexable pages appear.
     */
    #[Route('/llms.txt', name: 'llms', methods: ['GET'])]
    public function llms(): Response
    {
        $settings = $this->settings->load();
        if (!$settings['llmsTxt']) {
            throw $this->createNotFoundException('llms.txt ist für diese Website nicht aktiviert.');
        }
        $origin = rtrim($this->siteOrigin, '/');
        $home = $settings['homePath'];
        $name = (string) $this->db->fetchOne('SELECT name FROM website WHERE id = ?', ['default']);
        $lines = ['# '.('' !== $settings['legalName'] ? $settings['legalName'] : $name), ''];
        if ('' !== $settings['description']) {
            $lines[] = '> '.$settings['description'];
            $lines[] = '';
        }
        foreach (['page' => '## Seiten', 'post' => '## Beiträge'] as $type => $heading) {
            $rows = $this->db->fetchAllAssociative(
                "SELECT p.path, p.excerpt, r.document FROM page p JOIN revision r ON r.id = p.published_revision_id AND r.page_id = p.id WHERE p.website_id = ? AND p.robots = 'index' AND p.deleted_at IS NULL AND p.type = ? AND p.id <> ? ORDER BY p.path",
                ['default', $type, $settings['notFoundPageId']],
            );
            if ([] === $rows) {
                continue;
            }
            $lines[] = $heading;
            foreach ($rows as $row) {
                $document = json_decode($row['document'], true, 512, JSON_THROW_ON_ERROR);
                $summary = trim((string) ($row['excerpt'] ?? $document['description'] ?? ''));
                $lines[] = sprintf('- [%s](%s)%s', $document['title'] ?? $row['path'], $origin.$this->settings->publicPath($row['path'], $home), '' === $summary ? '' : ': '.$summary);
            }
            $lines[] = '';
        }

        return $this->text(implode("\n", $lines));
    }

    private function text(string $body): Response
    {
        return $this->harden(new Response($body, 200, ['Content-Type' => 'text/plain; charset=UTF-8']));
    }

    private function xml(string $body): Response
    {
        return $this->harden(new Response($body, 200, ['Content-Type' => 'application/xml; charset=UTF-8']));
    }

    private function harden(Response $response): Response
    {
        $response->headers->set('X-Content-Type-Options', 'nosniff');
        $response->headers->set('Cache-Control', 'no-store');

        return $response;
    }
}