presseportale/app/Http/Controllers/Web/SitemapController.php

<?php

namespace App\Http\Controllers\Web;

use App\Http\Controllers\Controller;
use App\Models\Category;
use App\Models\Company;
use App\Models\PressRelease;
use App\Scopes\PortalScope;
use App\Services\PressRelease\CanonicalAssigner;
use App\Services\Web\CategoryArchivePageData;
use App\Support\StaticPageCanonical;
use App\Support\WebPortalContext;
use Illuminate\Http\Request;
use Illuminate\Http\Response;
use Illuminate\Support\Facades\Cache;

/**
 * Host-aware XML-Sitemap: Der Hub listet Landing- und Rechts-/Info-Seiten,
 * die Spokes (presseecho, businessportal24) zusätzlich Kategorien, Newsrooms
 * und alle im Archiv sichtbaren Pressemitteilungen ihres Portals. Marketing-
 * Platzhalterseiten (team, presse, karriere, partner, kontakt, ueber-uns)
 * und die Suche bleiben bewusst draußen (R2.4/R3.2). Pro Host eine Stunde
 * gecacht.
 */
class SitemapController extends Controller
{
    /** @var list<string> */
    private const HUB_PAGES = [
        'faq', 'hilfe', 'gute-pressemitteilung', 'api', 'kuendigung',
        'was-kostet-eine-pressemitteilung', 'pressemitteilung-veroeffentlichen-ohne-abo', 'pressemitteilung-loeschen-oder-aendern-lassen', 'pressemitteilung-kostenlos-veroeffentlichen',
        'impressum', 'datenschutz', 'agb', 'richtlinien', 'melden',
    ];

    /** @var list<string> */
    private const PORTAL_PAGES = [
        'preise', 'veroeffentlichen', 'newsrooms', 'faq', 'hilfe', 'gute-pressemitteilung', 'api', 'rss',
        'was-kostet-eine-pressemitteilung', 'pressemitteilung-veroeffentlichen-ohne-abo', 'pressemitteilung-loeschen-oder-aendern-lassen', 'pressemitteilung-kostenlos-veroeffentlichen',
        'impressum', 'datenschutz', 'agb', 'richtlinien', 'melden',
    ];

    public function __invoke(Request $request): Response
    {
        $host = $request->getHost();
        $base = rtrim($request->getSchemeAndHttpHost(), '/');
        $isHub = str_contains($host, 'pressekonto');

        $xml = Cache::remember('web.sitemap.'.$host, 3600, function () use ($base, $isHub): string {
            $urls = $isHub ? $this->hubUrls($base) : $this->portalUrls($base);

            return view('web.sitemap', ['urls' => $urls])->render();
        });

        return response($xml, 200, ['Content-Type' => 'application/xml; charset=UTF-8']);
    }

    /**
     * Ob dieser Host die Seite führt (Anhang 2, E10): gemeinsame
     * Informationsseiten liegen auf allen drei Hosts, in die Sitemap gehört
     * nur die kanonische Adresse. Rechtstexte bleiben je Host gelistet.
     */
    private function leadsPage(string $base, string $page): bool
    {
        $canonicalBase = StaticPageCanonical::baseFor($page);

        return $canonicalBase === null || parse_url($canonicalBase, PHP_URL_HOST) === parse_url($base, PHP_URL_HOST);
    }

    /**
     * @return list<array{loc: string, lastmod: ?string}>
     */
    private function hubUrls(string $base): array
    {
        $urls = [['loc' => $base.'/', 'lastmod' => null]];

        foreach (self::HUB_PAGES as $page) {
            if (! $this->leadsPage($base, $page)) {
                continue;
            }

            $urls[] = ['loc' => $base.'/de/'.$page, 'lastmod' => null];
        }

        return $urls;
    }

    /**
     * @return list<array{loc: string, lastmod: ?string}>
     */
    private function portalUrls(string $base): array
    {
        $context = WebPortalContext::fromRequest();

        $urls = [['loc' => $base.'/de', 'lastmod' => null]];

        foreach (self::PORTAL_PAGES as $page) {
            if (! $this->leadsPage($base, $page)) {
                continue;
            }

            $urls[] = ['loc' => $base.'/de/'.$page, 'lastmod' => null];
        }

        // Kategorien über die lokalisierten Slugs (CategoryTranslation),
        // dazu je Kategorie die Jahresarchive (2.3).
        $categories = Category::query()
            ->where('is_active', true)
            ->whereIn('portal', $context->portalValues)
            ->with(['translations', 'children' => fn ($query) => $query->where('is_active', true)])
            ->get();

        foreach ($categories as $category) {
            $slug = $category->translations->firstWhere('locale', 'de')?->slug;

            if (! $slug) {
                continue;
            }

            $urls[] = ['loc' => $base.'/de/category/'.$slug, 'lastmod' => null];

            $categoryIds = $category->children->pluck('id')->prepend($category->id)->unique()->values()->all();

            foreach (array_keys(CategoryArchivePageData::yearsWithCounts($categoryIds, $context->portalValues, 'de')) as $year) {
                $urls[] = ['loc' => $base.'/de/category/'.$slug.'/archiv/'.$year, 'lastmod' => null];
            }
        }

        // Öffentliche Newsrooms: aktive Firmen mit mindestens einer im Archiv
        // sichtbaren PM des Portals (gleiche Regel wie NewsroomsPageData).
        $companies = Company::query()
            ->where('is_active', true)
            ->whereHas('pressReleases', fn ($query) => $query
                ->withoutGlobalScope(PortalScope::class)
                ->visibleInArchive()
                ->forPortalValues($context->portalValues))
            ->get(['slug', 'updated_at']);

        foreach ($companies as $company) {
            $urls[] = [
                'loc' => $base.'/de/newsroom/'.$company->slug,
                'lastmod' => $company->updated_at?->toAtomString(),
            ];
        }

        // Alle im Archiv sichtbaren Pressemitteilungen des Portals — bei
        // Doppelveröffentlichung nur die Erstquelle (Anhang 2, 1c.8): eine
        // Sitemap soll nur kanonische Adressen nennen.
        $assigner = app(CanonicalAssigner::class);
        $releases = PressRelease::query()
            ->withoutGlobalScope(PortalScope::class)
            ->visibleInArchive()
            ->forPortalValues($context->portalValues)
            ->when($assigner->enabled(), fn ($query) => $query->whereNull('canonical_press_release_id'))
            ->orderByDesc('published_at')
            ->get(['id', 'slug', 'portal', 'language', 'category_id', 'canonical_press_release_id', 'published_at', 'updated_at']);

        foreach ($releases as $release) {
            if (! $assigner->isCanonicalOn($release, $context->portal)) {
                continue;
            }

            $urls[] = [
                'loc' => $base.'/de/press-release/'.$release->slug,
                'lastmod' => ($release->updated_at ?? $release->published_at)?->toAtomString(),
            ];
        }

        return $urls;
    }
}