Sitemap lastmod and handling of past events

- SitemapListener: <lastmod> for event detail pages (event tstamp) and
  regular pages; events that ended more than 12 months ago are removed
  from the sitemap
- PastEventPolicy: events stay indexable for 12 months after they end
  (recurring events are searched before the next date), afterwards the
  detail page gets noindex,follow but stays reachable as archive
- detail template data: is_past and next_occurrence (next published
  event with the same title) for a notice with a link to the next date

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

src/EventListener/EventFullTemplateDataListener.php

@@ -4,7 +4,9 @@ declare(strict_types=1);
 
 namespace Mummert\EventManagerBundle\EventListener;
 
+use Contao\CalendarEventsModel;
 use Contao\CoreBundle\DependencyInjection\Attribute\AsHook;
+use Contao\CoreBundle\Routing\ContentUrlGenerator;
 use Contao\CoreBundle\Routing\ResponseContext\HtmlHeadBag\HtmlHeadBag;
 use Contao\CoreBundle\Routing\ResponseContext\ResponseContextAccessor;
 use Contao\StringUtil;
@@ -12,6 +14,7 @@ use Contao\Template;
 use Doctrine\DBAL\Connection;
 use Doctrine\DBAL\ParameterType;
 use Mummert\EventManagerBundle\Seo\EventMetaTitle;
+use Mummert\EventManagerBundle\Seo\PastEventPolicy;
 
 #[AsHook('parseTemplate', method: 'onParseTemplate')]
 class EventFullTemplateDataListener
@@ -21,6 +24,7 @@ class EventFullTemplateDataListener
     public function __construct(
         private readonly Connection $connection,
         private readonly ResponseContextAccessor $responseContextAccessor,
+        private readonly ContentUrlGenerator $contentUrlGenerator,
     ) {
     }
 
@@ -47,6 +51,7 @@ class EventFullTemplateDataListener
 
         $this->setMetaTitle($data);
         $this->setMetaDescriptionFallback($data, $locationRow);
+        $this->applyPastEventPolicy($template, $data, $eventId);
 
         $template->location_row = $locationRow;
         $template->organizations = $organizations;
@@ -138,6 +143,56 @@ class EventFullTemplateDataListener
         $headBag->setMetaDescription($description.'. Termin und Infos im Veranstaltungskalender Nossener Land.');
     }
 
+    /**
+     * Vergangene Events: Hinweis + Link auf den nächsten Termin gleichen Namens;
+     * länger als PastEventPolicy::INDEX_MONTHS vorbei: noindex,follow.
+     *
+     * @param array<string, mixed> $data
+     */
+    private function applyPastEventPolicy(Template $template, array $data, int $eventId): void
+    {
+        $template->is_past = PastEventPolicy::isPast($data);
+        $template->next_occurrence = null;
+
+        if (!$template->is_past) {
+            return;
+        }
+
+        if (PastEventPolicy::isOutsideIndexWindow($data)) {
+            $headBag = $this->responseContextAccessor->getResponseContext()?->get(HtmlHeadBag::class);
+
+            if ($headBag instanceof HtmlHeadBag) {
+                $headBag->setMetaRobots('noindex,follow');
+            }
+        }
+
+        // nächster veröffentlichter Termin mit gleichem (gespeicherten) Titel
+        $nextId = $this->connection->fetchOne(
+            "SELECT e.id FROM tl_calendar_events e
+              WHERE e.published = 1 AND e.id != ?
+                AND LOWER(TRIM(e.title)) = (SELECT LOWER(TRIM(o.title)) FROM tl_calendar_events o WHERE o.id = ?)
+                AND COALESCE(NULLIF(e.endDate, 0), e.startDate) >= ?
+                AND (e.start = '' OR e.start <= ?) AND (e.stop = '' OR e.stop > ?)
+              ORDER BY e.startDate ASC LIMIT 1",
+            [$eventId, $eventId, strtotime('today'), time(), time()],
+        );
+
+        $next = $nextId ? CalendarEventsModel::findById((int) $nextId) : null;
+
+        if (null === $next) {
+            return;
+        }
+
+        try {
+            $template->next_occurrence = [
+                'url' => $this->contentUrlGenerator->generate($next),
+                'date' => EventMetaTitle::formatDate((int) $next->startDate, (int) $next->endDate),
+            ];
+        } catch (\Throwable) {
+            // Kein Link möglich (z. B. Archiv ohne Zielseite): nur der Hinweis
+        }
+    }
+
     /** @return array<string, string> */
     private function fetchLocationRow(int $locationId): array
     {

src/EventListener/SitemapListener.php

@@ -0,0 +1,104 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Mummert\EventManagerBundle\EventListener;
+
+use Contao\CoreBundle\Event\SitemapEvent;
+use Doctrine\DBAL\Connection;
+use Mummert\EventManagerBundle\Seo\PastEventPolicy;
+use Symfony\Component\EventDispatcher\Attribute\AsEventListener;
+
+/**
+ * Ergänzt die Contao-Sitemap:
+ *
+ * - <lastmod> für Event-Detailseiten (tstamp des Events) und reguläre Seiten
+ *   (tstamp der Seite), damit Suchmaschinen geänderte Seiten gezielt neu abrufen
+ * - entfernt Events, deren Ende länger als PastEventPolicy::INDEX_MONTHS zurückliegt;
+ *   diese Seiten stehen auf noindex,follow (EventFullTemplateDataListener)
+ *
+ * Läuft nach dem SitemapListener des Calendar-Bundles (der die Events einfügt).
+ */
+#[AsEventListener(priority: -100)]
+class SitemapListener
+{
+    private const NS = 'http://www.sitemaps.org/schemas/sitemap/0.9';
+
+    public function __construct(
+        private readonly Connection $connection,
+    ) {
+    }
+
+    public function __invoke(SitemapEvent $event): void
+    {
+        $document = $event->getDocument();
+        $xpath = new \DOMXPath($document);
+        $xpath->registerNamespace('s', self::NS);
+
+        $events = $this->loadEvents();
+        $pages = $this->loadPages();
+
+        /** @var \DOMElement $url */
+        foreach (iterator_to_array($xpath->query('//s:url')) as $url) {
+            $loc = $xpath->query('s:loc', $url)->item(0);
+
+            if (!$loc instanceof \DOMElement || $xpath->query('s:lastmod', $url)->length > 0) {
+                continue;
+            }
+
+            $path = rawurldecode(trim((string) parse_url(trim($loc->textContent), PHP_URL_PATH), '/'));
+            $segments = '' === $path ? [] : explode('/', $path);
+            $lastmod = null;
+
+            if (\count($segments) >= 2 && isset($events[end($segments)])) {
+                $row = $events[end($segments)];
+
+                if (PastEventPolicy::isOutsideIndexWindow($row)) {
+                    $url->parentNode?->removeChild($url);
+
+                    continue;
+                }
+
+                $lastmod = (int) $row['tstamp'];
+            } elseif (isset($pages[$path])) {
+                $lastmod = $pages[$path];
+            }
+
+            if ($lastmod > 0) {
+                $element = $document->createElementNS(self::NS, 'lastmod', date('Y-m-d\TH:i:sP', $lastmod));
+                $loc->parentNode?->insertBefore($element, $loc->nextSibling);
+            }
+        }
+    }
+
+    /**
+     * @return array<string, array<string, mixed>> alias => Event (bei doppelten Aliassen der neueste Stand)
+     */
+    private function loadEvents(): array
+    {
+        $events = [];
+
+        foreach ($this->connection->fetchAllAssociative('SELECT alias, tstamp, startDate, endDate FROM tl_calendar_events WHERE published = 1 ORDER BY tstamp') as $row) {
+            if ('' !== (string) $row['alias']) {
+                $events[(string) $row['alias']] = $row;
+            }
+        }
+
+        return $events;
+    }
+
+    /**
+     * @return array<string, int> URL-Pfad => tstamp ("" = Startseite)
+     */
+    private function loadPages(): array
+    {
+        $pages = [];
+
+        foreach ($this->connection->fetchAllAssociative("SELECT alias, tstamp FROM tl_page WHERE type = 'regular' AND published = 1") as $row) {
+            $alias = (string) $row['alias'];
+            $pages['index' === $alias ? '' : $alias] = (int) $row['tstamp'];
+        }
+
+        return $pages;
+    }
+}

src/Seo/PastEventPolicy.php

@@ -0,0 +1,48 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Mummert\EventManagerBundle\Seo;
+
+/**
+ * Umgang mit vergangenen Veranstaltungen.
+ *
+ * Bis INDEX_MONTHS nach dem Ende bleiben Events indexierbar (wiederkehrende Events
+ * wie Weihnachtsmärkte werden vor dem nächsten Termin gesucht). Danach:
+ * noindex,follow und nicht mehr in der Sitemap. Die Seiten bleiben als Archiv erreichbar.
+ */
+final class PastEventPolicy
+{
+    public const INDEX_MONTHS = 12;
+
+    /**
+     * @param array<string, mixed> $event (startDate, endDate)
+     */
+    public static function endTimestamp(array $event): int
+    {
+        $end = (int) ($event['endDate'] ?? 0) ?: (int) ($event['startDate'] ?? 0);
+
+        // Ende des letzten Veranstaltungstags
+        return $end > 0 ? (int) strtotime('tomorrow', $end) - 1 : 0;
+    }
+
+    /**
+     * @param array<string, mixed> $event
+     */
+    public static function isPast(array $event, ?int $now = null): bool
+    {
+        $end = self::endTimestamp($event);
+
+        return $end > 0 && $end < ($now ?? time());
+    }
+
+    /**
+     * @param array<string, mixed> $event
+     */
+    public static function isOutsideIndexWindow(array $event, ?int $now = null): bool
+    {
+        $end = self::endTimestamp($event);
+
+        return $end > 0 && $end < (int) strtotime('-'.self::INDEX_MONTHS.' months', $now ?? time());
+    }
+}