Bugfix

src/EventListener/IndexPageListener.php

@@ -7,8 +7,12 @@ use MummertMedia\ContaoMeilisearchBundle\Service\PdfIndexService;
 
 class IndexPageListener
 {
+    public function __construct(
+        private PdfIndexService $pdfIndexService
+    ) {}
     public function onIndexPage(string $content, array &$data, array &$set): void
     {
+        $this->pdfIndexService->startCrawl();
         // Marker vorhanden?
         if (!str_contains($content, 'MEILISEARCH_JSON')) {
             return;
@@ -100,11 +104,9 @@ class IndexPageListener
         $pdfLinks = $this->findPdfLinks($content);
 
         if ($pdfLinks !== []) {
+            $this->pdfIndexService->startCrawl();
             error_log('PDF gefunden');
-
+            $this->pdfIndexService->handlePdfLinks($pdfLinks);
-            /** @var PdfIndexService $service */
-            $service = System::getContainer()->get(PdfIndexService::class);
-            $service->handlePdfLinks($pdfLinks);
         }
     }
 

src/EventListener/PdfCleanupListener.php

@@ -0,0 +1,18 @@
+<?php
+
+namespace MummertMedia\ContaoMeilisearchBundle\EventListener;
+
+use MummertMedia\ContaoMeilisearchBundle\Service\PdfIndexService;
+
+class PdfCleanupListener
+{
+    public function __construct(
+        private PdfIndexService $pdfIndexService
+    ) {}
+
+    public function onLastChunk(): void
+    {
+        error_log('Crawler beendet → PDF Cleanup startet');
+        $this->pdfIndexService->cleanupRemovedPdfs();
+    }
+}

src/Resources/config/services.yaml

@@ -10,3 +10,7 @@ services:
   MummertMedia\ContaoMeilisearchBundle\EventListener\IndexPageListener:
     tags:
       - { name: contao.hook, hook: indexPage, method: onIndexPage }
+
+  MummertMedia\ContaoMeilisearchBundle\EventListener\PdfCleanupListener:
+    tags:
+      - { name: kernel.event_listener, event: terminal42.escargot.last_chunk, method: onLastChunk }

src/Service/PdfIndexService.php

@@ -8,6 +8,29 @@ use Symfony\Component\DependencyInjection\ParameterBag\ParameterBagInterface;
 
 class PdfIndexService
 {
+
+    private int $crawlStart = 0;
+
+    public function startCrawl(): void
+    {
+        if ($this->crawlStart === 0) {
+            $this->crawlStart = time();
+            error_log('PDF Crawl Start: ' . $this->crawlStart);
+        }
+    }
+
+    public function cleanupRemovedPdfs(): void
+    {
+        if ($this->crawlStart === 0) {
+            return;
+        }
+
+        Database::getInstance()
+            ->prepare('DELETE FROM tl_search_pdf WHERE tstamp < ?')
+            ->execute($this->crawlStart);
+
+        error_log('PDF Cleanup abgeschlossen');
+    }
     private string $projectDir;
 
     public function __construct(ParameterBagInterface $params)