add files uuid

src/Command/MeilisearchFilesParseCommand.php

@@ -4,6 +4,7 @@ namespace MummertMedia\ContaoMeilisearchBundle\Command;
 
 use Contao\CoreBundle\Framework\ContaoFramework;
 use Contao\Database;
+use Contao\System;
 use Symfony\Component\Console\Command\Command;
 use Symfony\Component\Console\Input\InputInterface;
 use Symfony\Component\Console\Input\InputOption;
@@ -102,7 +103,11 @@ class MeilisearchFilesParseCommand extends Command
                 continue;
             }
 
-            $absolutePath = TL_ROOT . '/' . $normalized;
+            $root = defined('TL_ROOT')
+                ? TL_ROOT
+                : System::getContainer()->getParameter('kernel.project_dir') . '/public';
+
+            $absolutePath = $root . '/' . $normalized;
 
             if (!is_file($absolutePath)) {
                 $this->log('File missing, skip', [

src/EventListener/IndexPageListener.php

@@ -4,21 +4,18 @@ namespace MummertMedia\ContaoMeilisearchBundle\EventListener;
 
 use Contao\Config;
 use Contao\System;
-use Doctrine\DBAL\Connection;
+use MummertMedia\ContaoMeilisearchBundle\Service\MeilisearchFileHelper;
 
 class IndexPageListener
 {
     public function __construct(
-        private readonly Connection $connection,
+        private readonly MeilisearchFileHelper $fileHelper,
     ) {
     }
 
     private function debug(string $message, array $context = []): void
     {
-        $ctx = $context
+        $ctx = $context ? ' | ' . json_encode($context, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) : '';
-            ? ' | ' . json_encode($context, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE)
-            : '';
-
         error_log('[ContaoMeilisearch][IndexPageListener] ' . $message . $ctx);
     }
 
@@ -79,8 +76,6 @@ class IndexPageListener
                     $parsed['page']['keywords']  ?? null,
                 ];
 
-                $this->debug('Meta: keyword sources', ['sources' => $keywordSources]);
-
                 $keywords = [];
                 foreach ($keywordSources as $src) {
                     if (!is_string($src) || trim($src) === '') {
@@ -95,77 +90,41 @@ class IndexPageListener
                     $set['keywords'] = implode(' ', array_unique($keywords));
                 }
 
-                $this->debug('Meta: keywords result', [
+                // IMAGEPATH
-                    'keywords' => $set['keywords'] ?? null,
+                if (!empty($parsed['page']['searchimage'] ?? null)) {
-                ]);
+                    $set['imagepath'] = trim((string) $parsed['page']['searchimage']);
-
-                // IMAGEPATH (UUID)
-                $searchImage = $parsed['page']['searchimage'] ?? null;
-                $this->debug('Meta: searchimage candidate', ['searchimage' => $searchImage]);
-
-                if (!empty($searchImage)) {
-                    $set['imagepath'] = trim((string) $searchImage);
                 }
 
                 // STARTDATE
-                $startDate =
+                if (is_numeric($parsed['event']['startDate'] ?? null)) {
-                    $parsed['event']['startDate']
+                    $set['startDate'] = (int) $parsed['event']['startDate'];
-                    ?? $parsed['news']['startDate']
-                    ?? null;
-
-                $this->debug('Meta: startDate candidate', ['startDate' => $startDate]);
-
-                if (is_numeric($startDate) && (int) $startDate > 0) {
-                    $set['startDate'] = (int) $startDate;
                 }
 
                 // CHECKSUM
-                try {
+                $checksumSeed  = (string) ($data['checksum'] ?? '');
-                    $checksumSeed  = (string) ($data['checksum'] ?? '');
+                $checksumSeed .= '|' . ($set['keywords']  ?? '');
-                    $checksumSeed .= '|' . ($set['keywords']  ?? '');
+                $checksumSeed .= '|' . ($set['priority']  ?? '');
-                    $checksumSeed .= '|' . ($set['priority']  ?? '');
+                $checksumSeed .= '|' . ($set['imagepath'] ?? '');
-                    $checksumSeed .= '|' . ($set['imagepath'] ?? '');
+                $checksumSeed .= '|' . ($set['startDate'] ?? '');
-                    $checksumSeed .= '|' . ($set['startDate'] ?? '');
 
-                    $set['checksum'] = md5($checksumSeed);
+                $set['checksum'] = md5($checksumSeed);
-
-                    $this->debug('Checksum generated', [
-                        'seed_preview' => substr($checksumSeed, 0, 120) . (strlen($checksumSeed) > 120 ? '…' : ''),
-                        'checksum'     => $set['checksum'],
-                    ]);
-                } catch (\Throwable $e) {
-                    $this->debug('Failed to generate checksum', [
-                        'error' => $e->getMessage(),
-                        'class' => $e::class,
-                    ]);
-                }
             }
         }
 
         /*
          * =====================
-         * DATEI-ERKENNUNG + UPSERT
+         * DATEI-ERKENNUNG (NUR ERKENNUNG!)
          * =====================
          */
         if ((int) ($data['protected'] ?? 0) !== 0) {
-            $this->debug('Abort: protected page', ['protected' => $data['protected'] ?? null]);
             return;
         }
 
-        $indexFiles = (bool) Config::get('meilisearch_index_files');
+        if (!Config::get('meilisearch_index_files')) {
-
-        $this->debug('File indexing setting', [
-            'meilisearch_index_files' => $indexFiles,
-        ]);
-
-        if (!$indexFiles) {
-            $this->debug('Abort: file indexing disabled');
             return;
         }
 
         $links = $this->findAllLinks($content);
-        $this->debug('Links found', ['count' => count($links)]);
-
         $fileLinks = [];
 
         foreach ($links as $link) {
@@ -177,72 +136,15 @@ class IndexPageListener
 
         $this->debug('Indexable file links found', [
             'count' => count($fileLinks),
-            'types' => array_count_values(array_column($fileLinks, 'type')),
         ]);
 
         if ($fileLinks) {
-
-            $this->debug('About to use Doctrine DBAL connection');
-
-            $db = $this->connection;
-
-            $this->debug('Doctrine DBAL connection ready', [
-                'driver' => get_class($db->getDriver()),
-            ]);
-
-            $time = time();
-
             foreach ($fileLinks as $file) {
-                $url = strtok($file['url'], '#');
+                $this->fileHelper->collect(
-
+                    $file['url'],
-                $path = parse_url($url, PHP_URL_PATH);
+                    $file['type'],
-                $abs  = $path ? TL_ROOT . '/' . ltrim($path, '/') : null;
+                    (int) ($data['pid'] ?? 0)
-
-                $mtime = ($abs && is_file($abs)) ? filemtime($abs) : 0;
-                $checksum = md5($url . '|' . $mtime);
-
-                $existing = $db->fetchAssociative(
-                    'SELECT id, checksum FROM tl_search_files WHERE url = ?',
-                    [$url]
                 );
-
-                if ($existing) {
-                    $db->update(
-                        'tl_search_files',
-                        [
-                            'tstamp'      => $time,
-                            'last_seen'   => $time,
-                            'page_id'     => (int) ($data['pid'] ?? 0),
-                            'file_mtime'  => $mtime,
-                            'checksum'    => $checksum,
-                        ],
-                        ['id' => $existing['id']]
-                    );
-
-                    $this->debug('File updated', [
-                        'url'      => $url,
-                        'checksum' => $checksum,
-                    ]);
-                } else {
-                    $db->insert(
-                        'tl_search_files',
-                        [
-                            'tstamp'     => $time,
-                            'last_seen'  => $time,
-                            'type'       => $file['type'],
-                            'url'        => $url,
-                            'title'      => $file['linkText'] ?? basename($url),
-                            'page_id'    => (int) ($data['pid'] ?? 0),
-                            'file_mtime' => $mtime,
-                            'checksum'   => $checksum,
-                        ]
-                    );
-
-                    $this->debug('File inserted', [
-                        'url'      => $url,
-                        'checksum' => $checksum,
-                    ]);
-                }
             }
         }
 
@@ -292,11 +194,7 @@ class IndexPageListener
     private function detectIndexableFileType(string $url): ?string
     {
         $url = strtok($url, '#');
-
         $parts = parse_url($url);
-        if (!$parts) {
-            return null;
-        }
 
         if (!empty($parts['path'])) {
             $ext = strtolower(pathinfo($parts['path'], PATHINFO_EXTENSION));

src/Resources/contao/dca/tl_search_files.php

@@ -12,7 +12,8 @@ $GLOBALS['TL_DCA']['tl_search_files'] = [
                 'url'       => 'unique',
                 'type'      => 'index',
                 'checksum'  => 'index',
-                'last_seen' => 'index', // ⬅️ NEU (für Cleanup-Performance)
+                'uuid'      => 'index',
+                'last_seen' => 'index',
             ],
         ],
     ],
@@ -63,6 +64,10 @@ $GLOBALS['TL_DCA']['tl_search_files'] = [
             'sql' => "mediumtext NULL",
         ],
 
+        'uuid' => [
+            'sql' => "binary(16) NULL",
+        ],
+
         /*
          * md5(url + filemtime)
          * → erkennt Änderungen zuverlässig

src/Service/MeilisearchFileHelper.php

@@ -0,0 +1,259 @@
+<?php
+
+namespace MummertMedia\ContaoMeilisearchBundle\Service;
+
+use Contao\FilesModel;
+use Contao\StringUtil;
+use Contao\System;
+use Doctrine\DBAL\Connection;
+
+class MeilisearchFileHelper
+{
+    public function __construct(
+        private readonly Connection $connection,
+    ) {
+    }
+
+    /**
+     * Zentrale Datei-Verarbeitung
+     */
+    public function collect(string $url, string $type, int $pageId): void
+    {
+        $this->log('collect() start', [
+            'url'    => $url,
+            'type'   => $type,
+            'pageId' => $pageId,
+        ]);
+
+        // -------------------------------------------------
+        // 1. URL normalisieren
+        // -------------------------------------------------
+        $cleanUrl = strtok($url, '#');
+        $parts    = parse_url($cleanUrl);
+
+        if (!$parts) {
+            $this->log('Invalid URL, skip');
+            return;
+        }
+
+        // -------------------------------------------------
+        // 2. Externe Datei? → skip
+        // -------------------------------------------------
+        if (!empty($parts['host'])) {
+            $currentRequest = System::getContainer()
+                ->get('request_stack')
+                ->getCurrentRequest();
+
+            $pageHost = $currentRequest
+                ? parse_url($currentRequest->getSchemeAndHttpHost(), PHP_URL_HOST)
+                : null;
+
+            if ($pageHost && $parts['host'] !== $pageHost) {
+                $this->log('External file detected, skip', [
+                    'host' => $parts['host'],
+                ]);
+                return;
+            }
+        }
+
+        // -------------------------------------------------
+        // 3. Pfad-Kandidaten sammeln (ohne Annahmen!)
+        // -------------------------------------------------
+        $query = [];
+        if (!empty($parts['query'])) {
+            parse_str($parts['query'], $query);
+        }
+
+        $pathCandidates = [];
+
+        // direkter Pfad
+        if (!empty($parts['path'])) {
+            $pathCandidates[] = $parts['path'];
+        }
+
+        // Download-Parameter
+        foreach (['file', 'f', 'p'] as $param) {
+            if (!empty($query[$param])) {
+                $pathCandidates[] = $query[$param];
+            }
+        }
+
+        // normalisieren
+        $pathCandidates = array_values(array_unique(array_filter(array_map(
+            static function ($candidate) {
+                $candidate = rawurldecode(html_entity_decode((string) $candidate, ENT_QUOTES));
+                return ltrim($candidate, '/') ?: null;
+            },
+            $pathCandidates
+        ))));
+
+        $this->log('Path candidates (normalized)', [
+            'candidates' => $pathCandidates,
+        ]);
+
+        // -------------------------------------------------
+        // 4. FilesModel (DBAFS) auflösen → UUID
+        // -------------------------------------------------
+        $fileModel = null;
+
+        foreach ($pathCandidates as $candidate) {
+
+            // 1) direkt
+            $model = FilesModel::findByPath($candidate);
+            if ($model && $model->uuid) {
+                $fileModel = $model;
+                $this->log('Resolved via FilesModel (direct)', [
+                    'candidate' => $candidate,
+                    'path'      => $model->path,
+                ]);
+                break;
+            }
+
+            // 2) fallback: files/ davor
+            if (!str_starts_with($candidate, 'files/')) {
+                $model = FilesModel::findByPath('files/' . $candidate);
+                if ($model && $model->uuid) {
+                    $fileModel = $model;
+                    $this->log('Resolved via FilesModel (files/ prefix)', [
+                        'candidate' => $candidate,
+                        'path'      => $model->path,
+                    ]);
+                    break;
+                }
+            }
+        }
+
+        if (!$fileModel) {
+            $this->log('No Contao file model found, skip', [
+                'candidates' => $pathCandidates,
+            ]);
+            return;
+        }
+
+        $normalizedPath = (string) $fileModel->path;
+        $uuidBin        = $fileModel->uuid;
+        $uuid           = StringUtil::binToUuid($uuidBin);
+        $canonicalUrl   = '/' . ltrim($normalizedPath, '/');
+
+        $this->log('UUID resolved', [
+            'path' => $canonicalUrl,
+            'uuid' => $uuid,
+        ]);
+
+        // -------------------------------------------------
+        // 5. Datei im Filesystem prüfen
+        // -------------------------------------------------
+        $projectDir = System::getContainer()->getParameter('kernel.project_dir');
+        $abs        = $projectDir . '/public/' . $normalizedPath;
+
+        if (!is_file($abs)) {
+            $this->log('Resolved model but file missing on filesystem, skip', [
+                'path' => $normalizedPath,
+                'abs'  => $abs,
+            ]);
+            return;
+        }
+
+// -------------------------------------------------
+// 6. Redaktionellen Titel aus tl_files.meta
+// -------------------------------------------------
+        $title = null;
+        $meta  = StringUtil::deserialize($fileModel->meta, true);
+
+// 1) bevorzugte Sprache (falls vorhanden)
+        $lang = $GLOBALS['TL_LANGUAGE'] ?? null;
+        if ($lang && !empty($meta[$lang]['title'])) {
+            $title = trim((string) $meta[$lang]['title']);
+        }
+
+// 2) Fallback: erste verfügbare Sprache
+        if ($title === null && is_array($meta)) {
+            foreach ($meta as $langKey => $langMeta) {
+                if (!empty($langMeta['title'])) {
+                    $title = trim((string) $langMeta['title']);
+                    break;
+                }
+            }
+        }
+
+        if ($title) {
+            $this->log('Title resolved from tl_files', [
+                'title' => $title,
+            ]);
+        }
+
+        // -------------------------------------------------
+        // 7. Datei-Infos
+        // -------------------------------------------------
+        $mtime    = filemtime($abs) ?: 0;
+        $checksum = md5($normalizedPath . '|' . $mtime);
+        $now      = time();
+
+        // -------------------------------------------------
+        // 8. Upsert über UUID
+        // -------------------------------------------------
+        $existing = $this->connection->fetchAssociative(
+            'SELECT id FROM tl_search_files WHERE uuid = ?',
+            [$uuidBin]
+        );
+
+        if ($existing) {
+            $data = [
+                'tstamp'     => $now,
+                'last_seen'  => $now,
+                'type'       => $type,
+                'url'        => $canonicalUrl,
+                'page_id'    => $pageId,
+                'file_mtime' => $mtime,
+                'checksum'   => $checksum,
+            ];
+
+            if ($title !== null) {
+                $data['title'] = $title;
+            }
+
+            $this->connection->update(
+                'tl_search_files',
+                $data,
+                ['id' => $existing['id']]
+            );
+
+            $this->log('File updated by UUID', [
+                'uuid' => $uuid,
+            ]);
+        } else {
+            $this->connection->insert(
+                'tl_search_files',
+                [
+                    'tstamp'     => $now,
+                    'last_seen'  => $now,
+                    'type'       => $type,
+                    'url'        => $canonicalUrl,
+                    'title'      => $title ?? basename($normalizedPath),
+                    'page_id'    => $pageId,
+                    'file_mtime' => $mtime,
+                    'checksum'   => $checksum,
+                    'uuid'       => $uuidBin,
+                ]
+            );
+
+            $this->log('File inserted by UUID', [
+                'uuid' => $uuid,
+            ]);
+        }
+
+        $this->log('collect() end');
+    }
+
+    // -------------------------------------------------
+    // Logging
+    // -------------------------------------------------
+    private function log(string $message, array $context = []): void
+    {
+        $ctx = $context
+            ? ' | ' . json_encode($context, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE)
+            : '';
+
+        error_log('[ContaoMeilisearch][MeilisearchFileHelper] ' . $message . $ctx);
+    }
+}