add files uuid

src/EventListener/IndexPageListener.php

@@ -5,8 +5,8 @@ namespace MummertMedia\ContaoMeilisearchBundle\EventListener;
 use Contao\Config;
 use Contao\System;
 use Contao\FilesModel;
-use Contao\File;
 use Contao\StringUtil;
+use Contao\CoreBundle\Filesystem\VirtualFilesystemInterface;
 
 class IndexPageListener
 {
@@ -16,6 +16,8 @@ class IndexPageListener
 
     private function debug(string $message, array $context = []): void
     {
+        // Debug bewusst immer aktiv (bis du es wieder entfernst)
+        // Kontext kurz halten, damit Logs nicht explodieren
         $ctx = $context ? ' | ' . json_encode($context, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) : '';
         error_log('[ContaoMeilisearch][IndexPageListener] ' . $message . $ctx);
     }
@@ -29,20 +31,141 @@ class IndexPageListener
             'set_keys'  => array_keys($set),
         ]);
 
+        /*
+         * =====================
+         * SEITEN-METADATEN
+         * =====================
+         */
+        $hasMeta = str_contains($content, 'MEILISEARCH_JSON');
+        $this->debug('Meta marker scan', [
+            'contains_MEILISEARCH_JSON' => $hasMeta,
+            'content_length'            => strlen($content),
+        ]);
+
+        if ($hasMeta) {
+            try {
+                $parsed = $this->extractMeilisearchJson($content);
+                $this->debug('extractMeilisearchJson(): done', [
+                    'parsed_is_array' => is_array($parsed),
+                    'parsed_keys'     => is_array($parsed) ? array_keys($parsed) : null,
+                ]);
+            } catch (\Throwable $e) {
+                $this->debug('Failed to extract MEILISEARCH_JSON', [
+                    'error' => $e->getMessage(),
+                    'class' => $e::class,
+                ]);
+                $parsed = null;
+            }
+
+            if (is_array($parsed)) {
+
+                // PRIORITY
+                $priority =
+                    $parsed['event']['priority']
+                    ?? $parsed['news']['priority']
+                    ?? $parsed['page']['priority']
+                    ?? null;
+
+                $this->debug('Meta: priority candidate', ['priority' => $priority]);
+
+                if ($priority !== null && $priority !== '') {
+                    $set['priority'] = (int) $priority;
+                }
+
+                // KEYWORDS
+                $keywordSources = [
+                    $parsed['event']['keywords'] ?? null,
+                    $parsed['news']['keywords']  ?? null,
+                    $parsed['page']['keywords']  ?? null,
+                ];
+
+                $this->debug('Meta: keyword sources', ['sources' => $keywordSources]);
+
+                $keywords = [];
+                foreach ($keywordSources as $src) {
+                    if (!is_string($src) || trim($src) === '') {
+                        continue;
+                    }
+                    foreach (preg_split('/\s+/', trim($src)) as $word) {
+                        $keywords[] = $word;
+                    }
+                }
+
+                if ($keywords) {
+                    $set['keywords'] = implode(' ', array_unique($keywords));
+                }
+
+                $this->debug('Meta: keywords result', [
+                    'keywords' => $set['keywords'] ?? null,
+                ]);
+
+                // IMAGEPATH (UUID)
+                $searchImage = $parsed['page']['searchimage'] ?? null;
+                $this->debug('Meta: searchimage candidate', ['searchimage' => $searchImage]);
+
+                if (!empty($searchImage)) {
+                    $set['imagepath'] = trim((string) $searchImage);
+                }
+
+                // STARTDATE
+                $startDate =
+                    $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'] ?? '');
+                    $checksumSeed .= '|' . ($set['keywords']  ?? '');
+                    $checksumSeed .= '|' . ($set['priority']  ?? '');
+                    $checksumSeed .= '|' . ($set['imagepath'] ?? '');
+                    $checksumSeed .= '|' . ($set['startDate'] ?? '');
+
+                    $set['checksum'] = md5($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' => $e::class,
+                    ]);
+                }
+            }
+        }
+
         /*
          * =====================
          * DATEI-ERKENNUNG + UPSERT
          * =====================
          */
         if ((int) ($data['protected'] ?? 0) !== 0) {
+            $this->debug('Abort: protected page', ['protected' => $data['protected'] ?? null]);
             return;
         }
 
-        if (!Config::get('meilisearch_index_files')) {
+        $indexFiles = (bool) 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) {
@@ -52,6 +175,11 @@ class IndexPageListener
             }
         }
 
+        $this->debug('Indexable file links found', [
+            'count' => count($fileLinks),
+            'types' => array_count_values(array_column($fileLinks, 'type')),
+        ]);
+
         if ($fileLinks) {
             $db   = System::getContainer()->get('database_connection');
             $time = time();
@@ -60,39 +188,89 @@ class IndexPageListener
 
             foreach ($fileLinks as $file) {
                 try {
+                    // -------------------------------------------------
+                    // URL normalisieren (Fragment weg)
+                    // -------------------------------------------------
                     $url = strtok($file['url'], '#');
+                    $parts = parse_url($url);
 
-                    // =================================================
+                    // -------------------------------------------------
-                    // 🟢 valider lokaler Dateipfad
+                    // ⬅️ NEU: externe Links überspringen
-                    // =================================================
+                    // -------------------------------------------------
-                    $localFilePath = $this->getValidLocalFilePathFromUrl($url, $projectDir);
+                    if (!empty($parts['host'])) {
+                        // absolute URL → nur erlauben, wenn eigener Host
+                        $currentHost = parse_url(System::getContainer()->get('request_stack')->getCurrentRequest()?->getSchemeAndHttpHost() ?? '', PHP_URL_HOST);
+                        if ($currentHost && $parts['host'] !== $currentHost) {
+                            continue;
+                        }
+                    }
 
-                    // =================================================
+                    // -------------------------------------------------
-                    // UUID nur über echten Contao File-Model
+                    // ⬅️ NEU: lokalen Dateipfad ermitteln
-                    // =================================================
+                    // -------------------------------------------------
+                    $normalizedPath = null;
+
+                    // 1) direkter Pfad /files/…
+                    if (!empty($parts['path'])) {
+                        $candidate = ltrim(urldecode($parts['path']), '/');
+                        if (str_starts_with($candidate, 'files/')) {
+                            $normalizedPath = $candidate;
+                        }
+                    }
+
+                    // 2) Download-Parameter ?file= / ?p= / ?f=
+                    if (!$normalizedPath && !empty($parts['query'])) {
+                        parse_str($parts['query'], $query);
+                        foreach (['file', 'p', 'f'] as $param) {
+                            if (!empty($query[$param])) {
+                                $candidate = ltrim(urldecode(html_entity_decode((string) $query[$param], ENT_QUOTES)), '/');
+                                if (str_starts_with($candidate, 'files/')) {
+                                    $normalizedPath = $candidate;
+                                    break;
+                                }
+                            }
+                        }
+                    }
+
+                    // -------------------------------------------------
+                    // ⬅️ NEU: wenn keine lokale Datei → skip
+                    // -------------------------------------------------
+                    if (!$normalizedPath || !is_file($projectDir . '/public/' . $normalizedPath)) {
+                        continue;
+                    }
+
+                    // -------------------------------------------------
+                    // UUID aus normalisiertem Pfad
+                    // -------------------------------------------------
+                    $uuid    = null;
                     $uuidBin = null;
 
-                    if ($localFilePath !== null) {
+                    if (interface_exists(VirtualFilesystemInterface::class)) {
                         try {
-                            $objFile = new File($localFilePath);
+                            $vfs  = System::getContainer()->get(VirtualFilesystemInterface::class);
-                            $objModel = $objFile->getModel();
+                            $uuid = $vfs->pathToUuid($normalizedPath);
-
+                        } catch (\Throwable) {
-                            if ($objModel && $objModel->uuid) {
+                            $uuid = null;
-                                $uuidBin = $objModel->uuid; // bereits binary(16)
-                            }
-                        } catch (\Throwable $e) {
-                            // bewusst still
                         }
                     }
 
-                    // =================================================
+                    if (!$uuid) {
-                    // bestehende Logik unverändert
+                        $fileModel = FilesModel::findByPath($normalizedPath);
-                    // =================================================
+                        if ($fileModel) {
-                    $abs = $localFilePath
+                            $uuid = $fileModel->uuid;
-                        ? $projectDir . '/' . ltrim($localFilePath, '/')
+                        }
-                        : null;
+                    }
 
-                    $mtime    = ($abs && is_file($abs)) ? filemtime($abs) : 0;
+                    if ($uuid) {
+                        $uuidBin = StringUtil::uuidToBin($uuid);
+                    }
+
+                    // -------------------------------------------------
+                    // bestehende Logik (unverändert)
+                    // -------------------------------------------------
+                    $abs = $projectDir . '/public/' . $normalizedPath;
+
+                    $mtime    = is_file($abs) ? filemtime($abs) : 0;
                     $checksum = md5($url . '|' . $mtime);
 
                     $existing = $db->fetchAssociative(
@@ -133,48 +311,35 @@ class IndexPageListener
                     $this->debug('File upsert FAILED', [
                         'url'   => $file['url'] ?? null,
                         'error' => $e->getMessage(),
+                        'class' => $e::class,
+                        'code'  => $e->getCode(),
                     ]);
                 }
             }
         }
-    }
 
-    // =========================================================
-    // 🟢 PFAD-ERMITTLUNG
-    // =========================================================
-    private function getValidLocalFilePathFromUrl(string $url, string $projectDir): ?string
-    {
-        $arrUrl = parse_url($url);
-        if (!$arrUrl) {
-            return null;
-        }
 
-        $path = $arrUrl['path'] ?? null;
+        $this->debug('Hook end', [
-
+            'final_set_keys' => array_keys($set),
-        // Download-Parameter ?file=
+        ]);
-        if (!empty($arrUrl['query']) && preg_match('#file=(?<path>[^&]+)#i', $arrUrl['query'], $m)) {
-            $path = $m['path'];
-        }
-
-        if (!$path) {
-            return null;
-        }
-
-        $path = ltrim(urldecode($path), '/');
-
-        if (!str_starts_with($path, 'files/')) {
-            return null;
-        }
-
-        if (!is_file($projectDir . '/' . $path)) {
-            return null;
-        }
-
-        return $path;
     }
 
     /* === Hilfsmethoden unverändert === */
 
+    private function extractMeilisearchJson(string $content): ?array
+    {
+        if (!preg_match('/<!--\s*MEILISEARCH_JSON\s*(\{.*?\})\s*-->/s', $content, $m)) {
+            return null;
+        }
+
+        $json = preg_replace('/^\xEF\xBB\xBF/', '', trim($m[1]));
+        $data = json_decode($json, true);
+
+        return json_last_error() === JSON_ERROR_NONE && is_array($data)
+            ? $data
+            : null;
+    }
+
     private function findAllLinks(string $content): array
     {
         if (!preg_match_all(
@@ -200,7 +365,11 @@ 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));
@@ -211,10 +380,12 @@ class IndexPageListener
 
         if (!empty($parts['query'])) {
             parse_str($parts['query'], $query);
+
             foreach (['file', 'p', 'f'] as $param) {
                 if (!empty($query[$param])) {
-                    $candidate = rawurldecode((string) $query[$param]);
+                    $candidate = rawurldecode(html_entity_decode((string) $query[$param], ENT_QUOTES));
                     $ext = strtolower(pathinfo($candidate, PATHINFO_EXTENSION));
+
                     if (in_array($ext, ['pdf', 'docx', 'xlsx', 'pptx'], true)) {
                         return $ext;
                     }