first commit

src/Catalog/EntityCatalogService.php

@@ -4,77 +4,84 @@ declare(strict_types=1);
 
 namespace App\Catalog;
 
+use App\Config\CatalogIntentConfig;
+use App\Entity\Document;
+use App\Tag\TagTypes;
 use App\Tag\TagVectorSearchClient;
 use Doctrine\DBAL\Connection;
 use Symfony\Component\Uid\Uuid;
 
 /**
- * EntityCatalogService
+ * Builds deterministic catalog lists from a validated catalog entity term.
  *
- * Deterministische Katalog-Listen auf Basis eines Entity-Terms:
- * - TagVectorSearch (Score-Gate + Ambiguity-Check)
- * - DB Query auf document_tag + document (ACTIVE)
- * - Rückgabe als EIN Textblock (string) oder null (Fallback auf normalen Retrieval)
- *
- * Schritt-3 Änderung:
- * - Headline ist NICHT mehr hardcoded
- * - Headline basiert dynamisch auf dem gefundenen Tag
+ * This service is intentionally conservative:
+ * - only strong catalog_entity matches may open the catalog path
+ * - ambiguous matches fall back to normal retrieval
+ * - only ACTIVE documents are listed
  */
 final class EntityCatalogService
 {
-    private const MIN_SCORE = 0.55;
-    private const AMBIGUITY_DELTA = 0.05;
+    private const SEARCH_LIMIT = 3;
 
     public function __construct(
         private readonly TagVectorSearchClient $tagVectorClient,
-        private readonly Connection            $connection,
-    ) {}
+        private readonly Connection $connection,
+    ) {
+    }
 
     /**
-     * @return string|null Textblock oder null (wenn kein sicherer Catalog möglich ist)
+     * Returns a catalog text block or null when no safe catalog path exists.
      */
     public function listByTerm(string $entityTerm): ?string
     {
         $entityTerm = trim($entityTerm);
+
         if ($entityTerm === '') {
             return null;
         }
 
-        // 1) Tag-Vektorsuche (Top 3 für Ambiguity-Prüfung)
-        $hits = $this->tagVectorClient->search($entityTerm, 3);
+        $hits = $this->tagVectorClient->search($entityTerm, self::SEARCH_LIMIT);
 
         if ($hits === []) {
             return null;
         }
 
         $best = $hits[0];
+        $bestScore = (float) ($best['score'] ?? 0.0);
 
-        $bestScore = isset($best['score']) ? (float)$best['score'] : 0.0;
-        if ($bestScore < self::MIN_SCORE) {
+        if ($bestScore < CatalogIntentConfig::MIN_SCORE) {
+            return null;
+        }
+
+        if (($best['tag_type'] ?? null) !== TagTypes::CATALOG_ENTITY) {
             return null;
         }
 
-        // 2) Ambiguity: wenn Top2 zu nah ist → konservativ abbrechen
         if (isset($hits[1])) {
-            $secondScore = isset($hits[1]['score']) ? (float)$hits[1]['score'] : 0.0;
-            if (abs($bestScore - $secondScore) < self::AMBIGUITY_DELTA) {
+            $secondScore = (float) ($hits[1]['score'] ?? 0.0);
+
+            if (abs($bestScore - $secondScore) < CatalogIntentConfig::AMBIGUITY_DELTA) {
                 return null;
             }
         }
 
-        $tagHex = (string)($best['tag_id'] ?? '');
-        if ($tagHex === '') {
+        $tagId = trim((string) ($best['tag_id'] ?? ''));
+
+        if ($tagId === '') {
             return null;
         }
 
-        // OPTIONAL: Falls TagVectorSearchClient künftig tag_label zurückliefert,
-        // kann das hier direkt verwendet werden.
-        $tagLabel = isset($best['tag_label']) ? (string)$best['tag_label'] : null;
+        try {
+            $tagBinaryId = Uuid::fromString($tagId)->toBinary();
+        } catch (\Throwable) {
+            return null;
+        }
+
+        $tagLabel = trim((string) ($best['label'] ?? ''));
 
-        // 3) DB Query: alle ACTIVE Dokumente zu diesem Tag
         $rows = $this->connection->fetchAllAssociative(
             '
-            SELECT d.title
+            SELECT DISTINCT d.title
             FROM document d
             INNER JOIN document_tag dt ON dt.document_id = d.id
             WHERE dt.tag_id = :tagId
@@ -82,8 +89,8 @@ final class EntityCatalogService
             ORDER BY d.title ASC
             ',
             [
-                'tagId'  => Uuid::fromString($tagHex)->toBinary(),
-                'status' => 'ACTIVE',
+                'tagId' => $tagBinaryId,
+                'status' => Document::STATUS_ACTIVE,
             ]
         );
 
@@ -92,37 +99,42 @@ final class EntityCatalogService
         }
 
         $titles = [];
+
         foreach ($rows as $row) {
-            $t = trim((string)($row['title'] ?? ''));
-            if ($t !== '') {
-                $titles[] = $t;
+            $title = trim((string) ($row['title'] ?? ''));
+
+            if ($title === '') {
+                continue;
             }
+
+            $titles[$title] = $title;
         }
 
         if ($titles === []) {
             return null;
         }
 
-        return $this->buildTextBlock($tagLabel, $titles);
+        return $this->buildTextBlock(
+            $tagLabel !== '' ? $tagLabel : null,
+            array_values($titles)
+        );
     }
 
     /**
-     * Dynamische Headline:
-     * - Wenn Tag-Label vorhanden → verwenden
-     * - Sonst generischer Fallback
+     * Builds a stable human-readable list block for the prompt.
+     *
+     * @param list<string> $titles
      */
     private function buildTextBlock(?string $tagLabel, array $titles): string
     {
         $headline = 'Folgende Einträge sind verfügbar:';
 
-        if (\is_string($tagLabel) && \trim($tagLabel) !== '') {
-            $headline = sprintf(
-                'Folgende %s sind verfügbar:',
-                $tagLabel
-            );
+        if ($tagLabel !== null && trim($tagLabel) !== '') {
+            $headline = sprintf('Folgende %s sind verfügbar:', trim($tagLabel));
         }
 
         $lines = [];
+
         foreach ($titles as $title) {
             $lines[] = '- ' . $title;
         }