add catalog mode

src/Knowledge/Retrieval/NdjsonHybridRetriever.php

@@ -4,7 +4,9 @@ declare(strict_types=1);
 
 namespace App\Knowledge\Retrieval;
 
+use App\Catalog\EntityCatalogService;
 use App\Entity\ModelGenerationConfig;
+use App\Intent\CatalogIntentLite;
 use App\Intent\IntentLite;
 use App\Intent\SalesIntentLite;
 use App\Knowledge\QueryCleaner;
@@ -32,7 +34,9 @@ final class NdjsonHybridRetriever implements RetrieverInterface
         private readonly ModelGenerationConfigRepository $configRepository,
         private readonly QueryCleaner                    $queryCleaner,
         private readonly IntentLite                      $intentLite,
-        private readonly SalesIntentLite                 $salesIntentLite
+        private readonly SalesIntentLite                 $salesIntentLite,
+        private readonly CatalogIntentLite               $catalogIntent,
+        private readonly EntityCatalogService            $entityCatalogService
     )
     {
     }
@@ -54,6 +58,17 @@ final class NdjsonHybridRetriever implements RetrieverInterface
 
     public function retrieveInternal(string $prompt, ModelGenerationConfig $config): array
     {
+        // 🔵 ENTITY CATALOG EARLY EXIT (jetzt auch im Admin-Test aktiv)
+        $entityTerm = $this->catalogIntent->detect($prompt);
+
+        if ($entityTerm !== null) {
+            $catalogBlock = $this->entityCatalogService->listByTerm($entityTerm);
+
+            if ($catalogBlock !== null) {
+                return [$catalogBlock];
+            }
+        }
+
         $core = $this->runCore($prompt, $config, false);
 
         if ($core['ranked_chunk_ids'] === [] || $core['rows'] === []) {
@@ -111,8 +126,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
             return [];
         }
 
-        // 1) Production-like selection: wir selektieren Texte,
-        //    aber in Debug brauchen wir die ChunkIds dazu.
         $selectedChunkIds = $core['is_list_query']
             ? $this->selectChunkIdsListMode($core['ranked_chunk_ids'], $core['rows'], $core['limit'])
             : $this->selectChunkIdsSalesMode($core['ranked_chunk_ids'], $core['rows'], $core['limit']);
@@ -121,7 +134,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
             return [];
         }
 
-        // 2) Ausgabe inklusive Scores
         $out = [];
         $rank = 0;
 
@@ -179,7 +191,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
             $cleanQuery = $prompt;
         }
 
-        // Intent-based adjustments (identisch zur Produktionslogik)
         $threshold = self::VECTOR_SCORE_THRESHOLD;
         $topK = $vectorTopKBase;
 
@@ -216,7 +227,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
 
         $topK = max(1, min($topK, self::HARD_MAX_VECTORK));
 
-        // Tag routing (identisch)
         $candidateDocIds = $this->tagRouting->route($cleanQuery);
         $candidateSet = null;
 
@@ -224,7 +234,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
             $candidateSet = array_fill_keys($candidateDocIds, true);
         }
 
-        // Dual search (identisch)
         $globalHits = $this->vectorClient->search($cleanQuery, $topK);
 
         $scopedHits = [];
@@ -249,7 +258,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
         $rrfScores = [];
         $rawScores = [];
 
-        // RRF (identisch) + optional raw capture
         $this->applyRrfWithOptionalRaw($globalHits, $rrfScores, $rawScores, $threshold, false, $withScores);
         $this->applyRrfWithOptionalRaw(
             $scopedHits,
@@ -292,13 +300,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
         ];
     }
 
-    /**
-     * Gleiche Logik wie applyRrf(), aber optional mit raw-score capture.
-     *
-     * @param array<int, array{chunk_id:string, score:float}> $hits
-     * @param array<string,float> $rrfScores
-     * @param array<string,float> $rawScores
-     */
     private function applyRrfWithOptionalRaw(
         array $hits,
         array &$rrfScores,
@@ -322,7 +323,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
             $chunkId = (string)$hit['chunk_id'];
 
             if ($captureRaw) {
-                // wenn global+scoped vorkommt: bestes raw behalten
                 if (!isset($rawScores[$chunkId]) || $raw > $rawScores[$chunkId]) {
                     $rawScores[$chunkId] = $raw;
                 }
@@ -343,15 +343,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
         }
     }
 
-    // =========================================================
-    // DEBUG SELECTION HELPERS (identisch zu Produktionsregeln)
-    // =========================================================
-
-    /**
-     * List-Mode nutzt exakt collectTexts() Regeln, aber gibt ChunkIds zurück.
-     *
-     * @return string[]
-     */
     private function selectChunkIdsListMode(array $chunkIds, array $rows, int $limit): array
     {
         $seen = [];
@@ -384,11 +375,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
         return $out;
     }
 
-    /**
-     * Normal-Mode nutzt exakt collectSalesOptimized() Regeln, aber gibt ChunkIds zurück.
-     *
-     * @return string[]
-     */
     private function selectChunkIdsSalesMode(array $chunkIds, array $rows, int $limit): array
     {
         $out = [];
@@ -437,10 +423,6 @@ final class NdjsonHybridRetriever implements RetrieverInterface
         return $out;
     }
 
-    // =========================================================
-    // ORIGINAL METHODS (UNVERÄNDERT)
-    // =========================================================
-
     private function applyRrf(array $hits, array &$rrfScores, float $threshold, bool $boost = false): void
     {
         $rank = 0;