lexical logic

src/Ingest/VectorRebuildService.php

@@ -6,36 +6,50 @@ namespace App\Ingest;
 
 use App\Index\IndexMetaManager;
 use App\Knowledge\ChunkManager;
+use App\Knowledge\Retrieval\NdjsonLexicalIndexBuilder;
 use App\Vector\VectorIndexBuilder;
 
 final readonly class VectorRebuildService
 {
     public function __construct(
         private VectorIndexBuilder $vectorBuilder,
-        private IndexMetaManager   $metaManager,
-        private ChunkManager       $chunkManager,
-    ) {}
+        private NdjsonLexicalIndexBuilder $lexicalIndexBuilder,
+        private IndexMetaManager $metaManager,
+        private ChunkManager $chunkManager,
+    ) {
+    }
 
     /**
-     * Führt einen vollständigen, deterministischen FAISS-Rebuild aus.
+     * Executes a full deterministic rebuild of all derived retrieval artifacts.
      *
-     * Ablauf:
-     * 1. Rebuild des Vector Index aus index.ndjson
-     * 2. Chunk-Zählung via ChunkManager
-     * 3. Runtime-Stats atomar aktualisieren
+     * Flow:
+     * 1. Ensure index_meta.json exists
+     * 2. Rebuild vector index from index.ndjson
+     * 3. Rebuild lexical index from index.ndjson
+     * 4. Count chunks streaming-safe
+     * 5. Update runtime stats atomically
+     *
+     * Important:
+     * - Vector and lexical index are both derived from the same NDJSON source
+     * - rebuilding both here prevents drift between semantic and lexical retrieval layers
+     * - failures in either derived artifact should fail the rebuild as a whole
+     * @throws \Throwable
      */
     public function rebuild(?string $logPath = null): void
     {
-        // ✅ Stelle sicher, dass index_meta.json existiert
+        // Ensure metadata exists before derived index work starts.
         $this->metaManager->ensureExists();
 
-        // 1️⃣ Vector Index neu bauen
+        // 1) Rebuild semantic vector index.
         $this->vectorBuilder->rebuildFromNdjson($logPath);
 
-        // 2️⃣ Chunk Count streaming-safe zählen
+        // 2) Rebuild generic lexical index from the same NDJSON source.
+        $this->lexicalIndexBuilder->build();
+
+        // 3) Count chunks streaming-safe.
         $chunkCount = $this->chunkManager->countAllChunks();
 
-        // 3️⃣ Runtime-Stats aktualisieren (atomar)
+        // 4) Update runtime stats atomically.
         $this->metaManager->updateRuntimeStats($chunkCount);
     }
 }