add document dele and rebuild faiss index

src/Ingest/IngestFlow.php

@@ -10,18 +10,11 @@ use App\Knowledge\ChunkManager;
 use App\Knowledge\Ingest\KnowledgeIngestService;
 use App\Vector\VectorIndexBuilder;
 use Psr\Log\LoggerInterface;
+use Symfony\Component\Uid\Uuid;
 
 final readonly class IngestFlow
 {
-    /**
-     * Realistische Betriebsgrenze für dieses Systemdesign (CPU Embedding + FlatIP + Full Rebuild).
-     * Wird beim lokalen Ingest (Dokumentversion) enforced.
-     */
     public const CHUNK_LIMIT_HARD = 120000;
-
-    /**
-     * Ab hier nur Warnung (keine Blockade) – damit man frühzeitig reagieren kann.
-     */
     private const CHUNK_LIMIT_WARN = 100000;
 
     public function __construct(
@@ -37,16 +30,10 @@ final readonly class IngestFlow
     {
         $this->metaManager->validateAgainstCurrent();
 
-        // Entfernt alte Chunks dieses Dokuments -> danach ist "existing" der Basis-Index ohne dieses Dokument.
         $this->chunkManager->compactByDocument($version->getDocument()->getId());
 
-        // ------------------------------
-        // Chunk-Limit Guardrail (Hard Cap)
-        // ------------------------------
         $existing = $this->chunkManager->countAllChunks();
 
-        // buildChunkRecords() ist generatorbasiert; für einen sauberen Hard-Cap materialisieren wir lokal,
-        // damit wir vor dem Append abbrechen können (keine Partial Writes).
         $recordsIterable = $this->knowledgeIngestService->buildChunkRecords($version);
         $records = is_array($recordsIterable)
             ? $recordsIterable
@@ -83,6 +70,30 @@ final readonly class IngestFlow
         $this->updateChuckCount();
     }
 
+    /**
+     * HARD DELETE FLOW
+     *
+     * Removes all chunks belonging to a document from index.ndjson
+     * and rebuilds the vector index deterministically.
+     */
+    public function deleteDocument(Uuid $documentId): void
+    {
+        $this->metaManager->validateAgainstCurrent();
+
+        $this->logger->info('Deleting document from RAG index.', [
+            'document_id' => $documentId->toRfc4122(),
+        ]);
+
+        // Remove chunks for this document
+        $this->chunkManager->compactByDocument($documentId);
+
+        // Rebuild vector index from updated NDJSON
+        $this->vectorBuilder->rebuildFromNdjson();
+
+        // Update runtime stats
+        $this->updateChuckCount();
+    }
+
     public function globalReindex(): void
     {
         $allRecords = $this->knowledgeIngestService->buildAllActiveChunkRecords();
@@ -101,4 +112,4 @@ final readonly class IngestFlow
         $chunkCount = $this->chunkManager->countAllChunks();
         $this->metaManager->updateRuntimeStats($chunkCount);
     }
-}
+}