Add optional LLM cache deletion when deleting documents

• Add delete_llm_cache parameter to API
• Collect cache IDs from text chunks
• Delete cache after graph operations
• Update UI with new checkbox option
• Add i18n translations for cache option

lightrag/api/routers/document_routes.py

@@ -336,6 +336,10 @@ class DeleteDocRequest(BaseModel):
         default=False,
         description="Whether to delete the corresponding file in the upload directory.",
     )
+    delete_llm_cache: bool = Field(
+        default=False,
+        description="Whether to delete cached LLM extraction results for the documents.",
+    )
 
     @field_validator("doc_ids", mode="after")
     @classmethod
@@ -1487,6 +1491,7 @@ async def background_delete_documents(
     doc_manager: DocumentManager,
     doc_ids: List[str],
     delete_file: bool = False,
+    delete_llm_cache: bool = False,
 ):
     """Background task to delete multiple documents"""
     from lightrag.kg.shared_storage import (
@@ -1521,6 +1526,10 @@ async def background_delete_documents(
         )
         # Use slice assignment to clear the list in place
         pipeline_status["history_messages"][:] = ["Starting document deletion process"]
+        if delete_llm_cache:
+            pipeline_status["history_messages"].append(
+                "LLM cache cleanup requested for this deletion job"
+            )
 
     try:
         # Loop through each document ID and delete them one by one
@@ -1534,7 +1543,9 @@ async def background_delete_documents(
 
             file_path = "#"
             try:
-                result = await rag.adelete_by_doc_id(doc_id)
+                result = await rag.adelete_by_doc_id(
+                    doc_id, delete_llm_cache=delete_llm_cache
+                )
                 file_path = (
                     getattr(result, "file_path", "-") if "result" in locals() else "-"
                 )
@@ -2344,21 +2355,20 @@ def create_document_routes(
         Delete documents and all their associated data by their IDs using background processing.
 
         Deletes specific documents and all their associated data, including their status,
-        text chunks, vector embeddings, and any related graph data.
+        text chunks, vector embeddings, and any related graph data. When requested,
+        cached LLM extraction responses are removed after graph deletion/rebuild completes.
         The deletion process runs in the background to avoid blocking the client connection.
-        It is disabled when llm cache for entity extraction is disabled.
 
         This operation is irreversible and will interact with the pipeline status.
 
         Args:
-            delete_request (DeleteDocRequest): The request containing the document IDs and delete_file options.
+            delete_request (DeleteDocRequest): The request containing the document IDs and deletion options.
             background_tasks: FastAPI BackgroundTasks for async processing
 
         Returns:
             DeleteDocByIdResponse: The result of the deletion operation.
                 - status="deletion_started": The document deletion has been initiated in the background.
                 - status="busy": The pipeline is busy with another operation.
-                - status="not_allowed": Operation not allowed when LLM cache for entity extraction is disabled.
 
         Raises:
             HTTPException:
@@ -2366,15 +2376,6 @@ def create_document_routes(
         """
         doc_ids = delete_request.doc_ids
 
-        # The rag object is initialized from the server startup args,
-        # so we can access its properties here.
-        if not rag.enable_llm_cache_for_entity_extract:
-            return DeleteDocByIdResponse(
-                status="not_allowed",
-                message="Operation not allowed when LLM cache for entity extraction is disabled.",
-                doc_id=", ".join(delete_request.doc_ids),
-            )
-
         try:
             from lightrag.kg.shared_storage import get_namespace_data
 
@@ -2395,6 +2396,7 @@ def create_document_routes(
                 doc_manager,
                 doc_ids,
                 delete_request.delete_file,
+                delete_request.delete_llm_cache,
             )
 
             return DeleteDocByIdResponse(

lightrag/lightrag.py

@@ -2793,7 +2793,9 @@ class LightRAG:
         # Return the dictionary containing statuses only for the found document IDs
         return found_statuses
 
-    async def adelete_by_doc_id(self, doc_id: str) -> DeletionResult:
+    async def adelete_by_doc_id(
+        self, doc_id: str, delete_llm_cache: bool = False
+    ) -> DeletionResult:
         """Delete a document and all its related data, including chunks, graph elements.
 
         This method orchestrates a comprehensive deletion process for a given document ID.
@@ -2803,6 +2805,8 @@ class LightRAG:
 
         Args:
             doc_id (str): The unique identifier of the document to be deleted.
+            delete_llm_cache (bool): Whether to delete cached LLM extraction results
+                associated with the document. Defaults to False.
 
         Returns:
             DeletionResult: An object containing the outcome of the deletion process.
@@ -2814,6 +2818,7 @@ class LightRAG:
         """
         deletion_operations_started = False
         original_exception = None
+        doc_llm_cache_ids: list[str] = []
 
         # Get pipeline status shared data and lock for status updates
         pipeline_status = await get_namespace_data("pipeline_status")
@@ -2914,6 +2919,57 @@ class LightRAG:
             # Mark that deletion operations have started
             deletion_operations_started = True
 
+            if delete_llm_cache and chunk_ids:
+                if not self.llm_response_cache:
+                    logger.info(
+                        "Skipping LLM cache collection for document %s because cache storage is unavailable",
+                        doc_id,
+                    )
+                elif not self.text_chunks:
+                    logger.info(
+                        "Skipping LLM cache collection for document %s because text chunk storage is unavailable",
+                        doc_id,
+                    )
+                else:
+                    try:
+                        chunk_data_list = await self.text_chunks.get_by_ids(
+                            list(chunk_ids)
+                        )
+                        seen_cache_ids: set[str] = set()
+                        for chunk_data in chunk_data_list:
+                            if not chunk_data or not isinstance(chunk_data, dict):
+                                continue
+                            cache_ids = chunk_data.get("llm_cache_list", [])
+                            if not isinstance(cache_ids, list):
+                                continue
+                            for cache_id in cache_ids:
+                                if (
+                                    isinstance(cache_id, str)
+                                    and cache_id
+                                    and cache_id not in seen_cache_ids
+                                ):
+                                    doc_llm_cache_ids.append(cache_id)
+                                    seen_cache_ids.add(cache_id)
+                        if doc_llm_cache_ids:
+                            logger.info(
+                                "Collected %d LLM cache entries for document %s",
+                                len(doc_llm_cache_ids),
+                                doc_id,
+                            )
+                        else:
+                            logger.info(
+                                "No LLM cache entries found for document %s", doc_id
+                            )
+                    except Exception as cache_collect_error:
+                        logger.error(
+                            "Failed to collect LLM cache ids for document %s: %s",
+                            doc_id,
+                            cache_collect_error,
+                        )
+                        raise Exception(
+                            f"Failed to collect LLM cache ids for document {doc_id}: {cache_collect_error}"
+                        ) from cache_collect_error
+
             # 4. Analyze entities and relationships that will be affected
             entities_to_delete = set()
             entities_to_rebuild = {}  # entity_name -> remaining chunk id list
@@ -3236,6 +3292,25 @@ class LightRAG:
                 logger.error(f"Failed to delete document and status: {e}")
                 raise Exception(f"Failed to delete document and status: {e}") from e
 
+            if delete_llm_cache and doc_llm_cache_ids and self.llm_response_cache:
+                try:
+                    await self.llm_response_cache.delete(doc_llm_cache_ids)
+                    cache_log_message = f"Successfully deleted {len(doc_llm_cache_ids)} LLM cache entries for document {doc_id}"
+                    logger.info(cache_log_message)
+                    async with pipeline_status_lock:
+                        pipeline_status["latest_message"] = cache_log_message
+                        pipeline_status["history_messages"].append(cache_log_message)
+                    log_message = cache_log_message
+                except Exception as cache_delete_error:
+                    logger.error(
+                        "Failed to delete LLM cache for document %s: %s",
+                        doc_id,
+                        cache_delete_error,
+                    )
+                    raise Exception(
+                        f"Failed to delete LLM cache for document {doc_id}: {cache_delete_error}"
+                    ) from cache_delete_error
+
             return DeletionResult(
                 status="success",
                 doc_id=doc_id,

lightrag_webui/src/api/lightrag.ts

@@ -618,9 +618,13 @@ export const clearCache = async (): Promise<{
   return response.data
 }
 
-export const deleteDocuments = async (docIds: string[], deleteFile: boolean = false): Promise<DeleteDocResponse> => {
+export const deleteDocuments = async (
+  docIds: string[],
+  deleteFile: boolean = false,
+  deleteLLMCache: boolean = false
+): Promise<DeleteDocResponse> => {
   const response = await axiosInstance.delete('/documents/delete_document', {
-    data: { doc_ids: docIds, delete_file: deleteFile }
+    data: { doc_ids: docIds, delete_file: deleteFile, delete_llm_cache: deleteLLMCache }
   })
   return response.data
 }

lightrag_webui/src/components/documents/DeleteDocumentsDialog.tsx

@@ -44,6 +44,7 @@ export default function DeleteDocumentsDialog({ selectedDocIds, onDocumentsDelet
   const [confirmText, setConfirmText] = useState('')
   const [deleteFile, setDeleteFile] = useState(false)
   const [isDeleting, setIsDeleting] = useState(false)
+  const [deleteLLMCache, setDeleteLLMCache] = useState(false)
   const isConfirmEnabled = confirmText.toLowerCase() === 'yes' && !isDeleting
 
   // Reset state when dialog closes
@@ -51,6 +52,7 @@ export default function DeleteDocumentsDialog({ selectedDocIds, onDocumentsDelet
     if (!open) {
       setConfirmText('')
       setDeleteFile(false)
+      setDeleteLLMCache(false)
       setIsDeleting(false)
     }
   }, [open])
@@ -60,7 +62,7 @@ export default function DeleteDocumentsDialog({ selectedDocIds, onDocumentsDelet
 
     setIsDeleting(true)
     try {
-      const result = await deleteDocuments(selectedDocIds, deleteFile)
+      const result = await deleteDocuments(selectedDocIds, deleteFile, deleteLLMCache)
 
       if (result.status === 'deletion_started') {
         toast.success(t('documentPanel.deleteDocuments.success', { count: selectedDocIds.length }))
@@ -94,7 +96,7 @@ export default function DeleteDocumentsDialog({ selectedDocIds, onDocumentsDelet
     } finally {
       setIsDeleting(false)
     }
-  }, [isConfirmEnabled, selectedDocIds, deleteFile, setOpen, t, onDocumentsDeleted])
+  }, [isConfirmEnabled, selectedDocIds, deleteFile, deleteLLMCache, setOpen, t, onDocumentsDeleted])
 
   return (
     <Dialog open={open} onOpenChange={setOpen}>
@@ -155,6 +157,20 @@ export default function DeleteDocumentsDialog({ selectedDocIds, onDocumentsDelet
               {t('documentPanel.deleteDocuments.deleteFileOption')}
             </Label>
           </div>
+
+          <div className="flex items-center space-x-2">
+            <input
+              type="checkbox"
+              id="delete-llm-cache"
+              checked={deleteLLMCache}
+              onChange={(e) => setDeleteLLMCache(e.target.checked)}
+              disabled={isDeleting}
+              className="h-4 w-4 text-red-600 focus:ring-red-500 border-gray-300 rounded"
+            />
+            <Label htmlFor="delete-llm-cache" className="text-sm font-medium cursor-pointer">
+              {t('documentPanel.deleteDocuments.deleteLLMCacheOption')}
+            </Label>
+          </div>
         </div>
 
         <DialogFooter>

lightrag_webui/src/locales/ar.json

@@ -70,6 +70,7 @@
       "confirmButton": "نعم",
       "deleteFileOption": "حذف الملفات المرفوعة أيضًا",
       "deleteFileTooltip": "حدد هذا الخيار لحذف الملفات المرفوعة المقابلة على الخادم أيضًا",
+      "deleteLLMCacheOption": "حذف ذاكرة LLM المؤقتة للاستخراج أيضًا",
       "success": "تم بدء تشغيل خط معالجة حذف المستندات بنجاح",
       "failed": "فشل حذف المستندات:\n{{message}}",
       "error": "فشل حذف المستندات:\n{{error}}",

lightrag_webui/src/locales/en.json

@@ -70,6 +70,7 @@
       "confirmButton": "YES",
       "deleteFileOption": "Also delete uploaded files",
       "deleteFileTooltip": "Check this option to also delete the corresponding uploaded files on the server",
+      "deleteLLMCacheOption": "Also delete extracted LLM cache",
       "success": "Document deletion pipeline started successfully",
       "failed": "Delete Documents Failed:\n{{message}}",
       "error": "Delete Documents Failed:\n{{error}}",

lightrag_webui/src/locales/fr.json

@@ -70,6 +70,7 @@
       "confirmButton": "OUI",
       "deleteFileOption": "Supprimer également les fichiers téléchargés",
       "deleteFileTooltip": "Cochez cette option pour supprimer également les fichiers téléchargés correspondants sur le serveur",
+      "deleteLLMCacheOption": "Supprimer également le cache LLM d'extraction",
       "success": "Pipeline de suppression de documents démarré avec succès",
       "failed": "Échec de la suppression des documents :\n{{message}}",
       "error": "Échec de la suppression des documents :\n{{error}}",

lightrag_webui/src/locales/zh.json

@@ -70,6 +70,7 @@
       "confirmButton": "确定",
       "deleteFileOption": "同时删除上传文件",
       "deleteFileTooltip": "选中此选项将同时删除服务器上对应的上传文件",
+      "deleteLLMCacheOption": "同时删除实体关系抽取 LLM 缓存",
       "success": "文档删除流水线启动成功",
       "failed": "删除文档失败：\n{{message}}",
       "error": "删除文档失败：\n{{error}}",

lightrag_webui/src/locales/zh_TW.json

@@ -70,6 +70,7 @@
       "confirmButton": "確定",
       "deleteFileOption": "同時刪除上傳檔案",
       "deleteFileTooltip": "選取此選項將同時刪除伺服器上對應的上傳檔案",
+      "deleteLLMCacheOption": "同時刪除實體關係擷取 LLM 快取",
       "success": "文件刪除流水線啟動成功",
       "failed": "刪除文件失敗：\n{{message}}",
       "error": "刪除文件失敗：\n{{error}}",