OpenRAG 0.2.5__tar.gz → 0.3.0__tar.gz

{openrag-0.2.5/src/openrag.egg-info → openrag-0.3.0}/PKG-INFO

@@ -1,7 +1,17 @@
 Metadata-Version: 2.4
 Name: openrag
-Version: 0.2.5
+Version: 0.3.0
 Summary: OpenRAG is a comprehensive Retrieval-Augmented Generation platform that enables intelligent document search and AI-powered conversations.
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Requires-Python: >=3.13
 Description-Content-Type: text/markdown
 License-File: LICENSE
@@ -48,7 +58,7 @@ Dynamic: license-file
 [![GitHub stars](https://img.shields.io/github/stars/langflow-ai/openrag?style=social)](https://github.com/langflow-ai/openrag/stargazers)
 [![GitHub forks](https://img.shields.io/github/forks/langflow-ai/openrag?style=social)](https://github.com/langflow-ai/openrag/network/members)
 
-[![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/langflow-ai/openrag)
+[![Documentation](https://img.shields.io/badge/Documentation-773eff)](https://docs.openr.ag) [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/langflow-ai/openrag)
 
 </div>
 
@@ -58,6 +68,8 @@ OpenRAG is a comprehensive Retrieval-Augmented Generation platform that enables
 
 Users can upload, process, and query documents through a chat interface backed by large language models and semantic search capabilities. The system utilizes Langflow for document ingestion, retrieval workflows, and intelligent nudges, providing a seamless RAG experience. 
 
+Check out the [documentation](https://docs.openr.ag/) or get started with the [quickstart](https://docs.openr.ag/quickstart).
+
 Built with [Starlette](https://github.com/Kludex/starlette) and [Next.js](https://github.com/vercel/next.js). 
 Powered by [OpenSearch](https://github.com/opensearch-project/OpenSearch), [Langflow](https://github.com/langflow-ai/langflow), and [Docling](https://github.com/docling-project/docling).
 

{openrag-0.2.5 → openrag-0.3.0}/README.md

@@ -18,7 +18,7 @@
 [![GitHub stars](https://img.shields.io/github/stars/langflow-ai/openrag?style=social)](https://github.com/langflow-ai/openrag/stargazers)
 [![GitHub forks](https://img.shields.io/github/forks/langflow-ai/openrag?style=social)](https://github.com/langflow-ai/openrag/network/members)
 
-[![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/langflow-ai/openrag)
+[![Documentation](https://img.shields.io/badge/Documentation-773eff)](https://docs.openr.ag) [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/langflow-ai/openrag)
 
 </div>
 
@@ -28,6 +28,8 @@ OpenRAG is a comprehensive Retrieval-Augmented Generation platform that enables
 
 Users can upload, process, and query documents through a chat interface backed by large language models and semantic search capabilities. The system utilizes Langflow for document ingestion, retrieval workflows, and intelligent nudges, providing a seamless RAG experience. 
 
+Check out the [documentation](https://docs.openr.ag/) or get started with the [quickstart](https://docs.openr.ag/quickstart).
+
 Built with [Starlette](https://github.com/Kludex/starlette) and [Next.js](https://github.com/vercel/next.js). 
 Powered by [OpenSearch](https://github.com/opensearch-project/OpenSearch), [Langflow](https://github.com/langflow-ai/langflow), and [Docling](https://github.com/docling-project/docling).
 

{openrag-0.2.5 → openrag-0.3.0}/pyproject.toml

@@ -4,10 +4,22 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "openrag"
-version = "0.2.5"
+version = "0.3.0"
 description = "OpenRAG is a comprehensive Retrieval-Augmented Generation platform that enables intelligent document search and AI-powered conversations."
 readme = "README.md"
 requires-python = ">=3.13"
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Environment :: Console",
+    "Intended Audience :: Developers",
+    "Intended Audience :: Science/Research",
+    "License :: OSI Approved :: Apache Software License",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3 :: Only",
+    "Programming Language :: Python :: 3.13",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+    "Topic :: Software Development :: Libraries :: Python Modules",
+]
 dependencies = [
     "agentd>=0.2.2",
     "aiofiles>=24.1.0",

{openrag-0.2.5 → openrag-0.3.0}/src/agent.py

@@ -135,14 +135,13 @@ async def async_response_stream(
 
         full_response = ""
         chunk_count = 0
-        detected_tool_call = False  # Track if we've detected a tool call
+        detected_tool_call = False
         async for chunk in response:
             chunk_count += 1
             logger.debug(
                 "Stream chunk received", chunk_count=chunk_count, chunk=str(chunk)
             )
 
-            # Yield the raw event as JSON for the UI to process
             import json
 
             # Also extract text content for logging
@@ -186,7 +185,7 @@ async def async_response_stream(
                 if isinstance(chunk_data, dict):
                     # Check for any fields that might indicate tool usage
                     potential_tool_fields = {
-                        k: v for k, v in chunk_data.items() 
+                        k: v for k, v in chunk_data.items()
                         if any(keyword in str(k).lower() for keyword in ['tool', 'call', 'retrieval', 'function', 'result', 'output'])
                     }
                     if potential_tool_fields:
@@ -235,8 +234,8 @@ async def async_response_stream(
                                 "tool_name": "Retrieval",
                                 "status": "completed",
                                 "inputs": {"implicit": True, "backend_detected": True},
-                                "results": chunk_data.get('results') or chunk_data.get('outputs') or 
-                                         chunk_data.get('retrieved_documents') or 
+                                "results": chunk_data.get('results') or chunk_data.get('outputs') or
+                                         chunk_data.get('retrieved_documents') or
                                          chunk_data.get('retrieval_results') or []
                             }
                         }
@@ -625,8 +624,25 @@ async def async_langflow_chat(
             message_count=len(conversation_state["messages"]),
         )
 
+    # Extract sources from retrieval tool calls in the response
+    sources = []
+    if hasattr(response_obj, "output") and response_obj.output:
+        for output_item in response_obj.output:
+            item_type = getattr(output_item, "type", None)
+            if item_type in ("tool_call", "retrieval_call"):
+                for result in getattr(output_item, "results", None) or []:
+                    rd = result.model_dump() if hasattr(result, "model_dump") else (result if isinstance(result, dict) else {})
+                    if "text" in rd:
+                        sources.append({
+                            "filename": rd.get("filename", ""),
+                            "text": rd.get("text", ""),
+                            "score": rd.get("score", 0),
+                            "page": rd.get("page"),
+                            "mimetype": rd.get("mimetype"),
+                        })
+
     if not store_conversation:
-        return response_text, response_id
+        return response_text, response_id, sources
 
     # Store the conversation thread with its response_id
     if response_id:
@@ -662,7 +678,7 @@ async def async_langflow_chat(
     else:
         logger.warning("No response_id received from langflow, conversation not stored")
 
-    return response_text, response_id
+    return response_text, response_id, sources
 
 
 # Async langflow function with conversation storage (streaming)
@@ -698,47 +714,55 @@ async def async_langflow_chat_stream(
     response_id = None
     usage_data = None
     collected_chunks = []  # Store all chunks for function call data
+    error_occurred = False
 
-    async for chunk in async_stream(
-        langflow_client,
-        prompt,
-        flow_id,
-        extra_headers=extra_headers,
-        previous_response_id=previous_response_id,
-        log_prefix="langflow",
-    ):
-        # Extract text content to build full response for history
-        try:
-            import json
-
-            chunk_data = json.loads(chunk.decode("utf-8"))
-            collected_chunks.append(chunk_data)  # Collect all chunk data
-
-            if "delta" in chunk_data and "content" in chunk_data["delta"]:
-                full_response += chunk_data["delta"]["content"]
-            # Extract response_id from chunk
-            if "id" in chunk_data:
-                response_id = chunk_data["id"]
-            elif "response_id" in chunk_data:
-                response_id = chunk_data["response_id"]
-            # Capture usage from response.completed event
-            if chunk_data.get("type") == "response.completed":
-                response_obj = chunk_data.get("response", {})
-                usage_data = response_obj.get("usage")
-        except:
-            pass
-        yield chunk
+    try:
+        async for chunk in async_stream(
+            langflow_client,
+            prompt,
+            flow_id,
+            extra_headers=extra_headers,
+            previous_response_id=previous_response_id,
+            log_prefix="langflow",
+        ):
+            # Extract text content to build full response for history
+            try:
+                import json
+
+                chunk_data = json.loads(chunk.decode("utf-8"))
+                collected_chunks.append(chunk_data)  # Collect all chunk data
+
+                if "delta" in chunk_data and "content" in chunk_data["delta"]:
+                    full_response += chunk_data["delta"]["content"]
+                # Extract response_id from chunk
+                if "id" in chunk_data:
+                    response_id = chunk_data["id"]
+                elif "response_id" in chunk_data:
+                    response_id = chunk_data["response_id"]
+                
+                # Check for error status
+                if chunk_data.get("finish_reason") == "error" or chunk_data.get("status") == "failed":
+                    error_occurred = True
+                    logger.error("Error detected in Langflow stream chunk")
+                # Capture usage from response.completed event
+                if chunk_data.get("type") == "response.completed":
+                    response_obj = chunk_data.get("response", {})
+                    usage_data = response_obj.get("usage")
+            except:
+                pass
+            yield chunk
 
-    # Add the complete assistant response to message history with response_id, timestamp, and function call data
-    if full_response:
-        assistant_message = {
-            "role": "assistant",
-            "content": full_response,
-            "response_id": response_id,
-            "timestamp": datetime.now(),
-            "chunks": collected_chunks,  # Store complete chunk data for function calls
-        }
-        # Store usage data if available (from response.completed event)
+        # Add the complete assistant response to message history with response_id, timestamp, and function call data
+        if full_response:
+            assistant_message = {
+                "role": "assistant",
+                "content": full_response,
+                "response_id": response_id,
+                "timestamp": datetime.now(),
+                "chunks": collected_chunks,  # Store complete chunk data for function calls
+                "error": error_occurred,  # Mark if this was an error response
+            }
+            # Store usage data if available (from response.completed event)
         if usage_data:
             assistant_message["response_data"] = {"usage": usage_data}
         conversation_state["messages"].append(assistant_message)
@@ -760,6 +784,34 @@ async def async_langflow_chat_stream(
             logger.debug(
                 f"Stored langflow conversation thread for user {user_id} with response_id: {response_id}"
             )
+    except Exception as e:
+        # Log the error
+        logger.error(f"Error in langflow chat stream: {e}", exc_info=True)
+        error_occurred = True
+        
+        # Store error message in conversation history so it persists
+        error_message = {
+            "role": "assistant",
+            "content": f"Sorry, I encountered an error: {str(e)}",
+            "timestamp": datetime.now(),
+            "error": True,
+        }
+        conversation_state["messages"].append(error_message)
+        
+        # Try to store the conversation with error message
+        # Use a temporary response_id if we don't have one
+        if not response_id:
+            response_id = f"error_{user_id}_{int(datetime.now().timestamp())}"
+        
+        try:
+            conversation_state["last_activity"] = datetime.now()
+            await store_conversation_thread(user_id, response_id, conversation_state)
+            logger.debug(f"Stored conversation with error for user {user_id}")
+        except Exception as store_error:
+            logger.error(f"Failed to store error conversation: {store_error}")
+        
+        # Re-raise the exception so it propagates to the API layer
+        raise
 
 
 async def delete_user_conversation(user_id: str, response_id: str) -> bool:

{openrag-0.2.5 → openrag-0.3.0}/src/api/settings.py

@@ -55,6 +55,10 @@ class SettingsUpdateBody(BaseModel):
     watsonx_endpoint: Optional[str] = Field(None, min_length=1)
     watsonx_project_id: Optional[str] = Field(None, min_length=1)
     ollama_endpoint: Optional[str] = Field(None, min_length=1)
+    remove_ollama_config: Optional[bool] = None
+    remove_openai_config: Optional[bool] = None
+    remove_anthropic_config: Optional[bool] = None
+    remove_watsonx_config: Optional[bool] = None
 
 
 class OnboardingBody(BaseModel):
@@ -168,6 +172,7 @@ class OnboardingResponse(BaseModel):
     edited: bool
     sample_data_ingested: bool
     openrag_docs_filter_id: Optional[str] = None
+    task_id: Optional[str] = None
 
 class DoclingConfig(BaseModel):
     do_ocr: bool
@@ -360,6 +365,22 @@ async def get_settings(
         )
 
 
+def _first_configured_llm_provider(config, excluding: str) -> str:
+    """Return the first configured LLM provider that isn't `excluding`."""
+    for p in ["openai", "anthropic", "watsonx", "ollama"]:
+        if p != excluding and getattr(config.providers, p).configured:
+            return p
+    return "openai"
+
+
+def _first_configured_embedding_provider(config, excluding: str) -> str:
+    """Return the first configured embedding provider (openai/watsonx/ollama) that isn't `excluding`."""
+    for p in ["openai", "watsonx", "ollama"]:
+        if p != excluding and getattr(config.providers, p).configured:
+            return p
+    return "openai"
+
+
 async def update_settings(
     body: SettingsUpdateBody,
     session_manager=Depends(get_session_manager),
@@ -672,6 +693,99 @@ async def update_settings(
             config_updated = True
             provider_updated = True
 
+        if body.remove_ollama_config:
+            other_providers_configured = (
+                current_config.providers.openai.configured
+                or current_config.providers.anthropic.configured
+                or current_config.providers.watsonx.configured
+            )
+            if not other_providers_configured:
+                return JSONResponse(
+                    {"error": "Cannot remove Ollama configuration: configure another model provider first."},
+                    status_code=400,
+                )
+            current_config.providers.ollama.endpoint = ""
+            current_config.providers.ollama.configured = False
+            if current_config.agent.llm_provider == "ollama":
+                current_config.agent.llm_provider = _first_configured_llm_provider(current_config, "ollama")
+                current_config.agent.llm_model = ""
+            if current_config.knowledge.embedding_provider == "ollama":
+                current_config.knowledge.embedding_provider = _first_configured_embedding_provider(current_config, "ollama")
+                current_config.knowledge.embedding_model = ""
+            config_updated = True
+            provider_updated = True
+
+        if body.remove_openai_config:
+            other_providers_configured = (
+                current_config.providers.anthropic.configured
+                or current_config.providers.watsonx.configured
+                or current_config.providers.ollama.configured
+            )
+            if not other_providers_configured:
+                return JSONResponse(
+                    {"error": "Cannot remove OpenAI configuration: configure another model provider first."},
+                    status_code=400,
+                )
+            current_config.providers.openai.api_key = ""
+            current_config.providers.openai.configured = False
+            if current_config.agent.llm_provider == "openai":
+                fb = _first_configured_llm_provider(current_config, "openai")
+                current_config.agent.llm_provider = fb
+                current_config.agent.llm_model = ""
+            if current_config.knowledge.embedding_provider == "openai":
+                fb = _first_configured_embedding_provider(current_config, "openai")
+                current_config.knowledge.embedding_provider = fb
+                current_config.knowledge.embedding_model = ""
+            config_updated = True
+            provider_updated = True
+
+        if body.remove_anthropic_config:
+            other_providers_configured = (
+                current_config.providers.openai.configured
+                or current_config.providers.watsonx.configured
+                or current_config.providers.ollama.configured
+            )
+            if not other_providers_configured:
+                return JSONResponse(
+                    {"error": "Cannot remove Anthropic configuration: configure another model provider first."},
+                    status_code=400,
+                )
+            current_config.providers.anthropic.api_key = ""
+            current_config.providers.anthropic.configured = False
+            if current_config.agent.llm_provider == "anthropic":
+                fb = _first_configured_llm_provider(current_config, "anthropic")
+                current_config.agent.llm_provider = fb
+                current_config.agent.llm_model = ""
+            # Anthropic is not a valid embedding provider; no embedding reset needed
+            config_updated = True
+            provider_updated = True
+
+        if body.remove_watsonx_config:
+            other_providers_configured = (
+                current_config.providers.openai.configured
+                or current_config.providers.anthropic.configured
+                or current_config.providers.ollama.configured
+            )
+            if not other_providers_configured:
+                return JSONResponse(
+                    {"error": "Cannot remove IBM watsonx.ai configuration: configure another model provider first."},
+                    status_code=400,
+                )
+            current_config.providers.watsonx.api_key = ""
+            current_config.providers.watsonx.endpoint = ""
+            current_config.providers.watsonx.project_id = ""
+            current_config.providers.watsonx.configured = False
+            if current_config.agent.llm_provider == "watsonx":
+                fb = _first_configured_llm_provider(current_config, "watsonx")
+                current_config.agent.llm_provider = fb
+                current_config.agent.llm_model = ""
+            if current_config.knowledge.embedding_provider == "watsonx":
+                fb = _first_configured_embedding_provider(current_config, "watsonx")
+                current_config.knowledge.embedding_provider = fb
+                current_config.knowledge.embedding_model = ""
+            config_updated = True
+            provider_updated = True
+
         if provider_updated:
             await TelemetryClient.send_event(
                 Category.SETTINGS_OPERATIONS,
@@ -692,7 +806,7 @@ async def update_settings(
         # Update Langflow global variables and model values if provider settings changed
         await clients.refresh_patched_client()
 
-        if should_validate:
+        if should_validate or provider_updated:
             try:
                 flows_service = _get_flows_service()
 
@@ -705,8 +819,8 @@ async def update_settings(
                         current_config, session_manager
                     )
 
-                # Update model values if provider or model changed
-                if body.llm_provider is not None or body.llm_model is not None or body.embedding_provider is not None or body.embedding_model is not None:
+                # Update model values if provider or model changed (including removals that trigger fallback)
+                if body.llm_provider is not None or body.llm_model is not None or body.embedding_provider is not None or body.embedding_model is not None or provider_updated:
                     await _update_langflow_model_values(current_config, flows_service)
 
             except Exception as e:
@@ -975,6 +1089,8 @@ async def onboarding(
             )
             raise
 
+        task_id = None
+
         # Initialize the OpenSearch index if embedding model is configured
         if body.embedding_model or body.embedding_provider:
             try:
@@ -1004,7 +1120,7 @@ async def onboarding(
                     # Import the function here to avoid circular imports
                     from main import ingest_default_documents_when_ready
 
-                    await ingest_default_documents_when_ready(
+                    task_id = await ingest_default_documents_when_ready(
                         document_service,
                         task_service,
                         langflow_file_service,
@@ -1094,6 +1210,7 @@ async def onboarding(
             edited=True,  # Confirm that config is now marked as edited
             sample_data_ingested=should_ingest_sample_data,
             openrag_docs_filter_id=openrag_docs_filter_id,
+            task_id=task_id,
         )
 
     except Exception as e:
@@ -1121,7 +1238,7 @@ async def _create_openrag_docs_filter(
         return None
 
     # Get JWT token
-        jwt_token = user.jwt_token
+    jwt_token = user.jwt_token
 
     # In no-auth mode, set owner to None so filter is visible to all users
     # In auth mode, use the actual user as owner
@@ -1192,6 +1309,12 @@ async def _update_langflow_global_variables(config):
             )
             logger.info("Set WATSONX_PROJECT_ID global variable in Langflow")
 
+        if config.providers.watsonx.endpoint:
+            await clients._create_langflow_global_variable(
+                "WATSONX_URL", config.providers.watsonx.endpoint, modify=True
+            )
+            logger.info("Set WATSONX_URL global variable in Langflow")
+
         # OpenAI global variables
         if config.providers.openai.api_key:
             await clients._create_langflow_global_variable(
@@ -1208,11 +1331,19 @@ async def _update_langflow_global_variables(config):
 
         # Ollama global variables
         if config.providers.ollama.endpoint:
-            endpoint = transform_localhost_url(config.providers.ollama.endpoint)
-            await clients._create_langflow_global_variable(
-                "OLLAMA_BASE_URL", endpoint, modify=True
-            )
-            logger.info("Set OLLAMA_BASE_URL global variable in Langflow")
+
+            try:
+                endpoint = transform_localhost_url(config.providers.ollama.endpoint, is_langflow=True, is_podman=True)
+                await clients._create_langflow_global_variable(
+                    "OLLAMA_BASE_URL", endpoint, modify=True
+                )
+                logger.info("Set OLLAMA_BASE_URL global variable in Langflow (Podman)")
+            except Exception:
+                endpoint = transform_localhost_url(config.providers.ollama.endpoint, is_langflow=True, is_podman=False)
+                await clients._create_langflow_global_variable(
+                    "OLLAMA_BASE_URL", endpoint, modify=True
+                )   
+                logger.info("Set OLLAMA_BASE_URL global variable in Langflow (Docker)")
 
         if config.knowledge.embedding_model:
             await clients._create_langflow_global_variable(
@@ -1269,13 +1400,10 @@ async def _update_langflow_model_values(config, flows_service):
     try:
         # Update LLM model values
         llm_provider = config.agent.llm_provider.lower()
-        llm_provider_config = config.get_llm_provider_config()
-        llm_endpoint = getattr(llm_provider_config, "endpoint", None)
 
         await flows_service.change_langflow_model_value(
             llm_provider,
             llm_model=config.agent.llm_model,
-            endpoint=llm_endpoint,
         )
         logger.info(
             f"Successfully updated Langflow flows for LLM provider {llm_provider}"
@@ -1283,13 +1411,10 @@ async def _update_langflow_model_values(config, flows_service):
 
         # Update embedding model values
         embedding_provider = config.knowledge.embedding_provider.lower()
-        embedding_provider_config = config.get_embedding_provider_config()
-        embedding_endpoint = getattr(embedding_provider_config, "endpoint", None)
 
         await flows_service.change_langflow_model_value(
             embedding_provider,
             embedding_model=config.knowledge.embedding_model,
-            endpoint=embedding_endpoint,
         )
         logger.info(
             f"Successfully updated Langflow flows for embedding provider {embedding_provider}"
@@ -1517,6 +1642,7 @@ async def rollback_onboarding(
 
         # Mark config as not edited so user can go through onboarding again
         current_config.edited = False
+        current_config.onboarding.current_step = 0
 
         # Save the rolled back configuration manually to avoid save_config_file setting edited=True
         try:

{openrag-0.2.5 → openrag-0.3.0}/src/api/v1/chat.py

@@ -28,6 +28,21 @@ class ChatV1Body(BaseModel):
     filter_id: Optional[str] = None
 
 
+def _extract_sources(item: dict) -> list[dict]:
+    """Extract sources from a retrieval tool call item."""
+    sources = []
+    for result in item.get("results", []):
+        if isinstance(result, dict) and "text" in result:
+            sources.append({
+                "filename": result.get("filename", ""),
+                "text": result.get("text", ""),
+                "score": result.get("score", 0),
+                "page": result.get("page"),
+                "mimetype": result.get("mimetype"),
+            })
+    return sources
+
+
 async def _transform_stream_to_sse(raw_stream, chat_id_container: dict):
     """Transform raw Langflow streaming format to clean SSE events for v1 API."""
     full_text = ""
@@ -64,6 +79,13 @@ async def _transform_stream_to_sse(raw_stream, chat_id_container: dict):
                 full_text += delta_text
                 yield f"data: {json.dumps({'type': 'content', 'delta': delta_text})}\n\n"
 
+            # Emit sources from retrieval tool calls
+            item = chunk_data.get("item", {})
+            if item.get("type") in ("retrieval_call", "tool_call") and item.get("results"):
+                sources = _extract_sources(item)
+                if sources:
+                    yield f"data: {json.dumps({'type': 'sources', 'sources': sources})}\n\n"
+
             if not chat_id:
                 chat_id = chunk_data.get("id") or chunk_data.get("response_id")
 

{openrag-0.2.5 → openrag-0.3.0}/src/config/settings.py

@@ -697,6 +697,7 @@ class AppClients:
                 variable_name=name,
                 error=str(e),
             )
+            raise e
 
     async def _update_langflow_global_variable(self, name: str, value: str):
         """Update an existing global variable in Langflow via API"""

{openrag-0.2.5 → openrag-0.3.0}/src/dependencies.py

@@ -162,6 +162,7 @@ def get_optional_user(
 async def get_api_key_user_async(
     request: Request,
     api_key_service=Depends(get_api_key_service),
+    session_manager=Depends(get_session_manager),
 ) -> User:
     """
     Async dependency: require API key authentication.
@@ -208,9 +209,12 @@ async def get_api_key_user_async(
         provider="api_key",
     )
 
-    # API Key users don't typically have a JWT for OpenSearch OIDC, 
-    # but we can try to get an effective one if needed
-    user_with_token = dataclasses.replace(user, jwt_token=None)
+    # Register the API key user so get_effective_jwt_token can find them
+    if user.user_id not in session_manager.users:
+        session_manager.users[user.user_id] = user
+
+    effective_token = session_manager.get_effective_jwt_token(user.user_id, None)
+    user_with_token = dataclasses.replace(user, jwt_token=effective_token)
 
     request.state.user = user_with_token
     request.state.api_key_id = user_info["key_id"]