chore: use minimal langgraph app to read and delete checkpointer history

modules/features/chatBot/service.py

@@ -22,6 +22,7 @@ from modules.datamodels.datamodelChatbot import (
 from modules.datamodels.datamodelUam import User
 
 from langchain_core.messages import HumanMessage, AIMessage, BaseMessage
+from langgraph.graph import StateGraph, MessagesState, START, END
 from modules.shared.configuration import APP_CONFIG
 
 logger = logging.getLogger(__name__)
@@ -492,22 +493,65 @@ async def post_message_stream(
         )
 
 
+# Module-level singleton for minimal app used to read thread state
+_MINIMAL_APP = None
+
+
+def _build_minimal_app(*, checkpointer):
+    """Build a minimal LangGraph app for reading thread state.
+
+    This creates a valid graph with a no-op node that we never actually run.
+    LangGraph requires a valid graph structure (with edges from START) to compile,
+    even though we only use it to call aget_state() to read from the checkpointer.
+
+    Args:
+        checkpointer: The checkpointer to attach to the graph.
+
+    Returns:
+        A compiled StateGraph that can be used to read thread state.
+    """
+    graph = StateGraph(MessagesState)
+
+    # No-op node that returns the state unchanged
+    def noop(state: dict) -> dict:
+        return state
+
+    graph.add_node("noop", noop)
+    graph.add_edge(START, "noop")
+    graph.add_edge("noop", END)
+
+    return graph.compile(checkpointer=checkpointer)
+
+
+def _get_minimal_app():
+    """Get the module-level singleton minimal app.
+
+    Returns:
+        The cached minimal app, building it on first access.
+    """
+    global _MINIMAL_APP
+    if _MINIMAL_APP is None:
+        _MINIMAL_APP = _build_minimal_app(checkpointer=get_checkpointer())
+    return _MINIMAL_APP
+
+
 async def get_thread_messages_from_langgraph(
     *,
     thread_id: str,
-    app,
 ) -> List[dict]:
     """Retrieve and format messages from LangGraph checkpointer.
 
     Args:
         thread_id: The unique identifier for the chat thread.
-        app: The compiled LangGraph app with checkpointer.
 
     Returns:
-        List of message dicts with role, content, and timestamp.
+        List of message dicts with role and content.
     """
     ROLE_MAP = {"human": "user", "ai": "assistant"}
 
+    # Get the minimal app (singleton, built once)
+    app = _get_minimal_app()
+
     cfg = {"configurable": {"thread_id": thread_id}}
     state = await app.aget_state(cfg)
 
@@ -562,41 +606,8 @@ async def get_thread_detail_for_user(
     result = await session.execute(stmt)
     thread_mapping = result.scalar_one()
 
-    # Build the chatbot app to access LangGraph state
-    # Use same approach as post_message for consistency
-    tool_ids = permissions.get_chatbot_tools(user_id=user.id)
-    if not tool_ids:
-        raise ValueError("User does not have permission to use any chatbot tools")
-
-    model_name = permissions.get_chatbot_model(user_id=user.id)
-    system_prompt = permissions.get_system_prompt(user_id=user.id)
-
-    # Get tools from registry
-    registry = get_registry()
-    tools = registry.get_tool_instances(tool_ids=tool_ids)
-
-    # Get model and checkpointer
-    model = get_langchain_model(model_name=model_name)
-    checkpointer = get_checkpointer()
-
-    # Get context window size from config
-    context_window_size = int(
-        APP_CONFIG.get("CHATBOT_CONTEXT_WINDOW_TOKEN_SIZE", 100000)
-    )
-
-    # Create chatbot instance
-    chatbot = await Chatbot.create(
-        model=model,
-        memory=checkpointer,
-        system_prompt=system_prompt,
-        tools=tools,
-        context_window_size=context_window_size,
-    )
-
-    # Get messages from LangGraph checkpointer
-    message_dicts = await get_thread_messages_from_langgraph(
-        thread_id=thread_id, app=chatbot.app
-    )
+    # Get messages from LangGraph checkpointer (optimized - no full chatbot needed)
+    message_dicts = await get_thread_messages_from_langgraph(thread_id=thread_id)
 
     # Convert to MessageItem objects
     messages = [MessageItem(**m) for m in message_dicts]
@@ -638,39 +649,10 @@ async def delete_thread_for_user(
         thread_id=thread_id, user=user, session=session
     )
 
-    # Build the chatbot app to access the checkpointer
-    tool_ids = permissions.get_chatbot_tools(user_id=user.id)
-    if not tool_ids:
-        raise ValueError("User does not have permission to use any chatbot tools")
-
-    model_name = permissions.get_chatbot_model(user_id=user.id)
-    system_prompt = permissions.get_system_prompt(user_id=user.id)
-
-    # Get tools from registry
-    registry = get_registry()
-    tools = registry.get_tool_instances(tool_ids=tool_ids)
-
-    # Get model and checkpointer
-    model = get_langchain_model(model_name=model_name)
+    # Delete from LangGraph checkpointer (optimized - no app/tools/model needed)
     checkpointer = get_checkpointer()
-
-    # Get context window size from config
-    context_window_size = int(
-        APP_CONFIG.get("CHATBOT_CONTEXT_WINDOW_TOKEN_SIZE", 100000)
-    )
-
-    # Create chatbot instance
-    chatbot = await Chatbot.create(
-        model=model,
-        memory=checkpointer,
-        system_prompt=system_prompt,
-        tools=tools,
-        context_window_size=context_window_size,
-    )
-
-    # Delete from LangGraph checkpointer
     try:
-        await chatbot.app.checkpointer.adelete_thread(thread_id)
+        await checkpointer.adelete_thread(thread_id)
         logger.info(f"Deleted thread {thread_id} from LangGraph checkpointer")
     except Exception as e:
         logger.error(