Multiple States, 1 Conditional Edge, but 2 Results...

[ahmed33033]

Hi there!

I was learning how to work with graphs in LangGraph, and I stumbled across a perplexing issue while working with conditional edges.

Graph Summary

I have 3 nodes. The first node, node_1, is only connected to node_2, while node_2 is connected to both node_3 and END via a conditional edge.

States Summary

I have 3 different states. Both PersonState and PurchasesState has unique attributes (seen below). On the other hand, OverallState combines the aforementioned states.

The Issue

node_2 issues an update to the state, which is not reflected correctly in the routing function (that controls the conditional edge).

Code

from operator import add
from typing import TypedDict, Annotated
from langgraph.graph import StateGraph, START, END

class PersonState(TypedDict):
    name: str
    money: Annotated[int, add]
    
class PurchasesState(TypedDict):
    purchases: Annotated[list[str], add]
    total_cost: Annotated[int, add]

class OverallState(PersonState, PurchasesState):
    pass


# node_1 operation. 
def pay_salary(state: PersonState) -> OverallState:
    print(f"Node 1: {state}")
    return {"money": 1000}

# node_2 operation. 
def subtract_expenses(state: PurchasesState) -> OverallState:
    print(f"Node 2: {state}")
    return {'money': -100, "purchases": ["new iphone"], 'total_cost': 100}

# node_3 operation. 
def add_assistance(state: OverallState) -> OverallState:
    print(f"Node 3: {state}")
    return {"money": 100}

# Routing Function. 
def check_if_poor(state: OverallState):
    print(f"Routing function: {state}")
    # if poor, add asssitance
    if (state['money'] < 100): return "node_3"
    else: return END


graph = StateGraph(OverallState, input = PersonState)
graph.add_edge(START, 'node_1')
graph.add_node('node_1', pay_salary)
graph.add_edge('node_1', 'node_2')

graph.add_node('node_2', subtract_expenses)
graph.add_conditional_edges('node_2', check_if_poor)

graph.add_node('node_3', add_assistance)
graph.add_edge("node_3", END)

workflow = graph.compile()
output_dict = workflow.invoke({"name": "Ahmed"})
print(f"Final output: {output_dict}")

Code Output

Node 1: {'name': 'Ahmed', 'money': 0}
Node 2: {'purchases': [], 'total_cost': 0}
Routing function: {'money': -100, 'purchases': ['new iphone'], 'total_cost': 100}
Node 3: {'name': 'Ahmed', 'money': 900, 'purchases': ['new iphone'], 'total_cost': 100}
Final output: {'name': 'Ahmed', 'money': 1000, 'purchases': ['new iphone'], 'total_cost': 100}

The output line that I can't understand

Routing function: {'money': -100, 'purchases': ['new iphone'], 'total_cost': 100}

How does money have a value of -100? In the method pay_salary, the money attribute is set to 1000. Then, in the subtract_expenses method, I subtract 100 from it. So, the money attribute should have a value of 900, not -100.

The line that is causing the weird output (after debugging)

def subtract_expenses(state: PurchasesState) -> OverallState:

How is it causing the issue?

When I change the inputted state from PurchasesState to OverallState, it works as expected, i.e. the output printed in the routing function check_if_poor is:

Routing function: {'money': 900, 'purchases': ['new iphone'], 'total_cost': 100, 'name': 'Ahmed'}

What's my question?

Why does changing the type of the inputted state in the function subtract_expenses lead me to receive different values for the money attribute in the routing function check_if_poor?

Thanks in advance!

P.S. Please don't mind how inefficient the graph setup is for the problem I'm trying to solve. I had simplified my original code a good bit of my code to make it easier to pinpoint what the issue was.

[Okeefe-Niemann]

Not sure if you found an answer in the past month, but returning dictionaries in a node will overwrite that field in the state.

So, by returning {'money': -100, "purchases": ["new iphone"], 'total_cost': 100}, you are updating the "money" field in your state to be -100, the "purchases" field in your state to be a single entry list containing "new iphone", and "total_cost" field to 100.

if you are wanting to add these values to the state object, the correct syntax should be:

{'money': state.get("money") -100, "purchases": state.get("purchases").append("new iphone"), 'total_cost': state.get('total_cost') + 100}

relevant question (see response): #1788

[ahmed33033]

First off, thanks for engaging with the discussion post! And no, I still haven't found an answer.

To reply to your comment, no, returning a dictionary does not always overwrite a field in the graph's state. You can see that in my states' definitions, I'm using an Annotated type hint, which specifies an add function for fields money, purchases and total_cost. This means that when I return an update to those fields, it will utilize the python add function to combine the old and new values together. The add function works as you would expect with numbers. For lists, it "adds" them together by extending the old list with the values of the new list. The technical term for the add function is a reducer, which you can find out more about in the langgraph glossary.

You can clearly see the reducers in action when I change the input state of subtract_expenses from PurchasesState to OverallState, as outlined in my discussion post.

The issue is that the routing function check_if_poor is not receiving the updated state, even though the node add_assistance, which comes right after the routing function, receives the updated state like you would expect. As aforementioned, this issue is somehow solved by changing the input state of subtract_expenses.

I'm creating an issue from this post since the problem still exists in the updated version of langgraph.

[AnsenIO]

hi, running your code give me a different output from what you reported, probably it was a bug that got fixed meantime or you fixed the code already.

here is the output :

Node 1: {'name': 'Ahmed', 'money': 0}
Node 2: {'purchases': [], 'total_cost': 0}
Routing function: {'name': 'Ahmed', 'money': 900, 'purchases': ['new iphone'], 'total_cost': 100}
Final output: {'name': 'Ahmed', 'money': 900, 'purchases': ['new iphone'], 'total_cost': 100}

[ahmed33033]

Sorry for the late response! This bug was detailed in Issue #2504 and later fixed for LangGraph v. 0.36. Thanks for reminding me to update this discussion!