breakpoints.ipynb에서는 breakpoint의 개념과 사용예를 보여줍니다. 이 노트북의 원본은 langchain-breakpoints입니다.
먼저 간단한 케이스에 대한 breakpoint 예제 입니다.
from typing import TypedDict
from langgraph.graph import StateGraph, START, END
from langgraph.checkpoint.memory import MemorySaver
class State(TypedDict):
input: str
def step_1(state):
print("---Step 1---")
pass
def step_2(state):
print("---Step 2---")
pass
def step_3(state):
print("---Step 3---")
pass
workflow = StateGraph(State)
workflow.add_node("step_1", step_1)
workflow.add_node("step_2", step_2)
workflow.add_node("step_3", step_3)
workflow.add_edge(START, "step_1")
workflow.add_edge("step_1", "step_2")
workflow.add_edge("step_2", "step_3")
workflow.add_edge("step_3", END)
# Set up memory
memory = MemorySaver()
# Add
graph = workflow.compile(checkpointer=memory, interrupt_before=["step_3"])
이를 통해 구현된 workflow는 아래와 같습니다.
이제 아래와 같이 입력을 지정하고 실행합니다. Step3에서 멈춘다음에 입력을 받으려고 대기 합니다.
initial_input = {"input": "hello world"}
thread = {"configurable": {"thread_id": "1"}}
# Run the graph until the first interruption
for event in graph.stream(initial_input, thread, stream_mode="values"):
print(event)
user_approval = input("Do you want to go to Step 3? (yes/no): ")
if user_approval.lower() == 'yes':
# If approved, continue the graph execution
for event in graph.stream(None, thread, stream_mode="values"):
print(event)
else:
print("Operation cancelled by user.")
이때의 실행결과는 아래와 같습니다. Step 3을 실행하기 전에 멈춰선 후에 "yes"를 입력하면, breakpoint 이후로 실행을 계속합니다.
{'input': 'hello world'}
---Step 1---
---Step 2---
Do you want to go to Step 3? (yes/no): yes
---Step 3---
아래와 같은 Tool을 이용하는 경우에 Breakpoints를 이용할 수 있습니다.
아래와 같이 tool과 node에 대한 함수를 정의합니다.
@tool
def search(query: str):
"""Call to surf the web."""
return [
"It's sunny in San Francisco, but you better look out if you're a Gemini 😈."
]
tools = [search]
tool_node = ToolNode(tools)
model = chat.bind_tools(tools)
def should_continue(state):
messages = state["messages"]
last_message = messages[-1]
if not last_message.tool_calls:
return "end"
else:
return "continue"
def call_model(state):
messages = state["messages"]
response = model.invoke(messages)
return {"messages": [response]}
이제, 아래와 같이 workflow를 정의합니다.
workflow = StateGraph(MessagesState)
workflow.add_node("agent", call_model)
workflow.add_node("action", tool_node)
workflow.add_edge(START, "agent")
workflow.add_conditional_edges(
"agent",
should_continue,
{
"continue": "action",
"end": END,
},
)
workflow.add_edge("action", "agent")
memory = MemorySaver()
app = workflow.compile(checkpointer=memory, interrupt_before=["action"])
아래와 같이 실행합니다.
from langchain_core.messages import HumanMessage
thread = {"configurable": {"thread_id": "3"}}
inputs = [HumanMessage(content="search for the weather in sf now")]
for event in app.stream({"messages": inputs}, thread, stream_mode="values"):
event["messages"][-1].pretty_print()
action을 실행하기 전에 아래와 같이 멈춥니다.
================================ Human Message =================================
search for the weather in sf now
================================== Ai Message ==================================
Tool Calls:
search (toolu_bdrk_01KJqMCKm1nd7ej6w3xayBSy)
Call ID: toolu_bdrk_01KJqMCKm1nd7ej6w3xayBSy
Args:
query: san francisco weather
이제 아래와 같이 Resume을 요청합니다.
for event in app.stream(None, thread, stream_mode="values"):
event["messages"][-1].pretty_print()
이후 아래와 같이 나머지 동작을 수행합니다.
================================= Tool Message =================================
Name: search
["It's sunny in San Francisco, but you better look out if you're a Gemini \ud83d\ude08."]
================================== Ai Message ==================================
The search results show the current weather conditions in San Francisco. It looks like it is sunny there right now. However, the results also include a humorous astrological warning for people with the Gemini zodiac sign, which doesn't seem directly relevant to the weather query.
To summarize the key information from the search:
The current weather in San Francisco is sunny.