Langgraph条件边未正常工作：节点3/4无连接可视化问题求助

作为LangGraph新手，我尝试创建包含node_1、node_2、node_3、node_4四个节点的简单图，边规则如下：

• node_1直接连接node_2
• node_2通过条件边连接node_3和node_4
• node_3、node_4最终连接至END

但可视化时发现node_3和node_4无连接，似乎被遗漏。以下是代码、当前图和期望图：

原代码 #

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

class State(TypedDict):
  user_input: str
  node_output: Annotated[list[int],add]

count = 0

def first_node(state):
  print("I'm here")
  global count
  count+=1
  return {'node_output':[count]}

def second_node(TypedDict):
  print("He sent me here")
  global count
  count+=2
  return {'node_output':[count]}


def third_node(TypedDict):
  print('I was decided')
  global count
  count+=3
  return {'node_output':[count]}

def fourth_node(TypedDict):
  print("I'm here on chance")
  global count
  count+=4
  return {'node_output':[count]}

#starting graph building

builder1 = StateGraph(State)

builder1.add_node('node_1',first_node)
builder1.add_node('node_2',second_node)
builder1.add_node('node_3',third_node)
builder1.add_node('node_4',fourth_node)

builder1.add_edge(START,'node_1')
builder1.add_edge('node_1','node_2')
builder1.add_conditional_edges('node_2',decide_mood)
builder1.add_edge('node_3',END)
builder1.add_edge('node_4',END)

my_new_graph = builder1.compile()

display(Image(my_new_graph.get_graph().draw_mermaid_png()))

当前可视化图 #

期望可视化图 #

问题排查与修复 #

1. 缺失条件路由函数decide_mood #

代码中调用builder1.add_conditional_edges('node_2',decide_mood)，但从未定义decide_mood函数。LangGraph无法识别无效的条件边，导致node_3、node_4未被纳入图的有效连接链路中，可视化时自然不会显示。

2. 节点函数参数错误 #

second_node、third_node、fourth_node的参数写为TypedDict，这是类型定义而非变量名，正确参数应为state（对应自定义的State类型），参数错误可能导致节点逻辑执行异常。

3. 缺失Image导入 #

代码中使用display(Image(...))但未导入Image类，需补充对应库的导入。

修复后的完整代码 #

from typing_extensions import TypedDict
from typing import Annotated
from langgraph.graph import StateGraph, START, END
from operator import add
from IPython.display import Image  # 补充Image导入

class State(TypedDict):
    user_input: str
    node_output: Annotated[list[int], add]

count = 0

def first_node(state):
    print("I'm here")
    global count
    count += 1
    return {'node_output': [count]}

# 修正参数为state
def second_node(state):
    print("He sent me here")
    global count
    count += 2
    return {'node_output': [count]}

# 修正参数为state
def third_node(state):
    print('I was decided')
    global count
    count += 3
    return {'node_output': [count]}

# 修正参数为state
def fourth_node(state):
    print("I'm here on chance")
    global count
    count += 4
    return {'node_output': [count]}

# 定义条件路由函数，示例根据count值判断，可按需修改逻辑
def decide_mood(state):
    global count
    # 示例条件：count为奇数时到node_3，偶数到node_4
    if count % 2 == 1:
        return 'node_3'
    else:
        return 'node_4'

# 构建图
builder1 = StateGraph(State)

builder1.add_node('node_1', first_node)
builder1.add_node('node_2', second_node)
builder1.add_node('node_3', third_node)
builder1.add_node('node_4', fourth_node)

builder1.add_edge(START, 'node_1')
builder1.add_edge('node_1', 'node_2')
# 条件边需指定目标节点映射，或函数返回节点名
builder1.add_conditional_edges(
    'node_2',
    decide_mood,
    {'node_3': 'node_3', 'node_4': 'node_4'}
)
builder1.add_edge('node_3', END)
builder1.add_edge('node_4', END)

my_new_graph = builder1.compile()

display(Image(my_new_graph.get_graph().draw_mermaid_png()))

修复后重新编译运行，可视化图会正确显示node_2到node_3、node_4的条件边，以及node_3、node_4到END的连接，与期望图一致。

内容的提问来源于stack exchange，提问作者Abhishek K M