Langchain中OpenAI函数代理传参报错：TypeError参数数量不符

问题描述 #

我正在使用Langchain框架构建一个可帮我向商店下单的代理。用于下单的工具/函数需要多输入参数，但运行时持续报错：TypeError: __init__() takes 1 positional argument but 7 were given。

原代码 #

from placeorder import order_three
from langchain.chat_models import ChatOpenAI
from pydantic import BaseModel, Field
from typing import Optional, Type
from langchain.agents import AgentType
from langchain.tools import BaseTool
from langchain.agents import initialize_agent, Tool


class order_three(BaseModel):
    """Place an order"""
first_name: str =Field(...,description= "The first name of person placing the order")
last_name: str =Field(...,description= "The last name of person placing the order")
phone_number: str =Field(...,description= "The phone number  of person placing the order")
product: str =Field(...,description= "The product they want to buy")
date_of_birth: str =Field(...,description= "The date of birth of person placing the order")
quantity: str =Field(...,description= "The amount of the product they want to buy")

class order_three(BaseTool):
name = "place_order"
description = "Useful when the user wants to place an order. You should input the firstname, lastname, quantity, product, date of birth, and phone number"

def _run(self, first_name: str, last_name: str, quantity: str, date_of_birth: str, product: str, phone_number: str):
place_order_response = order_three(first_name, last_name, phone_number, date_of_birth, product, quantity)

return place_order_response

def _arun(self, first_name: str, last_name: str, quantity: str, date_of_birth: str, product: str, phone_number: str):
raise NotImplementedError("This tool does not support async")

args_schema: Optional[Type[BaseModel]] = order_three


tools = [order_three()]

llm = ChatOpenAI(temperature=0, model="gpt-3.5-turbo-0613")

open_ai_agent = initialize_agent(tools,
llm,
agent=AgentType.OPENAI_FUNCTIONS,
verbose=True)

print(open_ai_agent.run("I want to place an order. My name is John smith my phone number is 5555555555. my date of birth is 030419999 I want to order 8 large Pizzas"))

报错信息 #

Entering new AgentExecutor chain...

Invoking: place_order with {'first_name': 'John', 'last_name': 'Smith', 'phone_number': '5555555555', 'product': 'large Pizzas', 'date_of_birth': '03041999', 'quantity': '8'}

Traceback (most recent call last):
  File "/Users/izayahhudnut/Documents/Code/BudtenderAI/vite-project/testagent.py", line 45, in <module>
    print(open_ai_agent.run("I want to place an order. My name is John smith my phone number is 5555555555. my date of birth is 030419999 I want to order 8 large Pizzas"))
  File "/Users/izayahhudnut/Library/Python/3.9/lib/python/site-packages/langchain/chains/base.py", line 487, in run
    return self(args[0], callbacks=callbacks, tags=tags, metadata=metadata)[
  File "/Users/izayahhudnut/Library/Python/3.9/lib/python/site-packages/langchain/chains/base.py", line 292, in __call__
    raise e
  File "/Users/izayahhudnut/Library/Python/3.9/lib/python/site-packages/langchain/chains/base.py", line 286, in __call__
    self._call(inputs, run_manager=run_manager)
  File "/Users/izayahhudnut/Library/Python/3.9/lib/python/site-packages/langchain/agents/agent.py", line 1039, in _call
    next_step_output = self._take_next_step(
  File "/Users/izayahhudnut/Library/Python/3.9/lib/python/site-packages/langchain/agents/agent.py", line 894, in _take_next_step
    observation = tool.run(
  File "/Users/izayahhudnut/Library/Python/3.9/lib/python/site-packages/langchain/tools/base.py", line 356, in run
    raise e
  File "/Users/izayahhudnut/Library/Python/3.9/lib/python/site-packages/langchain/tools/base.py", line 330, in run
    else self._run(*tool_args, **tool_kwargs)
  File "/Users/izayahhudnut/Documents/Code/BudtenderAI/vite-project/testagent.py", line 25, in _run
    place_order_response = order_three(first_name, last_name, phone_number, date_of_birth, product, quantity)
TypeError: __init__() takes 1 positional argument but 7 were given

错误原因 #

1. 类名重复冲突：先后定义了两个order_three类，一个继承BaseModel，另一个继承BaseTool，后续代码中调用order_three(...)时，实际指向的是继承BaseTool的类，而BaseTool的构造函数不需要传入这些参数，导致参数数量不匹配。
2. 未正确调用导入的下单函数：代码开头从placeorder模块导入了order_three函数，但在_run方法中没有使用这个导入的函数，反而调用了自己定义的同名类。
3. 缩进错误：BaseModel的字段、BaseTool的属性和方法都没有正确缩进，这会导致语法错误，影响代码的正常运行。

修复后的代码 #

from placeorder import order_three  # 导入的实际下单函数
from langchain.chat_models import ChatOpenAI
from pydantic import BaseModel, Field
from typing import Optional, Type
from langchain.agents import AgentType
from langchain.tools import BaseTool
from langchain.agents import initialize_agent


# 重命名BaseModel类，避免和导入的函数、工具类重名
class OrderThreeArgs(BaseModel):
    """Place an order"""
    first_name: str = Field(..., description="The first name of person placing the order")
    last_name: str = Field(..., description="The last name of person placing the order")
    phone_number: str = Field(..., description="The phone number of person placing the order")
    product: str = Field(..., description="The product they want to buy")
    date_of_birth: str = Field(..., description="The date of birth of person placing the order")
    quantity: str = Field(..., description="The amount of the product they want to buy")


class OrderThreeTool(BaseTool):
    name = "place_order"
    description = "Useful when the user wants to place an order. You should input the firstname, lastname, quantity, product, date of birth, and phone number"
    args_schema: Optional[Type[BaseModel]] = OrderThreeArgs

    def _run(self, first_name: str, last_name: str, quantity: str, date_of_birth: str, product: str, phone_number: str):
        # 调用导入的实际下单函数
        place_order_response = order_three(first_name, last_name, phone_number, date_of_birth, product, quantity)
        return place_order_response

    def _arun(self, first_name: str, last_name: str, quantity: str, date_of_birth: str, product: str, phone_number: str):
        raise NotImplementedError("This tool does not support async")


tools = [OrderThreeTool()]

llm = ChatOpenAI(temperature=0, model="gpt-3.5-turbo-0613")

open_ai_agent = initialize_agent(
    tools,
    llm,
    agent=AgentType.OPENAI_FUNCTIONS,
    verbose=True
)

print(open_ai_agent.run("I want to place an order. My name is John smith my phone number is 5555555555. my date of birth is 03041999 I want to order 8 large Pizzas"))

修复说明 #

• 重命名了BaseModel类为OrderThreeArgs，避免和导入的下单函数、工具类名称冲突。
• 将工具类命名为OrderThreeTool，明确区分类和导入的函数。
• 修正了所有代码的缩进问题，确保符合Python语法规范。
• 在_run方法中明确调用从placeorder模块导入的order_three函数，而不是自己定义的类。

内容的提问来源于stack exchange，提问作者Izayah Hudnut