LangChain Document与Pydantic 2.x兼容报错求助：参数数量不符

这个错误源于LangChain的Document类内部沿用了Pydantic 1.x的validate方法签名，而Pydantic 2.x对该方法的参数要求已改变，即使开启arbitrary_types_allowed也无法解决签名冲突问题。以下是几种可行的解决办法：

方案1：使用LangChain适配的BaseModel #

LangChain提供了兼容不同Pydantic版本的BaseModel类，直接继承它即可规避兼容问题：

from typing import List

from langchain_core.documents.base import Document
from langchain_core.base_model import BaseModel


class ResponseBody(BaseModel):
    message: List[Document]

docs = [Document(page_content="This is a document")]
res = ResponseBody(message=docs)

方案2：为Document添加自定义序列化器 #

若坚持使用原生Pydantic 2.x的BaseModel，可通过Annotated结合序列化器处理Document类型：

from typing import List, Annotated
from pydantic import BaseModel, PlainSerializer, WithJsonSchema
from langchain_core.documents.base import Document


def serialize_document(doc: Document) -> dict:
    return {"page_content": doc.page_content, "metadata": doc.metadata}

class ResponseBody(BaseModel):
    message: List[Annotated[
        Document,
        PlainSerializer(serialize_document),
        WithJsonSchema({"type": "object", "properties": {"page_content": {"type": "string"}, "metadata": {"type": "object"}}})
    ]]

docs = [Document(page_content="This is a document")]
res = ResponseBody(message=docs)

方案3：开启LangChain的Pydantic 2兼容模式 #

通过代码配置让LangChain启用Pydantic 2兼容模式，之后即可正常使用原生BaseModel并开启arbitrary_types_allowed：

import os
os.environ["LANGCHAIN_USE_PYDANTIC_V2"] = "true"

然后使用原代码即可：

from typing import List

from langchain_core.documents.base import Document
from pydantic import BaseModel, ConfigDict


class ResponseBody(BaseModel):
    message: List[Document]
    model_config = ConfigDict(arbitrary_types_allowed=True)

docs = [Document(page_content="This is a document")]
res = ResponseBody(message=docs)

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