Source code for langchain_community.chains.llm_requests
"""链式操作,访问一个URL,然后使用LLM来解析结果。"""
from __future__ import annotations
from typing import Any, Dict, List, Optional
from langchain.chains import LLMChain
from langchain.chains.base import Chain
from langchain_core.callbacks import CallbackManagerForChainRun
from langchain_core.pydantic_v1 import Extra, Field, root_validator
from langchain_community.utilities.requests import TextRequestsWrapper
DEFAULT_HEADERS = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/87.0.4280.88 Safari/537.36" # noqa: E501
}
[docs]class LLMRequestsChain(Chain):
"""请求一个URL,然后使用LLM来解析结果的链。
**安全提示** :此链可以向任意URL发出GET请求,包括内部URL。
控制谁可以运行此链以及此链具有什么网络访问权限。
有关更多信息,请参见https://python.langchain.com/docs/security。"""
llm_chain: LLMChain # type: ignore[valid-type]
requests_wrapper: TextRequestsWrapper = Field(
default_factory=lambda: TextRequestsWrapper(headers=DEFAULT_HEADERS),
exclude=True,
)
text_length: int = 8000
requests_key: str = "requests_result" #: :meta private:
input_key: str = "url" #: :meta private:
output_key: str = "output" #: :meta private:
class Config:
"""此pydantic对象的配置。"""
extra = Extra.forbid
arbitrary_types_allowed = True
@property
def input_keys(self) -> List[str]:
"""将是提示所期望的任何键。
:元数据 私有:
"""
return [self.input_key]
@property
def output_keys(self) -> List[str]:
"""将始终返回文本键。
:元数据 私有:
"""
return [self.output_key]
@root_validator()
def validate_environment(cls, values: Dict) -> Dict:
"""验证环境中是否存在API密钥和Python包。"""
try:
from bs4 import BeautifulSoup # noqa: F401
except ImportError:
raise ImportError(
"Could not import bs4 python package. "
"Please install it with `pip install bs4`."
)
return values
def _call(
self,
inputs: Dict[str, Any],
run_manager: Optional[CallbackManagerForChainRun] = None,
) -> Dict[str, Any]:
from bs4 import BeautifulSoup
_run_manager = run_manager or CallbackManagerForChainRun.get_noop_manager()
# Other keys are assumed to be needed for LLM prediction
other_keys = {k: v for k, v in inputs.items() if k != self.input_key}
url = inputs[self.input_key]
res = self.requests_wrapper.get(url)
# extract the text from the html
soup = BeautifulSoup(res, "html.parser")
other_keys[self.requests_key] = soup.get_text()[: self.text_length]
result = self.llm_chain.predict( # type: ignore[attr-defined]
callbacks=_run_manager.get_child(), **other_keys
)
return {self.output_key: result}
@property
def _chain_type(self) -> str:
return "llm_requests_chain"