Dataherald

Dataherald is a natural language-to-SQL.

This page covers how to use the Dataherald API within LangChain.

Installation and Setup

• Install requirements with

pip install dataherald

• Go to dataherald and sign up here
• Create an app and get your API KEY
• Set your API KEY as an environment variable DATAHERALD_API_KEY

Wrappers

Utility

There exists a DataheraldAPIWrapper utility which wraps this API. To import this utility:

from langchain_community.utilities.dataherald import DataheraldAPIWrapper

API Reference:

• DataheraldAPIWrapper

For a more detailed walkthrough of this wrapper, see this notebook.

Tool

You can use the tool in an agent like this:

from langchain_community.utilities.dataherald import DataheraldAPIWrapper
from langchain_community.tools.dataherald.tool import DataheraldTextToSQL
from langchain_openai import ChatOpenAI
from langchain import hub
from langchain.agents import AgentExecutor, create_react_agent, load_tools

api_wrapper = DataheraldAPIWrapper(db_connection_id="<db_connection_id>")
tool = DataheraldTextToSQL(api_wrapper=api_wrapper)
llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0)
prompt = hub.pull("hwchase17/react")
agent = create_react_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
agent_executor.invoke({"input":"Return the sql for this question: How many employees are in the company?"})

API Reference:

• DataheraldAPIWrapper
• DataheraldTextToSQL
• ChatOpenAI
• AgentExecutor
• create_react_agent
• load_tools

Output

> Entering new AgentExecutor chain...
I need to use a tool that can convert this question into SQL.
Action: dataherald
Action Input: How many employees are in the company?Answer: SELECT
    COUNT(*) FROM employeesI now know the final answer
Final Answer: SELECT
    COUNT(*)
FROM
    employees

> Finished chain.
{'input': 'Return the sql for this question: How many employees are in the company?', 'output': "SELECT \n    COUNT(*)\nFROM \n    employees"}

For more information on tools, see this page.