/Hazelnut

Primary LanguagePythonMIT LicenseMIT

Hazelnut

Hazelnut: Nutty Precision for your SQL Queries

Table of Contents

Introduction

Hazelnut is an LLM-powered SQL query builder and executer designed to simplify and automate database interactions. With Hazelnut, you can generate and execute SQL queries efficiently, making database management easier than ever.

Project Demo

Here is a demonstration of Hazelnut in action:

Hazelnut Demo

Features

  • LLM-Powered Query Generation: Generate SQL queries using natural language input.
  • Query Execution: Execute the generated queries directly on your database and see the result.
  • Database schema aware: Knows the entire schema/structure of the given database to answer queries schematically.
  • Context aware chatbot: Can response to questions while also considering the previous conversation.
  • User-Friendly Interface: Easy-to-use interface for seamless interaction.

Installation

To get started with Hazelnut, follow these steps:

  1. Clone the Repository:

    git clone https://github.com/Aditi-Asati/Hazelnut.git
cd Hazelnut

  2. Build Docker Image:

    docker build -t hazelnut .

API Usage

Hazelnut provides a set of API endpoints to facilitate the generation and execution of SQL queries through a FastAPI backend. Below are the details of each endpoint:

1. Submit Form

Endpoint: POST http://localhost:8000/submit

Description: This endpoint is used to submit database connection details. It returns a session ID that is used for subsequent requests.

Request Body:

{
    "host": "your_database_host",
    "port": your_database_port,
    "username": "your_database_username",
    "password": "your_database_password",
    "database": "your_database_name"
}

Response

{
    "session_id": "generated_session_id"
}

1. Chat

Endpoint: POST http://localhost:8000/chat/{session_id}

Description: This endpoint accepts a session ID as a path parameter and a question as part of the request body. It returns the generated SQL query.

Request Body:

{
    "question": "your_question_here",
    "credentials": {
        "host": "your_database_host",
        "port": your_database_port,
        "username": "your_database_username",
        "password": "your_database_password",
        "database": "your_database_name"
}
}

Response

{
    "answer": "generated_sql_query"
}

1. Execute query

Endpoint: POST http://localhost:8000/execute

Description: This endpoint executes the provided SQL query on the database and returns the query results.

Request Body:

{
    "question": "generated_sql_query",
    "credentials": {
        "host": "your_database_host",
        "port": your_database_port,
        "username": "your_database_username",
        "password": "your_database_password",
        "database": "your_database_name"
    }
}

Response

{
    "result": [
        // list of rows with query results
    ],
    "columns": [
        // list of column names
    ]
}

Guidelines

  1. Start the Application:

    docker run -p 8000:8000 -p 8501:8501 hazelnut

  2. Interact with Hazelnut:

    • Use the interface to input natural language queries.
    • Review and execute the generated SQL queries.

  3. Example:

    • Input: "Show me all customers from New York"
    • Generated SQL: SELECT * FROM customers WHERE city = 'New York';

License

Hazelnut is licensed under the MIT License. See the LICENSE file for more details.