🦜 Chat with SQL DB — Powered by LangChain & Groq

This project is an interactive Streamlit-based web application that allows users to chat with a SQL database using natural language. Powered by LangChain and integrated with Groq’s LLaMA3 model, this tool enables users to query and explore both SQLite and MySQL databases without writing SQL directly.

---

💡 Project Overview

LangChain SQL Chat App simplifies database interaction using advanced natural language processing. With support for both SQLite (local or uploaded) and MySQL (remote) databases, users can ask questions such as:

• “Show all students who scored above 90.”
• “What is the average marks per section?”
• “List all students from the Web Developer class.”

The app translates these queries to SQL using LangChain’s agent-based architecture and displays the results in real-time.

---

⚙️ How It Works

1. Choose Your Database

• From the sidebar, select between:
  • Upload or provide path to a local SQLite database, or
  • Connect to a remote MySQL database using host, username, password, and DB name.

2. API Key for Groq

• Enter your Groq API Key in the sidebar to initialize the LLaMA3 model for natural language querying.

3. Model & Agent Setup

• A LangChain agent is created using:
  • The selected database.
  • ChatGroq LLM (LLaMA3-8b-8192).
  • SQLDatabaseToolkit to support schema understanding and tool invocation.

4. Chat Interface

• A familiar chat interface allows users to ask natural language questions.
• The agent uses the LLM to interpret, generate SQL queries, and return results.
• Chat history is retained within the session and can be cleared with a button.

---

📌 Notes

• No need to store your GROQ API key in a .env file — the key is securely entered by the user through the Streamlit UI at runtime.
• Sample database creation script (sqlite.py) is included to populate and test with student records.
• Ensure your Groq API key has sufficient usage limits for LLaMA3 queries.

---

🖼️ Some Screenshots of Work

📁 Connecting to SQLite Database via File Upload

The application allows you to upload a .sqlite file directly, and it automatically connects and extracts the database schema for natural language queries.

---

---

🧰 Viewing Tables using SQL Workbench

This screenshot shows the SQL Workbench interface where the database structure can be reviewed and validated before or during chatbot interaction.

---

🌐 Connecting to Remote SQL Server

Instead of uploading a file, you can connect to a remote MySQL server by entering four key details: Host, Username, Password, and Database Name.

---

⚙️ Setup Instructions

Follow the steps below to run the project locally:

1️⃣ Clone the Repository

git clone https://github.com/prince2004patel/chat-with-sql-db.git
cd chat-with-sql-db

2️⃣ Create and Activate Conda Environment

• It’s recommended to use a virtual environment to avoid conflicts:

conda create --name chat-with-sql-db python=3.10 -y
conda activate chat-with-sql-db

3️⃣ Install Dependencies

pip install -r requirements.txt

4️⃣ Run the App

streamlit run app.py