SQLify

A Streamlit web application that converts natural-language business questions into safe, read-only SQLite queries using a local Hugging Face transformer model.

Features

• Connects dynamically to any SQLite database and extracts schema.
• Uses Hugging Face transformers to generate SQL from natural language.
• Strictly validates generated SQL for read-only safety (blocks INSERT, UPDATE, DROP, multi-statement, etc.).
• Clean Streamlit UI with query history, settings, and result display.

Setup Instructions

1. Install Dependencies

Make sure you have Python 3.11 installed. Create a virtual environment and install the requirements:

python -m venv venv
# On Windows:
venv\Scripts\activate
# On Linux/Mac:
source venv/bin/activate

pip install -r requirements.txt

2. Configure Environment Variables

Create a .env file in the root directory (or rename .env.example to .env) and update the paths:

# Hugging Face Model ID (e.g. "budecosystem/sql-millennials-13b") 
# Or an absolute local path to your model folder
MODEL_ID_OR_PATH="budecosystem/sql-millennials-13b"

# Absolute path to your SQLite database
DATABASE_PATH="REPLACE_WITH_MY_SQLITE_DB"

3. Run the App

Start the Streamlit server:

streamlit run app.py

Known Limitations

• The application currently supports SQLite only.
• Designed for small/medium local transformer models. Ensure you have adequate RAM/VRAM for the model you specify.
• V1 does not include user authentication or containerization.

Example Questions

• "Show total sales by month"
• "How many users signed up last week?"
• "List the top 5 products by revenue"