Natural Language to SQL Converter

A simple natural language to SQL converter

Tech Stack

• Python
• Pandas
• OpenAI API
• SQLAlchemy / SQLite

How it works

1. Pandas converts the data file (in this case, instruments.json) into a data frame
2. A temporary SQLite database is created using SQLAlchemy and the data frame is pushed to the database
3. Your request from the user_input variable is converted into a usable prompt
4. The prompt is sent to OpenAI's Completion API
5. Grab the resulting query and run it in the database

The Prompt Sent to OpenAI

### sqlite SQL table, with its properties:
# 
# instruments({columns in data})
# 
### A query to answer: {natural language prompt}
SELECT

The SELECT keyword at the end of the prompt is meant to force the LLM to produce a SELECT statement only.
This app is unable to produce potentially harmful DROP, DELETE, and UPDATE statements because the Completion API stops producing tokens after a semi-colon, which signifies the end of the SELECT statement.

How to run on machine

1. Get an API key from OpenAI, and place it in a file called secret.py in the root
2. You can replace the data if you wish. Check the Pandas docs to see supported files.
3. Create a virtual environment by running python3 -m venv env
4. Start the environment by running source env/bin/activate
5. Run pip install -r requirements.txt to install dependencies
6. Run at anytime by running python3 app.py