expenSieve: LLM-based Receipt Processing Web App

This is a Flask-based web application that allows users to upload multiple receipt images (or PDFs), extract structured expense data using OpenAI's GPT-4o Vision model, and save the parsed information to a CSV file. The application is designed to be mobile-friendly and supports multiple file uploads at once.

Overview

This web application allows users to upload multiple receipts (images or PDFs), extract structured expense data using OpenAI's GPT-4o Vision model, and save the processed transactions to a CSV file. The app is built with Flask for the backend and JavaScript for the frontend.

Features

• Upload multiple receipts (JPG, PNG, or PDF format)
• Process receipts using OpenAI's GPT-4o for structured data extraction
• Display extracted data in an editable table
• Preview receipt thumbnails in the table
• Save expense data to a CSV file
• Mobile-friendly UI

Installation

Prerequisites

• Python 3.8+
• Flask
• OpenAI API Key (if using GPT-4o for processing)
• Node.js (optional, for frontend enhancements)

Setup

1. Clone the repository:

   git clone https://github.com/your-username/expense-receipt-processor.git
   cd expense-receipt-processor

2. Install dependencies:

   pip install -r requirements.txt

3. Set up environment variables: Create a .env file and add your OpenAI API key:

   OPENAI_API_KEY=your_api_key_here
   

4. Run the Flask app:

   python app.py

5. Open the web app in a browser:

   http://127.0.0.1:5000
   

Usage

1. Select and upload one or multiple receipt images/PDFs.
2. View extracted structured data in the table.
3. Edit any incorrect values directly in the table.
4. Click Save to store the transactions in the CSV file.

File Structure

expense-receipt-processor/
¦-- static/
¦   +-- uploads/         # Stores uploaded files
¦   +-- processed/       # Stores processed images
¦-- templates/
¦   +-- index.html       # Frontend UI
¦-- app.py               # Flask backend
¦-- requirements.txt     # Python dependencies
¦-- .env                 # Environment variables
¦-- README.md            # This file

API Endpoints

POST /upload

• Input: Multipart form with one or more receipt files
• Output: JSON with extracted receipt data and image preview URLs

POST /save

• Input: JSON with edited expense details
• Output: Confirmation message & updated spreadsheet data

Future Enhancements

• Automatic email parsing for receipts
• Multi-page PDF handling
• Cloud storage integration (OneDrive/Google Drive)

Author

Sammit Jain