Baker_Project (made by Amar Kanakamedala and Sahil Joshi)

This tool helps you:

• Extract text from PDF files using a selected OCR engine
• Convert that text into structured metadata using an LLM
• Store that metadata in our database
• Search texts in our database based on keywords

More information can be found on OpenAPI (the automatic interactive documentation systems).

Getting Started

Follow these steps to run the application:

1. Create a Python virtual environment

   python3 -m venv myenv

2. Activate the virtual environment

source myenv/bin/activate  # macOS/Linux
myenv\Scripts\activate.bat  # Windows CMD
myenv\Scripts\Activate.ps1  # Windows PowerShell

3. Install all dependencies

pip install -r requirements.txt

4. Set Environment Variables

• APP_ENV, DEBUG, HOST, PORT
• DATABASE_URL
• OPENAI_API_KEY
• MISTRAL_API_KEY
• CORS_ORIGINS
• FRONTEND_DIST_DIR
• ELASTICSEARCH_URL

5. Run the application Run

   • fastapi dev src/backend/main.py
   • python src/frontend/frontend.py

6. Understand the application

• Documentation at http://127.0.0.1:8000/docs
• Server (FastAPI) started at http://127.0.0.1:8000
• Frontend (NiceGUI) ready to go on http://localhost:8080, and http://10.134.128.10:8080
• Click on "Process Data" in frontend to see functionality (Search functionality to be added)

Options Available

OCR Engine

• Mistral – Lightweight, fast OCR

Metadata Extraction (LLMs)

• ChatGPT – OpenAI GPT-4-based extraction