Recipe Summarizer (FastAPI + OpenAI/Gemini)

This project scrapes a food recipe web page and summarizes its content using your configured LLM. It is specialized to handle cooking recipes and provides chef-level, easy-to-understand summaries. If the provided website is not a food recipe, the service will inform you it cannot summarize non-recipe content.

• A production-ready FastAPI service with Basic Authentication
• A reusable summarization module that summarizes recipes with a "multicuisine chef" persona
• A Jupyter notebook for experimentation

Project Structure

• api_service.py: FastAPI app exposing POST /v1/summarize.
• website_summarizer.py: Orchestrates recipe scraping and calls the LLM.
• scraper.py: Extracts website text (Selenium/BS4 based).
• practice.ipynb: Sample notebook usage.
• pyproject.toml: Dependencies managed by uv.

Requirements

• Python 3.12+
• uv (recommended) or pip
• Chrome/Chromedriver available for Selenium (if using Selenium paths)

Setup

1. Install dependencies:

uv sync

2. Create a .env in the project root and set keys:

OPENAI_API_KEY=sk-...
GEMINI_API_KEY=... 
API_USERNAME=your_user
API_PASSWORD=your_pass

Notes:

• You can use OpenAI or Gemini via the existing code paths. Ensure keys are set accordingly.

Run the API

uv run uvicorn api_service:app --host 0.0.0.0 --port 8000

Health check:

curl -s http://localhost:8000/health

Summarize a recipe (POST JSON, Basic Auth):

curl -X POST "http://localhost:8000/v1/summarize" \
  -H "Authorization: Basic $(printf '%s' 'your_user:your_pass' | base64)" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://www.food.com/recipe/creamy-garlic-penne-pasta-43023"}'

Response:

{
  "url": "https://www.food.com/recipe/creamy-garlic-penne-pasta-43023",
  "summary": "..."
}

Notebook Usage

• Launch Jupyter with the same environment:

uv run jupyter notebook

• In the notebook, avoid print(display(Markdown(...))); use display(Markdown(...)) or return Markdown(...) as the last cell expression.

Troubleshooting

• If curl appears to hang:
  • Ensure server is running and reachable: lsof -i :8000
  • Try IPv4 loopback explicitly: curl -v http://127.0.0.1:8000/health
  • Kill previous processes on the same port: kill -9 $(lsof -t -i :8000)
• Selenium timeouts: ensure Chrome is installed and accessible; adjust waits in scraper.py.

Security

• Basic Auth is required when API_USERNAME/API_PASSWORD are set.
• Tighten CORS and logging in production; use a process manager (systemd, Docker, or similar).

License

MIT (or your preferred license).