Web2API TinyFish AI Hackathon

Web2API is a FastAPI web app for turning messy websites into structured outputs.

It includes two workflows:

• Storefront extraction: uses TinyFish + OpenAI to extract product-like results and render them in a table and UI preview
• Website analysis: fetches a live page, infers schema/page structure, and generates sample API endpoints

Features

• FastAPI backend
• TinyFish-powered extraction
• OpenAI-powered schema shaping and showcase generation
• Product table preview
• UI-based product preview
• Website analysis studio
• Generated sample API endpoints

Local run

python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
uvicorn main:app --reload

Open:

• http://127.0.0.1:8000/
• http://127.0.0.1:8000/docs

Environment variables

Create a .env file with:

TINYFISH_API_KEY=your_tinyfish_key
OPENAI_API_KEY=your_openai_key

Optional:

TINYFISH_URL=https://agent.tinyfish.ai/v1/automation/run

Deploy

For Render:

• Build command: pip install -r requirements.txt
• Start command: uvicorn main:app --host 0.0.0.0 --port $PORT

Notes

• The website-analysis workflow stores results in memory, so they reset on redeploy/restart.
• This is acceptable for a hackathon demo but not persistent production storage.