PSU Menu Analyzer

Web application that scrapes Penn State campus dining menus, analyzes food items using Google Gemini AI, and provides personalized meal recommendations based on dietary preferences.

What It Does

• Scrapes daily menus from Penn State dining websites
• Analyzes food items with AI for health scores (0-100)
• Filters by dietary preferences (vegetarian, vegan, protein priority, exclusions)
• Ranks recommendations by meal (Breakfast, Lunch, Dinner)
• Provides nutrition information links
• Caches results for 24 hours to reduce API calls

Controls:

• Select campus dining location from dropdown
• Toggle dietary preferences (vegetarian, vegan, exclude beef/pork, prioritize protein)
• Click "Analyze Today's Menu" button
• Preferences are saved to localStorage

How It Works

1. Menu Scraping: Flask backend scrapes menu data from Penn State dining websites using BeautifulSoup
2. Data Processing: Parses HTML to extract food items, meal times, and nutrition links
3. AI Analysis: Sends food items to Google Gemini 3.1 Flash API for health scoring
4. Preference Filtering: Applies user dietary restrictions and preferences server-side
5. Ranking: Sorts items by health score and protein content (if prioritized)
6. Caching: Stores results in pickle files with MD5 hash keys for 24-hour cache
7. Display: Frontend displays ranked recommendations with scores and analysis

Dependencies

• Flask - Backend web framework
• Flask-CORS - Cross-origin resource sharing
• beautifulsoup4 - HTML parsing and web scraping
• requests - HTTP requests for menu scraping
• aiohttp - Async HTTP requests
• python-dotenv - Environment variable management
• gunicorn - WSGI server for production
• Google Gemini 3.1 Flash API - AI-powered nutritional analysis
• Tailwind CSS - Frontend styling framework

Technical Details

• Backend: Flask REST API with CORS enabled
• Frontend: HTML/JavaScript with Tailwind CSS
• Health Score Range: 0-100 (based on protein, cooking method, nutritional balance)
• Supported Locations: 16+ Penn State campus dining locations
• Cache: 24-hour pickle file cache with MD5 hash keys
• Scraping: BeautifulSoup parsing of Penn State dining HTML
• AI Model: Google Gemini 3.1 Flash
• Note: Results may be inaccurate on weekends due to missing menu data