This sample application performs news article scraping with advanced classification and summarization capabilities. The application fetches article URLs from RSS feeds by polling various news outlets and maintains a cache of discovered URLs. It leverages OpenAI's ChatGPT-4o to categorize each article according to a specific Greek municipality. After categorization, the application generates a concise summary of the article.

The summaries are stored in Firebase for each municipality. The application uses Python for the scraper and NextJS for the frontend map. Additionally, it uses a Leaflet map to display emoji previews that capture the mood of the news, depending on the municipality area. These emojis provide a quick visual representation of the news sentiment within each region.

You can view a live demo at https://news.paravantis.org/.

• To run this app, set up a Firebase project with Firestore
• Use Docker to spin up a containerized version with docker compose up -d

To run the Python scraper backend:

• Create a virtual environment and install the dependencies with pip install -r requirements.txt
• Install the scraper package locally for development with pip install -e .
• Setup the required .env and firebase.json project files
• Launch the scraper with python run_instant.py

To run the NextJS frontend client:

• Install the dependencies with npm install
• Setup the required .env project files
• Launch the development server with npm run start