YLE Asiaotsikot

A web application that fetches YLE's (Finnish Broadcasting Company) most popular news articles and uses AI to improve clickbait headlines, making them more informative and accurate.

YLE Asiaotsikot is an automated news feed generator that:

• Fetches the latest news from YLE's RSS feed
• Uses AI (Google Claude) to analyze articles and improve clickbait titles
• Displays articles with corrected titles while showing the original title for comparison
• Updates automatically twice daily (around 7:10 and 16:10)

Note: This is not an official YLE website.

Frontend is a statically built SvelteKit and Tailwind CSS single-page app. All feeds (today's and previous days') are fetched from the R2 bucket in the browser.

Backend consists of a periodically running node script that fetches news articles, processes them with AI, and stores them in a SQLite database to cache results. It then uploads the daily feed as <date>.json to a Cloudflare R2 bucket, which the frontend reads.

Infrastructure is handled with Github Actions, Pages, GHCR & Cloudflare R2. On every push to master the backend Docker image is built and pushed to ghcr.io, and the frontend is built and deployed to GitHub Pages.

yle-etusivu/
├── backend/          # News fetching & AI processing service
│   └── src/
│       ├── main.ts                    # Cron jobs, RSS fetching, AI title improvement
│       ├── r2.ts                      # Cloudflare R2 upload
│       ├── title-improvement-template.ts
│       └── utils.ts
├── frontend/         # SvelteKit web application
│   └── src/
│       ├── lib/
│       │   ├── Feed.svelte           # Main feed component
│       │   ├── api.ts
│       │   └── utils.ts
│       └── routes/
│           ├── +page.svelte          # Homepage
│           ├── [date]/               # Historical views
│           └── faq/                  # FAQ page
├── shared/           # Shared code (entities, types, database)
│   └── src/
│       ├── entities/                 # MikroORM entities
│       │   ├── Article.ts
│       │   ├── Collection.ts
│       │   └── ArticleInCollection.ts
│       ├── types/
│       │   └── feed.ts
│       └── mikro-orm.config.ts
├── data/             # SQLite database storage
├── docker-compose.yml
├── Dockerfile
└── package.json

Create a .env file in the project root:

# Required for AI title improvement
ANTHROPIC_API_KEY=your_anthropic_api_key

# Cloudflare R2
R2_ACCOUNT_ID=...
R2_API_TOKEN=...       # needs "Workers R2 Storage: Edit"
R2_BUCKET_NAME=...

# Optional
ARTICLE_COUNT=15  # Number of articles to fetch (default: 15)
SKIP_INITIAL_RUN=true

See backend/.env.example, frontend/.env.example, package.json

Github Copilot has been used to assist in writing parts of the codebase, and most of this README.