How ccbikai/hacker-news works
Project Type and Purpose
• This is a full-stack, serverless web application combined with a Cloudflare Worker–driven data‐processing workflow.
• Purpose: every day fetch Hacker News top posts, summarize with OpenAI, synthesize audio via Edge TTS, store results in Cloudflare R2/KV, and expose via a Next.js-based website and RSS feed.
Key Architectural Components
a. Scheduled Trigger
• A time-based Cron trigger configured in Cloudflare Workers (via wrangler.jsonc) to kick off the workflow once per day.
b. Data-Processing Worker & Workflow
• worker/index.ts + workflow/*.ts files implement:
– Fetch top Hacker News articles (external HN API).
– Call OpenAI API to generate Chinese summaries.
– Call Edge TTS (Minimax Audio) to produce audio clips.
– Merge or process audio in a headless browser environment.
– Store JSON metadata and text in KV; store audio files in R2.
c. Storage Layer
• Cloudflare KV: stores metadata, text transcripts, pointers to audio URLs.
• Cloudflare R2: stores TTS-generated audio files.
d. Web Frontend
• Next.js 13 “app” directory:
– app/page.tsx: homepage listing episodes.
– app/post/[date]/page.tsx: individual episode pages.
– app/rss.xml/route.ts: RSS feed endpoint.
– app/static/[...path]/route.ts: serves static assets from R2.
• shadcn-ui + Tailwind for styling; OpenNext adapter for Cloudflare deployment.
e. External Services
• Hacker News API (data source).
• OpenAI API (text summarization).
• Edge TTS (audio generation via Minimax Audio).
f. Clients
• Web browsers (Next.js UI).
• Podcast apps via RSS feed.
• Any HTTP-capable RSS reader.
Relationships & Data Flows
• Cron → Worker (HTTP trigger or scheduled) → workflow code
• workflow → Hacker News API → receives articles
• workflow → OpenAI API → returns summaries
• workflow → Edge TTS → returns audio blobs
• workflow → Cloudflare R2 (audio) + KV (JSON/text)
• Next.js UI & RSS → read from KV & R2 → render pages & feeds
• Client (browser or podcast app) ← HTTP ← Next.js app or RSS endpoint
Architectural Patterns & Principles
• Serverless, event-driven: Cron triggers + Workers + KV/R2 storage.
• Micro-workflow: discrete steps (fetch → summarize → synthesize → store).
• Separation of concerns: data pipeline (Worker+workflow) vs. presentation layer (Next.js).
• Cloud-native: all components deployed on Cloudflare edge for low latency.