Ground News Austria

Ground News Austria lets users browse and search clustered news topics, open topic detail pages with related articles and generated summaries, compare coverage across left, center, and right-leaning Austrian publishers, follow links back to original articles, and manage a basic account with profile and history features. The frontend is implemented with Angular and reusable UI components, while the Spring Boot REST API exposes topic, search, authentication, user, and admin pipeline endpoints secured with JWT.

Articles are collected by Python scraper workers using a configurable source registry, stored as raw snapshots in MinIO, and moved through RabbitMQ as versioned pipeline events. Spring services persist data in PostgreSQL with Flyway migrations, use pgvector for semantic article clustering, Redis for repeated embedding/context cache lookups, Stanford CoreNLP for German entity extraction, and optional Ollama/Spring AI integration for topic summaries and contextual notes.

Tech Stack

Area	Technologies
Frontend	Angular 20, TypeScript, RxJS, Tailwind CSS, Angular Material/CDK, Cypress
Backend	Java 25, Spring Boot 3.5, Spring Security, Spring AMQP, Spring AI, Flyway, MapStruct
Data	PostgreSQL 17 with pgvector, Redis, MinIO
Messaging	RabbitMQ 4 with management UI and dead-letter queues
AI/NLP	Ollama, Spring AI, Stanford CoreNLP German models, embeddings
Scraping	Python workers, custom publisher scrapers, configurable source registry
Ops	Docker Compose, Kustomize/Kubernetes, Prometheus metrics, Testcontainers

Run locally

Copy the example environment file if you want to override defaults:

cp .env.example .env

Start the core application:

docker compose up --build

Useful Docker Compose profiles:

Command	What it starts
docker compose up --build	Frontend, backend, PostgreSQL, Redis, RabbitMQ, MinIO
docker compose --profile pipeline up --build	Core stack plus scraper worker
docker compose --profile llm up --build	Core stack plus Ollama
docker compose --profile pipeline --profile llm up --build	Full local pipeline

Default local endpoints:

Service	URL
Frontend	http://localhost:4200
Backend API	http://localhost:8080
OpenAPI UI	http://localhost:8080/swagger-ui/index.html
RabbitMQ Management	http://localhost:15672
MinIO Console	http://localhost:9001

Run a smoke check after the stack is up:

./scripts/smoke-local.sh

Running the Scraper

Start the shared stack:

docker compose up database redis rabbitmq minio backend

Then run the worker locally:

cd news_scraper
uv sync
uv run cli.py worker

Trigger a crawl request for a configured source:

uv run cli.py publish-crawl derstandard --max-articles 5

Fetch EventRegistry-backed articles into the same pipeline:

EVENTREGISTRY_API_KEY=... uv run cli.py newsapi --max-articles 100 --days-back 30

Verification

Backend integration tests use Testcontainers and require Docker access.

./scripts/verify-backend.sh
./scripts/verify-frontend.sh
./scripts/verify-scraper.sh

Kubernetes

Kustomize manifests live in deployment/k8s.

For a local cluster:

docker compose build backend frontend
kubectl apply -k deployment/k8s/overlays/local

The Kubernetes manifests use development defaults. Replace secrets, image names, hostnames, CORS origins, resource settings, and LLM/runtime configuration before deploying to a hosted environment.