Aero-offers

This project aims at reviving aero-offers.com - invaluable source of price trends for gliders and other aircrafts, originally developed and maintained by @rthaenert

Development

Project structure (building blocks / deployment units)

• ui - vue.js application. Deployed as Azure Static Web App (for free).

• backend

  • web api:
    • python flask app with few REST endpoints
    • deployed as a lightweight Azure Container App (running 24/7)
  • update offers scheduled job:
    • scrapes new offers from few ads portals (e.g. soaring.de) and stores them in the database
    • runs classifier for not yet classified offers to find exact manufacturer and model, utilizing Gemini LLM api, which provides excellent accuracy compared to legacy, rule-based classifier.
    • it's much more resource-heavy than API, bundled in the same Docker image as api, deployed as Azure Container Job.

• db - Azure CosmosDb NoSQL fully managed database (for free; check #135 for decision record).

• terraform - deployment infrastructure to Azure

Deployment

Trunk Based Development and Continuous Deployment is utilized here - all changes pushed/merged to main are automatically deployed to production env. Currently, the project is running in Azure, managed via Terraform.

TODO

• Scraper: update db offer if price or location (or any other parameter) has changed
• aero-offers.com
• improve aircraft types structure and introduce 2 levels: glider (e.g Discus 2c 18m) and model (Discus 2cFES 18m) as prices between models sometimes differ significantly
• admin panel for manual (re) classification (or community-based)
• crawler for Facebook Marketplace - do they have an api?
• crawler for https://www.aircraft24.de
• crawler for https://plane-sale.com

Running backend locally without Python

docker compose up --build - starts CosmosDb and Python Flask backend

Prerequisites for local development with hot reloads

• python 3.13+
• uv (pip install uv)
• docker (compose)
• nodejs

Start CosmosDb emulator in docker (available for debugging via localhost:1234):

docker-compose up cosmosdb

Init python environment

cd backend
uv sync

Set up auto-formatting and lint fix before commit (recommended:

git config core.hooksPath hooks

Start backend api (python app):

cd backend
./run_api_in_docker.sh

Start UI (vue app):

cd ui
npm run dev

UI has own, detailed README.md file.

Run crawlers/spiders & classifier:

cd backend
./run_spiders.sh && ./run_classifier.sh