brightsky 2.2.7

JSON API for DWD's open weather data.

Bright Sky

JSON API for DWD's open weather data.

The DWD (Deutscher Wetterdienst), as Germany's meteorological service, publishes a myriad of meteorological observations and calculations as part of their Open Data program.

Bright Sky is an open-source project aiming to make some of the more popular data — in particular weather observations from the DWD station network and weather forecasts from the MOSMIX model — available in a free, simple JSON API.

Looking for something specific?

I just want to retrieve some weather data

You can use the free public Bright Sky instance!

I want to run my own instance of Bright Sky

Check out the infrastructure repo!

I want to parse DWD weather files from the command line or in Python

The parsing core for Bright Sky is maintained in a separate package named dwdparse, which has no dependencies outside the standard library. If you find that's not quite serving your needs, check out wetterdienst.

I want to contribute to Bright Sky's source code

Read on. :)

On Bright Sky's versioning

Starting from version 2.0, where we extracted the parsing core into a separate package, Bright Sky is no longer intended to be used as a Python library, but only as the service available at brightsky.dev.

Consequentially, we adjust our version numbers from the perspective of that service and its users – i.e., we will increase the major version number only when we introduce backwards-incompatible (or otherwise very major) changes to the actual JSON API interface, e.g. by changing URLs or parameters. This means that increases of the minor version number may introduce backwards-incompatible changes to the internals of the brightsky package, including the database structure. If you use brightsky as a Python library, please version-pin to a minor version, e.g. by putting brightsky==2.0.* in your requirements.txt.

Quickstart

Running a full-fledged API instance

Note: These instructions are aimed at running a Bright Sky instance for development and testing. Check out our infrastructure repository if you want to set up a production-level API instance.

Just run docker-compose up and you should be good to go. This will set up a PostgreSQL database (with persistent storage in .data), run a Redis server, and start the Bright Sky worker and webserver. The worker periodically polls the DWD Open Data Server for updates, parses them, and stores them in the database. The webserver will be listening to API requests on port 5000.

Architecture

Bright Sky is a rather simple project consisting of four components:

• The brightsky worker, which leverages the logic contained in the brightsky Python package to retrieve weather records from the DWD server, parse them, and store them in a database. It will periodically poll the DWD servers for new data.

• The brightsky webserver (API), which serves as gate to our database and processes all queries for weather records coming from the outside world.

• A PostgreSQL database consisting of two relevant tables:

  • sources contains information on the locations for which we hold weather records, and
  • weather contains the history of actual meteorological measurements (or forecasts) for these locations.

  The database structure can be set up by running the migrate command, which will simply apply all .sql files found in the migrations folder.

• A Redis server, which is used as the backend of the worker's task queue.

Most of the tasks performed by the worker and webserver can also be performed independently. Run docker-compose run --rm brightsky to get a list of available commands.

Hacking

Constantly rebuilding the brightsky container while working on the code can become cumbersome, and the default setting of parsing records dating all the way back to 2010 will make your development database unnecessarily large. You can set up a more lightweight development environment as follows:

1. Create a virtual environment and install our dependencies: python -m virtualenv .venv && source .venv/bin/activate && pip install -r requirements.txt && pip install -e .

2. Start a PostgreSQL container: docker-compose run --rm -p 5432:5432 postgres

3. Start a Redis container: docker-compose run --rm -p 6379:6379 redis

4. Point brightsky to your containers, and configure a tighter date threshold for parsing DWD data, by adding the following .env file:

   BRIGHTSKY_DATABASE_URL=postgres://postgres:pgpass@localhost
   BRIGHTSKY_BENCHMARK_DATABASE_URL=postgres://postgres:pgpass@localhost/benchmark
   BRIGHTSKY_REDIS_URL=redis://localhost
   BRIGHTSKY_MIN_DATE=2020-01-01
   

You should now be able to directly run brightsky commands via python -m brightsky, and changes to the source code should be effective immediately.

Tests

Large parts of our test suite run against a real Postgres database. By default, these tests will be skipped. To enable them, make sure the BRIGHTSKY_DATABASE_URL environment variable is set when calling pytest, e.g. via:

BRIGHTSKY_DATABASE_URL=postgres://postgres:pgpass@localhost/brightsky_test pytest

Beware that adding this environment variable to your .env file will not work as that file is not read by pytest. The database will be dropped and recreated on every test run, so don't use your normal Bright Sky database. ;)

Acknowledgements

Bright Sky's development is boosted by the priceless guidance and support of the Open Knowledge Foundation's Prototype Fund program, and is generously funded by Germany's Federal Ministry of Education and Research. Obvious as it may be, it should be mentioned that none of this would be possible without the painstaking, never-ending effort of the Deutscher Wetterdienst.