Test environment for GeoDjango projects with PostGIS and SpatiaLite support out of the box.

You can use this as a test image with BitBucket Pipelines, an example:

image: wooyek/geodjango

pipelines:
  default:
    - step:
      script:        
        # Pipelines overrides default docker entry point, we need to run it maually
        - docker-entrypoint.sh  
        - tox

Tox is is the best way to encapsulate your tests, here's an example:

[tox]
envlist = py35,py27
skipsdist = True

[testenv]
passenv =
    DJANGO_SETTINGS_MODULE
    DATABASE_PASSWORD
    DATABASE_USER
    DATABASE_HOST
    DATABASE_NAME
    DATABASE_TEST_NAME
    

setenv =
    TOX_ENVBINDIR = {envbindir}
    LIBRARY_PATH = /usr/local/lib
    CPATH=/usr/local/include
    PYTHONIOENCODING = utf-8

commands =
    coverage erase
    coverage run setup.py test
    coverage report
    coverage xml

deps =
    -rrequirements.txt
    -rrequirements-dev.txt

If you don't have Docker installed, you can use Vagrant development environment:

vagrant up
vagrant ssh

Replace wooyek/geodjango with your own image name, replace /vagrant/ with a proper path if you don't use Vagrant:

docker build -t wooyek/geodjango /vagrant/
docker push wooyek/geodjango

We can always check docker machine environment variables by this command (you should see default values set in Dockerfile):

docker run --rm wooyek/geodjango env

We can use host environment, but we still need to to pass environment variable names in a command.

I find the cleaniest easiest way is to create a separate file containing all the environment variables. Let's prepare our environment file with some variable substitution from a sample file. This is a one time operations.

ENV_FILE=/vagrant/sample/environment.env
source <(sed -E 's/[^#]+/export &/' ${ENV_FILE})
envsubst < ${ENV_FILE} | tee /vagrant/environment.env

Let's check them out:

docker run --rm --env-file /vagrant/environment.env wooyek/geodjango env

Now we can run tests from a sample project. We will map local vagrant folder to a docker volume to make them accessible from inside a running container. We'll also override DJANGO_SETTINGS_MODULE to point to all-in-one settings file, instead of more specialized website.settings.testing module used when settings are split info multiple modules:

docker run --rm -it --name qa \
    --volume=/vagrant:/vagrant \
    --workdir="/vagrant" \
    --memory=4g \
    --memory-swap=4g \
    --entrypoint=/bin/bash \
    --env-file /vagrant/environment.env \
    --hostname ${HOSTNAME} \
    -e DJANGO_SETTINGS_MODULE=website.settings \
    wooyek/geodjango \
    docker-entrypoint.sh tox -c /vagrant/sample/awesome-project

There should be a one passed test filtering on distance.

Docker images take some space, if you need to clear that out use one of those prune commands:

docker system prune
docker image prune
docker container prune