Accessing environment variables with a magic module.

>>> import os >>> from envcfg.raw.python import CONFIGURE_OPTS >>> >>> CONFIGURE_OPTS '--enable-shared --enable-universalsdk=/ --with-universal-archs=intel' >>> CONFIGURE_OPTS == os.environ['PYTHON_CONFIGURE_OPTS'] True 

It works with many frameworks such as Django and Flask. Then you can store your config in the environment variables instead of framework-specific config files. It is recommended by 12-Factor.

$ pip install python-envcfg 

• import envcfg.raw.foo as config: Import each FOO_* environment variable as string.
• import envcfg.json.foo as config: Import each FOO_* environment variable as JSON body.
• import envcfg.smart.foo as config: Try to import each FOO_* environment variable as JSON body, if fail then import it as string.

There is an example:

1. Defines environment variables with a prefix:

   $ cat .env # should not checked into VCS # values are valid JSON expressions MYAPP_DEBUG=true MYAPP_SECRET_KEY='"7950ad141c7e4b3990631fcdf9a1d909"' MYAPP_SQLALCHEMY_DATABASE_URI='"sqlite:///tmp/myapp.sqlite3"' 

2. Creates Flask app and loads config from python-envcfg:

   $ cat myapp.py ... app = Flask(__name__) app.config.from_object('envcfg.json.myapp') # MYAPP_ -> .myapp ... 

3. Enters your app with those environment variables:

   $ env $(cat .env | xargs) python myapp.py 

1. Creates a django project and moves all sensitive config items into the environment variables:

   $ cat djapp/settings.py # codebase-scope config ... INSTALLED_APPS = ( 'django.contrib.admin', ) ... $ cat .env # environment-scope config, should not checked into VCS # values are valid JSON expressions DJAPP_SECRET_KEY='"wo9g2o#jws=u"' DJAPP_DEBUG=true DJAPP_TEMPLATE_DEBUG=true 

2. Adds importing statements in the end of settings.py module:

   $ tail -n 2 djapp/settings.py # importing all config items stored in the environment variables from envcfg.json.djapp import * # noqa 

3. Runs your Django app with environment variables:

   $ env $(cat .env | xargs) python manage.py runserver 

1. Defines environment variables with a prefix:

   $ cat .env export TORAPP_PORT='8888' export TORAPP_MYSQL_HOST='"127.0.0.1"' export TORAPP_MYSQL_DATABASE='"database"' 

2. Creates a Tornado project and loads config:

   $ cat torapp/server.py from tornado.web import Application, RequestHandler from tornado.ioloop import IOLoop from tornado.options import define, options from tordb import Connection def options_from_object(*args, **kwargs): module = __import__(*args, **kwargs) for name, value in vars(module).items(): name = name.lower() if name in options._options: options._options[name].set(value) class IndexHandler(RequestHandler): def initialize(self): self.db = Connection(options.mysql_host, options.mysql_database) def get(self): pass # some database operations with ``self.db`` application = Application([ (r'/', IndexHandler), ]) define('port', type=int) define('mysql_host', type=unicode) define('mysql_database', type=unicode) options_from_object('envcfg.json.torapp', fromlist=['torapp']) if __name__ == '__main__': application.listen(options.port) IOLoop.instance().start() 

3. Runs your Tornado app:

   $ env $(cat .env | xargs) python server.py 

In development, we can work with per-project environments but no more typing source foo/bar.

I recommend to put your project-specified environment variables in {PROJECT_ROOT}/.env and mark the .env as ignored in your VCS. For example, you can write /.env in .gitignore if you are using Git, and put a .env.example as a copying template for new-cloned projects.

And then, you can use some utility such as honcho or autoenv to apply the .env automatically.

For honcho:

$ echo 'MYPROJECT_DEBUG=true' >> .env $ echo 'web: python manage.py runserver' >> Procfile $ honcho run python manage.py check-debug True $ honcho start web Starting development server at http://127.0.0.1:5000/ ... 

For autoenv:

$ echo 'MYPROJECT_DEBUG=true' >> myproject/.env $ cd myproject $ python manage.py check-debug True $ python manage.py runserver Starting development server at http://127.0.0.1:5000/ ... 

If you want to report bugs or request features, please create issues on GitHub Issues.

• https://github.com/henriquebastos/python-decouple
• https://github.com/pallets/click