Python application configuration

Nowadays, we have to configure many configuration parameters in our applications. These parameters sometimes depend on the execution environment.

We need to build a maintainable code whose configuration should be externalised. This way, we can run the same package in different environments. More information here

Another important concept is “cloud-native”. This concept refers to how we should develop our applications, facilitating running and configuring them in cloud environments.

In this article, we are going to see how to configure our Python applications in a “cloud-native” way.

In addition, exist other ways to configure parameters as secrets. These cover the secret’s lifecycle and how it is provided, which should be independent that our programming language.

Environment variables

The environment variables are a basic way to externalise the configuration.

To read the variables in our Python code, we will use the os library.

import os

user = os.getenv('USER')
print("[USER]:", user)

Run it:

❯ python tmp.py
[USER]: db

Configuration files

The configuration files have the advantage to manage many variables easily. These variables are in one or more files that can be replaced depending on the environment.

The python-dotenv library helps us to load the configuration files. By default, the library loads a (K, V) properties file, but we can change its format.

Firstly, install the library:

pip install python-dotenv

Now, we only have to use it to load the configuration:

from dotenv import load_dotenv
import os

load_dotenv(".env", override=True)

user = os.getenv('USER')
print("[USER]:", user)

The function load_dotenv can be configured:

  • file: by default, the library searches the file .env, but we can indicate another.
  • Override: this property indicates if we like to give priority the file variables over system variables.

Create the .env file:

❯ echo 'USER=dborrego' > .env

And, run the Python code:

❯ python tmp.py
[USER]: dborrego

References

python  cloud-native  twelve-factor 

See also