Library to load configuration from various sources.

Nectarine requires Python 3.7 or above, and has been tested using Python 3.7 and 3.8.

Nectarine can be installed through pip.

$ pip3 install nectarine        # Install the basic version
$ pip3 install nectarine[yaml]  # Install the YAML extension

Nectarine loads the configuration from different back-ends called providers. The table below lists the available providers.

Nectarine gives you the ability to describe a configuration using Python's dataclasses, and load it from various sources (program arguments, configuration files, ...).

Let's consider a program that consumes events from a RabbitMQ broker (for example, using the awesome pika library!). That little program should of course allow the user to configure connection parameters such as the broker's host and listening port, the credentials to use, etc.

Our program will accept configuration from 3 different providers:

• the program arguments
• the program environment
• a JSON configuration file, say ./conf.json with fields for the host / port / username / password

The idea here is to allow the user to quickly override the configuration file by using the environment or the arguments.

In order to represent those connection parameters, we create the following dataclass:

from dataclasses import dataclass

@dataclass
class RabbitMQ:
    host: str
    username: str
    password: str
    port: int = 5671

The next step is to communicate with the user, obtain the configuration, and create an instance of RabbitMQ with it. Let's see how we can achieve that with Nectarine.

Nectarine provides a load function, which needs two things: the type to load, and a list of providers to load from.

We already have our target type (which is RabbitMQ), thus we need to select our providers:

• for program arguments, the arguments provider
• for program environment, the env provider
• for the JSON file, the json provider

Now, we can call the load function like below and retrieve its result:

from nectarine import load, arguments, env, json

rabbitmq = load(
    target=RabbitMQ,
    providers=[
        arguments(),
        env(prefix="RABBITMQ_"),
        json("./conf.json")
    ]
)
print(rabbitmq)

That's pretty much it! The rabbitmq variable is now loaded with our configuration when we run the program:

$ echo '{"host": "localhost", "port": 5671, "username": "user", "password": "pass"}' > ./conf.json
$ ./consumer.py
RabbitMQ(host='localhost', username='user', password='pass', port=5671)
$ ./consumer.py --port 1234
RabbitMQ(host='localhost', username='user', password='pass', port=1234)

• Multiple configuration providers: program arguments, environment, configuration files, ...
• Value overriding: higher-priority providers override values from lower-priority providers
• Complex structures: nested dataclasses and collections are supported
• Type checking: configuration data is checked against the type hints provided in your dataclasses
• Dataclasses features: default values or factories can be specified just as in regular dataclasses