pydic 1.0.0

What is a Service Container

A Service Container (or dependency injection container) is simply a python object that manages the instantiation of services (objects). For example, suppose you have a simple python class that delivers email messages. Without a service container, you must manually create the object whenever you need it:

from myapplication.mailer import Mailer

mailer = Mailer('sendmail')
mailer.send('felix@example.com', ...)

This is easy enough. The imaginary Mailer class allows you to configure the method used to deliver the email messages (e.g. sendmail, smtp, etc).

But what if you wanted to use the mailer service somewhere else? You certainly don’t want to repeat the mailer configuration every time you need to use the Mailer object. What if you needed to change the transport from sendmail to smtp everywhere in the application? You’d need to hunt down every place you create a Mailer service and change it.

The Services container allows you to standardize and centralize the way objects are constructed in your application.

Creating/Configuring Services in the Container

A better answer is to let the service container create the Mailer object for you. In order for this to work, you must teach the container how to create the Mailer service. This is done via configuration definitions:

...
from pydic import Services

definitions = {
    'my_mailer': {
        'class':        'myapplication.mailer.Mailer',
        'arguments':    ['sendmail']
    }
}

services = Services(definitions)
...

When you ask for the my_mailer service from the container services.get('my_mailer'), the container constructs the object and returns it.

This is another major advantage of using the service container. Namely, a service is never constructed until it’s needed. If you define a service and never use it, the service is never created. This saves memory and increases the speed of your application. This also means that there’s very little or no performance hit for defining lots of services. Services that are never used are never constructed.

As an added bonus, the Mailer service is only created once and the same instance is returned each time you ask for the service. This is almost always the behavior you’ll need (it’s more flexible and powerful).

You can pass the arguments as list or dict.

Also you can call functions after object instantiation with:

...
definitions = {
    'my_mailer': {
        'class':        'myapplication.mailer.Mailer',
        'arguments':    ['sendmail'],
        'calls': [
            [ 'set_name', 'Felix Carmona'],
            [ 'inject_something',  [1, 2, 3]],
            [ 'inject_something',  [2, 3]],
            [ 'set_location',  {'city': 'Barcelona', 'country': 'Spain'}]
        ]
    }
}
...

Once the container has been constructed with the definitions, the available methods for the service container object are:

• set(key, value): Sets a service object by name.

• get(key): Returns a service object by name.

• has(key): Returns True if the service definition exists or if the service object is instantiated, False otherwise.

• remove(key): Removes a service object and service definition by name.

• add(parameters): Adds a dict of services objects.

• keys(): Returns the services keys.

Using the Parameters to build Services

The creation of new services (objects) via the container is pretty straightforward. Parameters make defining services more organized and flexible:

...
parameters = Parameters(
    {
        'my_mailer_class':     'myapplication.mailer.Mailer',
        'my_mailer_transport': 'sendmail'
    }
)

definitions = {
    'my_mailer': {
        'class':        '{{ my_mailer_class }}',
        'arguments':    ['{{ my_mailer_transport }}']
    }
}

services = Services(definitions, parameters)
...

The end result is exactly the same as before - the difference is only in how you defined the service. By surrounding the my_mailer.class and my_mailer.transport strings in double bracket keys ({{ }}) signs, the services container knows to look for parameters with those names. Parameters can deep reference other parameters that references other parameters, and will be resolved anyway.

The purpose of parameters is to feed information into services. Of course there was nothing wrong with defining the service without using any parameters. Parameters, however, have several advantages:

• separation and organization of all service “options” under a single parameters key

• parameter values can be used in multiple service definitions

The choice of using or not using parameters is up to you.

Referencing (Injecting) Services

You can of course also reference services

Start the string with @ to reference a service, example:

...
parameters = Parameters(
    {
        'my_mailer_class':     'myapplication.mailer.Mailer',
        'my_mailer_transport': 'sendmail'
    }
)

definitions = {
    'my_mailer': {
        'class':        '{{ my_mailer_class }}',
        'arguments':    ['{{ my_mailer_transport }}']
    },
    'my_mailer_manager': {}
        'class': 'myapplication.mailer.MailerManager',
        'arguments': ['@my_mailer']
    }
}

services = Services(definitions, parameters)
...

the my_mailer service will be injected in the my_mailer_manager

pydic is open-sourced software licensed under the MIT license