mirror of
https://github.com/ets-labs/python-dependency-injector.git
synced 2024-11-23 01:56:58 +03:00
128 lines
5.4 KiB
ReStructuredText
128 lines
5.4 KiB
ReStructuredText
Dependency injection and inversion of control in Python
|
|
-------------------------------------------------------
|
|
|
|
.. meta::
|
|
:keywords: Python,DI,Dependency injection,IoC,Inversion of Control
|
|
:description: This article describes benefits of dependency injection and
|
|
inversion of control for Python applications. Also it
|
|
contains some Python examples that show how dependency
|
|
injection and inversion could be implemented.
|
|
|
|
History
|
|
~~~~~~~
|
|
|
|
Originally, dependency injection pattern got popular in languages with static
|
|
typing, like Java. Dependency injection framework can
|
|
significantly improve flexibility of the language with static typing. Also,
|
|
implementation of dependency injection framework for language with static
|
|
typing is not something that one can do shortly, it could be quite complex
|
|
thing to be done well.
|
|
|
|
While Python is very flexible interpreted language with dynamic typing, there
|
|
is a meaning that dependency injection doesn't work for it as well, as it does
|
|
for Java. Also there is a meaning that dependency injection framework is
|
|
something that Python developer would not ever need, cause dependency injection
|
|
could be implemented easily using language fundamentals.
|
|
|
|
Discussion
|
|
~~~~~~~~~~
|
|
|
|
It is true.
|
|
|
|
Partly.
|
|
|
|
Dependency injection, as a software design pattern, has number of
|
|
advantages that are common for each language (including Python):
|
|
|
|
+ Dependency Injection decreases coupling between a class and its dependency.
|
|
+ Because dependency injection doesn't require any change in code behavior it
|
|
can be applied to legacy code as a refactoring. The result is clients that
|
|
are more independent and that are easier to unit test in isolation using
|
|
stubs or mock objects that simulate other objects not under test. This ease
|
|
of testing is often the first benefit noticed when using dependency
|
|
injection.
|
|
+ Dependency injection can be used to externalize a system's configuration
|
|
details into configuration files allowing the system to be reconfigured
|
|
without recompilation (rebuilding). Separate configurations can be written
|
|
for different situations that require different implementations of
|
|
components. This includes, but is not limited to, testing.
|
|
+ Reduction of boilerplate code in the application objects since all work to
|
|
initialize or set up dependencies is handled by a provider component.
|
|
+ Dependency injection allows a client to remove all knowledge of a concrete
|
|
implementation that it needs to use. This helps isolate the client from the
|
|
impact of design changes and defects. It promotes reusability, testability
|
|
and maintainability.
|
|
+ Dependency injection allows a client the flexibility of being configurable.
|
|
Only the client's behavior is fixed. The client may act on anything that
|
|
supports the intrinsic interface the client expects.
|
|
|
|
.. note::
|
|
|
|
While improved testability is one the first benefits of using dependency
|
|
injection, it could be easily overwhelmed by monkey-patching technique,
|
|
that works absolutely great in Python (you can monkey-patch anything,
|
|
anytime). At the same time, monkey-patching has nothing similar with
|
|
other advantages defined above. Also monkey-patching technique is
|
|
something that could be considered like too dirty to be used in production.
|
|
|
|
The complexity of dependency injection pattern implementation in Python is
|
|
definitely quite lower than in other languages (even with dynamic typing).
|
|
|
|
.. note::
|
|
|
|
Low complexity of dependency injection pattern implementation in Python
|
|
still means that some code should be written, reviewed, tested and
|
|
supported.
|
|
|
|
Talking about inversion of control, it is a software design principle that
|
|
also works for each programming language, not dependending on its typing type.
|
|
|
|
Inversion of control is used to increase modularity of the program and make
|
|
it extensible.
|
|
|
|
Main design purposes of using inversion of control are:
|
|
|
|
+ To decouple the execution of a task from implementation.
|
|
+ To focus a module on the task it is designed for.
|
|
+ To free modules from assumptions about how other systems do what they do and
|
|
instead rely on contracts.
|
|
+ To prevent side effects when replacing a module.
|
|
|
|
Example
|
|
~~~~~~~
|
|
|
|
Let's go through next example:
|
|
|
|
.. literalinclude:: ../../examples/ioc_di_demos/car_engine.py
|
|
:language: python
|
|
:linenos:
|
|
|
|
``Car`` **creates** an ``Engine`` during its creation. Really? Does it make
|
|
more sense than creating an ``Engine`` separately and then
|
|
**inject (put) it into** ``Car`` when ``Car`` is being created?
|
|
|
|
.. literalinclude:: ../../examples/ioc_di_demos/car_engine_ioc.py
|
|
:language: python
|
|
:linenos:
|
|
|
|
Previous example may look more obvious and gives a chance to start getting
|
|
other benefits of dependency injection and inversion of control, but creation
|
|
of ``Car`` instances became a bit harder cause now ``Engine`` injections
|
|
should be done manually every time when ``Car`` instances are being created.
|
|
|
|
Let's automate ``Engine`` into ``Car`` injections using *Dependency Injector*:
|
|
|
|
|
|
.. literalinclude:: ../../examples/ioc_di_demos/car_engine_ioc_container.py
|
|
:language: python
|
|
:linenos:
|
|
|
|
.. note::
|
|
|
|
``Container`` from previous example is an inversion of control container.
|
|
It contains a collection of component providers that could be injected
|
|
into each other.
|
|
|
|
Assuming this, ``Container`` could be one and the only place, where
|
|
application's structure is being managed on the high level.
|