Skip to main content

Transparent dependency injection.

Project description

https://img.shields.io/pypi/v/antidote.svg https://img.shields.io/pypi/l/antidote.svg https://img.shields.io/pypi/pyversions/antidote.svg https://travis-ci.org/Finistere/antidote.svg?branch=master https://codecov.io/gh/Finistere/antidote/branch/master/graph/badge.svg https://readthedocs.org/projects/antidote/badge/?version=latest

Antidotes is a declarative dependency injection micro-framework for Python 3.5+ designed for ease of use.

Why Dependency Injection ?

In short antidote avoids you the hassle of instantiating and managing your services. You declare them at their definition, and inject them wherever needed with simple decorators, which do not change how you interact with your objects. Unit testing is not impacted as one can override any injection and control the available dependencies easily.

For the longer version: https://antidote.readthedocs.io/en/stable/why.html

Why Antidote ?

While there are several dependency injection libraries, there was none which really convinced me. Most of them did not satisfy all of those requirements:

  • Use of type hints: Be consistent with type hints as supported by mypy and use them to inject dependencies. Other means to inject dependencies should be possible.
  • Maturity: Support different kind of dependencies, decent test coverage.
  • Easy to integrate with existing code: Introducing the library in a existing application implies changing the whole wiring of the services/configuration. It should be as easy as possible.
  • Not encourage magic: For example, using the arguments name implicitly, so by default, to find dependencies is magic. While having this possibility is good, as it can make sense in some cases, the library shouldn’t do this by default.
  • Be nice with developers and their IDE: Use of type hints in the library, no **kwargs for function arguments (so auto-completion works), should be as easy as possible to find definition of dependencies with a right click and “Go to definition”, etc…

Features Highlight

Core functionalities:

  • Injection of all kinds of functions (method, classmethod, bound or not, …) through type hints and optionally from argument’s name and/or with explicitly specified dependencies.
  • Dependency cycle detection
  • Thread-safety and limited performace impact (see injection benchmark).
  • Easily extendable, through dependency providers. All after-mentioned kind of dependencies are implemented with it. It is designed to support custom kind of dependencies from the ground up. So if you want custom magic or whatever, you can have it !

Kind of dependencies:

  • Services and factories: provides an instance of a class.
  • Tags: Dependencies can be tagged, and as such all of them matching a specific tag can be retrieved.
  • Configuration: Constants which are lazily evaluated.
  • Lazy function calls: Results of a function call is lazily provided.
  • Defintion of interface and services implementing those.

Installation

To install Antidote, simply run this command:

pip install antidote

Quick Start

How does injection looks like ? Here is a simple example:

from antidote import inject, register

# Declare Service as a dependency that can be injected
@register
class Service:
    pass

 # uses the type hint
 @inject
 def f(service: Service):
     pass

 f()  # Service will be automatically injected if not provided
 f(Service())  # Want to override injection for tests ? easy

 # Explicitly provide the dependency
 @inject(dependencies=dict(service=Service))
 def f(service):
     pass

 # uses the position of the arguments
 @inject(dependencies=(Service,))
 def f(service):
     pass

Want more ? Here is a more complete example with configurations, services, factories:

"""
Simple example where a MovieDB interface is defined which can be used
to retrieve the best movies. In our case the implementation uses IMDB
to dot it.
"""
from functools import reduce

import antidote


class MovieDB:
    def get_best_movies(self):
        pass


class ImdbAPI:
    """
    Class from an external library.
    """

    def __init__(self, *args, **kwargs):
        """ Initializes the IMDB API. """


# Usage of constants for configuration makes refactoring easier and is
# less error-prone. Moreover Conf will only be instantiated if necessary.
class Conf(metaclass=antidote.LazyConstantsMeta):
    # The metaclass adds custom behavior for constants (upper case attributes).
    # Conf.IMDB_HOST is a dependency id
    # but Conf().IMDB_HOST is the actual value making it easy to work with.
    IMDB_HOST = 'imdb.host'
    IMDB_API_KEY = 'imdb.api_key'

    def __init__(self):
        # Load configuration from somewhere
        self._raw_conf = {
            'imdb': {
                'host': 'dummy_host',
                'api_key': 'dummy_api_key'
            }
        }

    def get(self, key):
        """ 'a.b' -> self._raw_conf['a']['b'] """
        return reduce(dict.get, key.split('.'), self._raw_conf)


# Declare a factory which should be called to instantiate Database.
# The order of the arguments is here used to map the dependencies.
# A dictionary mapping arguments name to their dependency could also
# have been used.
@antidote.factory(dependencies=(Conf.IMDB_HOST, Conf.IMDB_API_KEY))
def imdb_factory(host: str, api_key: str) -> ImdbAPI:
    """
    Configure your database.
    """
    return ImdbAPI(host=host, api_key=api_key)


# implements specifies that IMDBMovieDB should be used whenever MovieDB is requested.
@antidote.implements(MovieDB)
# Registering IMDBMovieDB makes it available in Antidote. (required for @implements)
@antidote.register
class IMDBMovieDB(MovieDB):
    # Here the dependencies of __init__() are injected by default as @register treats
    # it as the factory of the service.
    # Note that IMDBMovieDB does not build itself ImdbAPI, which makes testing
    # easier.
    def __init__(self, imdb_api: ImdbAPI):
        self._imdb_api = imdb_api

    def get_best_movies(self):
        pass


# Inject dependencies in f(), by default only type annotations are used. But
# arguments name, explicit mapping, etc.. can also be used.
@antidote.inject
def f(movie_db: MovieDB):
    """ Do something with your database. """


# Can be called without arguments now.
f()

assert antidote.world.get(MovieDB) is antidote.world.get(IMDBMovieDB)

# You can still explicitly pass the arguments to override
# injection.
conf = Conf()
f(IMDBMovieDB(imdb_factory(
    # equivalent to conf._raw_conf['db.host'], mainly to make your tests easier.
    host=conf.IMDB_HOST,
    api_key=conf._raw_conf['imdb']['api_key'],
)))

Interested ? Check out the documentation or try it directly ! There are still features left such as tags or custom kinds of dependencies.

Documentation

The documentation is available at https://antidote.readthedocs.io/en/stable.

Injection benchmark is available at injection benchmarks.

Bug Reports / Feature Requests

Any feedback is always welcome, feel free to submit issues and enhancement requests ! :) For any questions, open an issue on Github.

How to Contribute

  1. Check for open issues or open a fresh issue to start a discussion around a feature or a bug.
  2. Fork the repo on GitHub. Run the tests to confirm they all pass on your machine. If you cannot find why it fails, open an issue.
  3. Start making your changes to the master branch.
  4. Writes tests which shows that your code is working as intended. (This also means 100% coverage.)
  5. Send a pull request.

Be sure to merge the latest from “upstream” before making a pull request!

Pull requests should avoid to:

  • make it harder to integrate Antidote into existing code.
  • break backwards compatibility.
  • create features difficult to understand for an IDE, such as converting a string dependency id to a non singleton object somehow. An user may do this, but antidote shouldn’t.

Pull requests will not be accepted if:

  • classes and non trivial functions have not docstrings documenting their behavior.
  • tests do not cover all of code changes.

Do not hesitate to send a pull request, even if incomplete, to get early feedback ! :)

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for antidote, version 0.7.3
Filename, size File type Python version Upload date Hashes
Filename, size antidote-0.7.3-cp35-cp35m-manylinux1_i686.whl (940.5 kB) File type Wheel Python version cp35 Upload date Hashes View
Filename, size antidote-0.7.3-cp35-cp35m-manylinux1_x86_64.whl (1.2 MB) File type Wheel Python version cp35 Upload date Hashes View
Filename, size antidote-0.7.3-cp35-cp35m-manylinux2014_i686.whl (940.5 kB) File type Wheel Python version cp35 Upload date Hashes View
Filename, size antidote-0.7.3-cp35-cp35m-manylinux2014_x86_64.whl (1.2 MB) File type Wheel Python version cp35 Upload date Hashes View
Filename, size antidote-0.7.3-cp36-cp36m-manylinux1_i686.whl (971.0 kB) File type Wheel Python version cp36 Upload date Hashes View
Filename, size antidote-0.7.3-cp36-cp36m-manylinux1_x86_64.whl (1.2 MB) File type Wheel Python version cp36 Upload date Hashes View
Filename, size antidote-0.7.3-cp36-cp36m-manylinux2014_i686.whl (971.0 kB) File type Wheel Python version cp36 Upload date Hashes View
Filename, size antidote-0.7.3-cp36-cp36m-manylinux2014_x86_64.whl (1.2 MB) File type Wheel Python version cp36 Upload date Hashes View
Filename, size antidote-0.7.3-cp37-cp37m-manylinux1_i686.whl (970.0 kB) File type Wheel Python version cp37 Upload date Hashes View
Filename, size antidote-0.7.3-cp37-cp37m-manylinux1_x86_64.whl (1.2 MB) File type Wheel Python version cp37 Upload date Hashes View
Filename, size antidote-0.7.3-cp37-cp37m-manylinux2014_i686.whl (970.0 kB) File type Wheel Python version cp37 Upload date Hashes View
Filename, size antidote-0.7.3-cp37-cp37m-manylinux2014_x86_64.whl (1.2 MB) File type Wheel Python version cp37 Upload date Hashes View
Filename, size antidote-0.7.3-cp38-cp38-manylinux1_i686.whl (1.0 MB) File type Wheel Python version cp38 Upload date Hashes View
Filename, size antidote-0.7.3-cp38-cp38-manylinux1_x86_64.whl (1.4 MB) File type Wheel Python version cp38 Upload date Hashes View
Filename, size antidote-0.7.3-cp38-cp38-manylinux2014_i686.whl (1.0 MB) File type Wheel Python version cp38 Upload date Hashes View
Filename, size antidote-0.7.3-cp38-cp38-manylinux2014_x86_64.whl (1.4 MB) File type Wheel Python version cp38 Upload date Hashes View
Filename, size antidote-0.7.3.tar.gz (66.5 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page