Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Dependency injection the python way, the good way
Branch: master



Dependency injection the python way, the good way. Not a port of Guice or Spring.

Key features

  • Fast.
  • Thread-safe.
  • Simple to use.
  • Does not steal class constructors.
  • Does not try to manage your application object graph.
  • Transparently integrates into tests.
  • Supports Python 2.7 and Python 3.3+.



pip install inject


# Import the inject module.
import inject

# `inject.instance` requests dependencies from the injector.
def foo(bar):
    cache = inject.instance(Cache)'bar', bar)

# `inject.params` injects dependencies as keyword arguments or positional argument.
@inject.params(cache=Cache, user=CurrentUser)
def baz(foo, cache=None, user=None):'foo', foo, user)

# this can be called in different ways:
# with injected arguments

# with positional arguments
baz('foo', my_cache)

# with keyword arguments
baz('foo', my_cache, user=current_user)

# `inject.param` is deprecated, use `inject.params` instead.
@inject.param('cache', Cache)
def bar(foo, cache=None):'foo', foo)

# `inject.attr` creates properties (descriptors) which request dependencies on access.
class User(object):
    cache = inject.attr(Cache)

    def __init__(self, id): = id

    def save(self):'users', self)

    def load(cls, id):
        return cls.cache.load('users', id)

# Create an optional configuration.
def my_config(binder):
    binder.install(my_config2)  # Add bindings from another config.
    binder.bind(Cache, RedisCache('localhost:1234'))

# Configure a shared injector.

# Instantiate User as a normal class. Its `cache` dependency is injected when accessed.
user = User(10)

# Call the functions, the dependencies are automatically injected.

Usage with Django

Django can load some modules multiple times which can lead to InjectorException: Injector is already configured. You can use configure_once which is guaranteed to run only once when the injector is absent:

import inject


In tests use inject.clear_and_configure(callable) to create a new injector on setup, and optionally inject.clear() to clean up on tear down:

class MyTest(unittest.TestCase):
    def setUp(self):
        inject.clear_and_configure(lambda binder: binder
            .bind(Cache, Mock() \
            .bind(Validator, TestValidator())

    def tearDown(self):


After configuration the injector is thread-safe and can be safely reused by multiple threads.

Binding types

  • Instance bindings always return the same instance:

    redis = RedisCache(address='localhost:1234')
    def config(binder):
        binder.bind(Cache, redis)
  • Constructor bindings create a singleton on injection:

    def config(binder):
        # Creates a redis cache singleton on first injection.
        binder.bind_to_constructor(Cache, lambda: RedisCache(address='localhost:1234'))
  • Provider bindings call the provider on injection:

    def get_my_thread_local_cache():
    def config(binder):
        # Executes the provider on each injection.
        binder.bind_to_provider(Cache, get_my_thread_local_cache)
  • Runtime bindings automatically create singletons on injection, require no configuration. For example, only the Config class binding is present, other bindings are runtime:

    class Config(object):
    class Cache(object):
        config = inject.attr(Config)
    class Db(object):
        config = inject.attr(Config)
    class User(object):
        cache = inject.attr(Cache)
        db = inject.attr(Db)
        def load(cls, user_id):
            return cls.cache.load('users', user_id) or cls.db.load('users', user_id)
    inject.configure(lambda binder: binder.bind(Config, load_config_file()))
    user = User.load(10)


It is possible to use any hashable object as a binding key. For example:

import inject

inject.configure(lambda binder: \
    binder.bind('host', 'localhost') \
    binder.bind('port', 1234))

Why no scopes?

I've used Guice and Spring in Java for a lot of years, and I don't like their scopes. python-inject by default creates objects as singletons. It does not need a prototype scope as in Spring or NO_SCOPE as in Guice because python-inject does not steal your class constructors. Create instances the way you like and then inject dependencies into them.

Other scopes such as a request scope or a session scope are fragile, introduce high coupling, and are difficult to test. In python-inject write custom providers which can be thread-local, request-local, etc.

For example, a thread-local current user provider:

import inject
import threading

# Given a user class.
class User(object):

# Create a thread-local current user storage.
_LOCAL = threading.local()

def get_current_user():
    return getattr(_LOCAL, 'user', None)

def set_current_user(user):
    _LOCAL.user = user

# Bind User to a custom provider.
inject.configure(lambda binder: binder.bind_to_provider(User, get_current_user))

# Inject the current user.
def foo(user):



Apache License 2.0

Something went wrong with that request. Please try again.