Dependency Injector is a dependency injection framework for Python.

Hello,

all your docs are stored in a HTTP server, however, many company proxies/AntiVir block any HTTP site.

Just mentioning that moving to a https will make your docs available to a bigger audience (including myself :)).

Regards, A.

Hi,

Now that python typing (type hints) is very common, it would be great that dependency-injector will have type hints / annotations. Currently when using Container / Provider classes mypy complains since it doesn't have information about a provider returned type, etc.

For example:

provider: Provider[Animal] = providers.Factory(Cat)

animal = provider()  # mypy can know that animal is of type Animal

My current workaround is to put type: ignore in many places but it is less desired since I won't get errors at type checking time.

Thanks

Hi, I just run into this issue with the Configuration provider. After scratching my head for a bit, I managed to find a workaround, but I was wondering if this is actually a bug or just something wrong I am doing. Any help would be appreciated!

Steps to reproduce

containers.py

from dependency_injector import providers, containers

class MyService(object):
    def __init__(self, **kwargs):
        self.key = kwargs.pop('key')

    def trigger(self):
        pass

class MyDevice(object):
    def __init__(self, **kwargs):
        # doesn't raise an error because it's an instance of
        # dependency_injector.providers.Singleton
        self.service = kwargs.pop('service')

    def do_something(self):
        # raises "AttributeError: 'NoneType' object has no attribute 'get'"
        self.service().trigger()

class ServiceContainer(containers.DeclarativeContainer):
    config = providers.Configuration()

    myservice = providers.Singleton(MyService, config=config.myservice)


class Container(containers.DeclarativeContainer):
    config = providers.Configuration()

    services = providers.Container(ServiceContainer, config=config.services)

    mydevice = providers.Factory(MyDevice)

If I run app.py

import sys

from containers import Container

container = Container()
container.config.from_yaml('config.yaml')
container.init_resources()
container.wire(modules=[sys.modules[__name__]])

mydevice = container.mydevice(service=container.services.myservice)
mydevice.do_something()

with config.yaml

foo:
  bar: 42

it raises the following error

File "/home/stefano/personal/test-error/containers.py", line 15, in do_something self.service().trigger() File "src/dependency_injector/providers.pyx", line 168, in dependency_injector.providers.Provider.call File "src/dependency_injector/providers.pyx", line 2245, in dependency_injector.providers.Singleton._provide File "src/dependency_injector/providers.pxd", line 550, in dependency_injector.providers.__factory_call File "src/dependency_injector/providers.pxd", line 536, in dependency_injector.providers.__callable_call File "src/dependency_injector/providers.pxd", line 495, in dependency_injector.providers.__call File "src/dependency_injector/providers.pxd", line 387, in dependency_injector.providers.__provide_keyword_args File "src/dependency_injector/providers.pxd", line 310, in dependency_injector.providers.__get_value File "src/dependency_injector/providers.pyx", line 168, in dependency_injector.providers.Provider.call File "src/dependency_injector/providers.pyx", line 1232, in dependency_injector.providers.ConfigurationOption._provide File "src/dependency_injector/providers.pyx", line 1467, in dependency_injector.providers.Configuration.get AttributeError: 'NoneType' object has no attribute 'get'

Workaround To avoid the issue, I have to pass the whole config to ServiceContainer

class ServiceContainer(containers.DeclarativeContainer):
    config = providers.Configuration()

    myservice = providers.Singleton(MyService, config=config.services.myservice)


class Container(containers.DeclarativeContainer):
    config = providers.Configuration()

    services = providers.Container(ServiceContainer, config=config)

    mydevice = providers.Factory(MyDevice)

Running the application now, raises the following (as expected)

File "/home/stefano/personal/test-error/containers.py", line 18, in do_something self.service().trigger() File "src/dependency_injector/providers.pyx", line 168, in dependency_injector.providers.Provider.call File "src/dependency_injector/providers.pyx", line 2245, in dependency_injector.providers.Singleton._provide File "src/dependency_injector/providers.pxd", line 550, in dependency_injector.providers.__factory_call File "src/dependency_injector/providers.pxd", line 536, in dependency_injector.providers.__callable_call File "src/dependency_injector/providers.pxd", line 526, in dependency_injector.providers.__call File "/home/stefano/personal/test-error/containers.py", line 5, in init self.key = kwargs.pop('key') KeyError: 'key'

I am not quite sure if this is expected behavior or not. Methods annotated as @classmethod end up getting extra parameters injected. The following code demonstrates. I discovered this while using Closing, but filled out the example a bit as I discovered that it is a general issue for Provide.

import sys
from dependency_injector import containers, providers
from dependency_injector.wiring import Provide, Closing


def my_factory():
    return 'test-factory'


def my_resource():
    yield 'test-resource'
    print('Closing')


class Container(containers.DeclarativeContainer):

    factory = providers.Factory(my_factory)
    resource = providers.Resource(my_resource)


def do_function_thing(r:str=Closing[Provide[Container.resource]]) -> None:
    print('from function', r)


class MyClass():

    def do_instance_thing(self, r:str=Closing[Provide[Container.resource]]) -> None:
        print('from instance', r)

    @classmethod
    def do_class_thing(cls, r:str=Closing[Provide[Container.resource]]) -> None:
        print('from class', r)

    @classmethod
    def non_closing_class_thing(cls, r:str=Provide[Container.factory]) -> None:
        print('non-closing from class', r)


container = Container()
container.init_resources()
container.wire(modules=[sys.modules[__name__]])


do_function_thing()
c = MyClass()
c.do_instance_thing()

# both of these end up getting multiple values for r:
c.non_closing_class_thing()
c.do_class_thing()

The resulting output is:

from function test-resource
Closing
from instance test-resource
Closing
Traceback (most recent call last):
  File "clstest.py", line 49, in <module>
    c.non_closing_class_thing()
  File "/Users/scott/repos/github.com/scott2b/Starlight/.venv/lib/python3.8/site-packages/dependency_injector/wiring.py", line 296, in _patched
    result = fn(*args, **to_inject)
TypeError: non_closing_class_thing() got multiple values for argument 'r'

I have not been able to find any documentation/examples on how to setup an endpoint using dynamic containers. All i've seen is: def create_transaction( dbengine: MyDbEngine = Depends(Provide[DiContainer.dbengine_provider]) )

Of course this doesn't work for dynamic containers since i have no declarative container called DiContainer. Am i correct here? I'm sure i'm missing something here being new to python etc...

thx ~MarvinB

I'm hoping this is just because I'm new at this.

I'm developing an application, and currently working on converting to use dependency-injector. I've written a container and have been able to get configuration and a data store factory working. However, I'm running into an issue when wiring up injections with a Flask blueprint.

When I follow the documentation and call the function to wire the container, I get a RuntimeError (actually from Flask):

Traceback (most recent call last):
  File "c:\source\personal\pydas\src\pydas\__init__.py", line 61, in create_app
    app.container.wire(
  File "src/dependency_injector/containers.pyx", line 250, in dependency_injector.containers.DynamicContainer.wire
  File "C:\Source\Personal\pyDAS\.env\lib\site-packages\dependency_injector\wiring.py", line 238, in wire
    if inspect.isfunction(member):
  File "C:\Python38\lib\inspect.py", line 169, in isfunction
    return isinstance(object, types.FunctionType)
  File "C:\Source\Personal\pyDAS\.env\lib\site-packages\werkzeug-2.0.0rc1-py3.8.egg\werkzeug\local.py", line 379, in __get__
    obj = instance._get_current_object()
  File "C:\Source\Personal\pyDAS\.env\lib\site-packages\werkzeug-2.0.0rc1-py3.8.egg\werkzeug\local.py", line 499, in _get_current_object
    return self.__local()  # type: ignore
  File "C:\Source\Personal\pyDAS\.env\lib\site-packages\flask-1.1.2-py3.8.egg\flask\globals.py", line 38, in _lookup_req_object
    raise RuntimeError(_request_ctx_err_msg)
RuntimeError: Working outside of request context.

This typically means that you attempted to use functionality that needed
an active HTTP request.  Consult the documentation on testing for
information about how to avoid this problem.

Although it's from Flask, it is only raised when calling the wire function, passing in the modules to my blueprints. I've refactored my code to no longer use current_app, thinking that might have made an impact. My primary app factory can be found here: https://github.com/bvanfleet/pydas/blob/u/bvanfleet/di.support/src/pydas/init.py

Has this ever been encountered before? I tried searching through the other issues, but couldn't find an obvious answer.

Hi, for my usecase I want to use aio_pika

And my singleton resource connection is created via:

connection = await connect(
        "amqp://guest:guest@localhost/", loop=loop
    )

and shutdown via

await connection.close()

but it looks like something like this is not supported yet:

async def aio_pika_connection():
 connection = await connect(
        "amqp://guest:guest@localhost/", loop=loop
    )
 yield connection
 await connection.close()

class Container(containers.DeclarativeContainer):

    config = providers.Configuration()

    aio_pika_connection= providers.Resource(
        aio_pika_connection
    )

Any hints how to deal with this?

Hi,

I use this module daily at work and for me fits perfectly! Thanks so much for your work!

Question: Is there any way to load ini file directly to dependency_injector.providers.Configuration? Currently I solve this by implementing in Core my configuration loader and then override config values. Other way is to use directly python ConfigParser, load ini file and then override Core.config. I don't want to use directly ConfigParser because I had too much customized this class.

So there is another question: Is there any awy to use customized ConfigParser to paste parameters from ini to Core.config?

best regards!

Hello, I have a class that is in charge to load some pickle files for a "machine learning" purpose.

Specifically, I call my class, called Regressor, in this way, through the usage of a .ini file reporting all the desired settings:

from dependency_injector import containers, providers
from dependency_injector.wiring import inject, Provide
from src.service.costestimator.configuration.configuration import Container

class Regressor():
    def __init__(self,
                 regressors_list = Provide[Container.config.regressors_scalers.regressors_list.as_(str)],
                 regressors_path = Provide[Container.config.regressors_scalers.regressors_path.as_(str)],
                 scalers_path = Provide[Container.config.regressors_scalers.scalers_path.as_(str)],
                 file_type_regressors = Provide[Container.config.regressors_scalers.file_type_regressors.as_(str)],
                 file_type_scalers = Provide[Container.config.regressors_scalers.file_type_scalers.as_(str)]):

When it comes to the "core" part of the code, it fails to load the pickle file:

    def load_model(self):
        if self.regressors_out is None:
            resulting_reg = {}
            resulting_scal = {}
            
            for regressor in self.regressors_list:
                with open(self.regressors_path.joinpath(self.regressors_path, f'{regressor}.{self.file_type_regressors}'), 'rb') as reg:
                    resulting_reg[regressor] = pickle.load(reg)

The error I get is the following: ImportError: cannot import name '_ccallback_c' from 'scipy._lib'

The funny part is that: if I instantiate my Regressor class using the "Factory approach", it perfectly works.

Any idea? Thank you in advance for any response

Hello there, I noticed that an explicit import of a package works smoothly, while a dinamic import causes the wire function to not work properly

When it works:

`from dependency_injector import containers, providers from dependency_injector.wiring import inject, Provide from pathlib import Path

from src.service.costestimator.costestimator import costestimator as CE

if name == 'main': service = 'costestimator' customer = 'john_doe' plant = '0'

# Retrieving paths
root_dir = Path.cwd().parent
ini_file = service + '_' + customer + '_' + plant + '.ini'
ini_path = root_dir.joinpath(root_dir, 'config', 'params', ini_file)

# Loading needed libraries
container = importlib.import_module('src.service.' + service + '.configuration.configuration')
configuration = container.Container()
configuration.config.from_ini(ini_path)
configuration.wire(modules=[sys.modules[__name__]])

CE_1 = CE()`

In this case, the output is constituted by prints of fields coming from an .ini file (i.e.: the desired output).

In the following case, it does not work: ` from dependency_injector import containers, providers from dependency_injector.wiring import inject, Provide from pathlib import Path import importlib

if name == 'main': service = 'costestimator' customer = 'john_doe' plant = '0'

# Retrieving paths
root_dir = Path.cwd().parent
ini_file = service + '_' + customer + '_' + plant + '.ini'
ini_path = root_dir.joinpath(root_dir, 'config', 'params', ini_file)

# Loading needed libraries
container = importlib.import_module('src.service.' + service + '.configuration.configuration')
configuration = container.Container()
configuration.config.from_ini(ini_path)
configuration.wire(modules=[sys.modules[__name__]])

getattr(importlib.import_module('src.service.' + service + '.' + service), service)()

`

in this case, the result is a print of the Provide objects.

The init function of the class costestimator is written as: class costestimator(): def __init__(self, customer_name = Provide[Container.config.customer.customer_name.as_(str)], customer_plant = Provide[Container.config.customer.customer_plant.as_(str)]):

I am trying to create a container that optionally takes a dependency, and otherwise provides a value derived from another provider. The (IMO) hacky solution I have so far is a custom provider which either provides a Callable, or a default value if the callable has an error. Then I use this with the Callable being the dependency provider.

My questions are (1) is there a better way? and (2) even using this method, DefaultCallable defined below seems like a hack -- how can I improve?

T = TypeVar("T")

class DefaultCallable(providers.Provider):

    __slots__ = ("_callable", "_default")

    def __init__(
        self, callable: Callable[..., T], default: T, *args, **kwargs
    ):
        self._default = default
        self._callable = providers.Callable(callable, *args, **kwargs)

        super().__init__()

    def __deepcopy__(self, memo):
        copied = memo.get(id(self))
        if copied is not None:
            return copied

        # TODO: type?
        copied = self.__class__(
            cast(Callable[..., T], self._callable.provides),
            providers.deepcopy(self._default, memo),
            *providers.deepcopy(self._callable.args, memo),
            **providers.deepcopy(self._callable.kwargs, memo),
        )
        self._copy_overridings(copied, memo)
        return copied

    def _provide(self, args, kwargs):
        try:
            return self._callable(*args, **kwargs)
        except Exception:
            # TODO: why do we need to check if is provider?
            # type?
            if getattr(cast(Any, self._default), "__IS_PROVIDER__", False):
                return cast(Any, self._default)()
            else:
                return self._default

# Used like

class Foo(containers.DeclarativeContainer):

    #: specify dv for pattern discovery (optional)
    dv_in: Provider[xr.DataArray] = providers.Dependency(
        instance_of=xr.DataArray
    )

    #: dv for pattern discovery (specified or default)
    dv: Provider[xr.DataArray] = DefaultCallable(
        # cast(Callable[..., xr.DataArray], dv_in), type??
        cast(Any, dv_in),
        problem.training.provided["dv"],
    )

Trying to understand how to correctly inject dependencies into a class constructor, the following is a sandbox example where I've extracted the key parts from a broader project.

from dependency_injector import containers, providers
from dependency_injector.wiring import Provide, inject

class Config():
    def __init__(self, config_file: str = None):
        self.config_file = config_file

class Container(containers.DeclarativeContainer):
    config = providers.Singleton(Config)

class Engine():
    @inject
    def __init__(
        self,
        config: Config = Provide[Container.config],
    ):
        self.config = config

    def init(self) -> None:
        print(f'Reading from config: {self.config.config_file}')

def container_factory(config_file: str = None) -> Container:
    container = containers.DynamicContainer()
    config_provider = providers.Factory(Config, config_file=config_file) \
        if config_file else providers.Factory(Config)
    container.config = config_provider
    return container

def main():
    container = container_factory('/tmp/config.yml')
    container.wire(modules=[__name__])
    engine = Engine()
    engine.init()

if __name__ == '__main__':
    main()

I'm getting an error here from the init log: AttributeError: 'Provide' object has no attribute 'config_file'

My expected behavior is that config is an instance of the Config class that has been passed the config_file value in its constructor. Instead I seem to be given an instance of the Provide class?

I have a few questions about this:

• The documentation is very vague about what wire actually does, and what it expects for its modules/packages arguments. What should these values be? The module and/or package that needs to have dependencies injected?
• I'm assuming the problem here is that the wire has failed so it's unable to inject correctly?
• I'm having a hard time finding examples that demonstrate where the @inject decorator needs to go for a class -- is it on the class declaration or on the actual __init__ def, if I wish to inject into the constructor?

If it's relevant, I use virtualenv .venv to create a virtual env at the root of the project and am using poetry install and poetry shell to load the environment and run it.

I have a dependency injection container that holds a coroutine. This coroutine takes some object as a parameter. Also that object is managed by the container - it is created via a factory method but the method itself is asynchronous. The code snippet below depicts clearly how it is done:

import asyncio

from dependency_injector import containers, providers


async def create_a():
    await asyncio.sleep(1)
    return 1


async def run(a: int):
    await asyncio.sleep(1)
    return a


class Container(containers.DeclarativeContainer):
    a = providers.Factory(
        create_a
    )

    run_a = providers.Coroutine(
        run,
        a=a,
    )


async def main():
    container = Container()
    asyncio.create_task(container.run_a())


if __name__ == "__main__":
    loop = asyncio.new_event_loop()
    asyncio.set_event_loop(loop)
    loop.run_until_complete(main())

If create_a was a normal synchronous function, then the code above would run as expected. However, with an async create_a an exception is raised while trying to create a task for container.run():

Traceback (most recent call last):
  File "C:\Users\-\Desktop\coro\main.py", line 35, in <module>
    loop.run_until_complete(main())
  File "C:\Python310\lib\asyncio\base_events.py", line 641, in run_until_complete
    return future.result()
  File "C:\Users\-\Desktop\coro\main.py", line 29, in main
    asyncio.create_task(container.run())
  File "C:\Python310\lib\asyncio\tasks.py", line 337, in create_task
    task = loop.create_task(coro)
  File "C:\Python310\lib\asyncio\base_events.py", line 433, in create_task
    task = tasks.Task(coro, loop=self, name=name)
TypeError: a coroutine was expected, got <Future pending>

At first glance, I thought that I need to await container.run() to acquire the actual coroutine object and then pass it to asyncio.create_task() but this way await just runs the run coroutine and blocks execution which is not what I want. What am I doing wrong?

class Service:
    ...

from dependency_injector import containers, providers


class Container(containers.DeclarativeContainer):
    # obtain `environment` as a passed in parameter
    environment = providers.Dependency()

    # utilise the provided `environment`
    service = providers.Factory(Service, env=environment.provided)

from dependency_injector.wiring import Provide, inject


@inject
def start(service: Service = Provide['service']) -> None:
    ...


if __name__ == "__main__":
    # example of passing a param into our container
    container = Container(environment='DEBUG')
    container.wire(modules=[__name__])

    start()

Given the above example, with a dependency provided as environment I would like to know if something like this exists within the dependency_injector package, I had a look at the documentation and couldn't really figure out a way to achieve this, and this is as far as I got which doesn't work

Traceback (most recent call last): File "", line 1178, in _find_and_load File "", line 1149, in _find_and_load_unlocked File "", line 690, in _load_unlocked File "", line 940, in exec_module File "", line 241, in _call_with_frames_removed File "D:\Projects\python\user-test\user\route\user_route.py", line 17, in @user_router.get( ^^^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\site-packages\fastapi\routing.py", line 630, in decorator self.add_api_route( File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\site-packages\fastapi\routing.py", line 569, in add_api_route route = route_class( ^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\site-packages\fastapi\routing.py", line 442, in init get_parameterless_sub_dependant(depends=depends, path=self.path_format), ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\site-packages\fastapi\dependencies\utils.py", line 135, in get_parameterless_sub_dependant return get_sub_dependant(depends=depends, dependency=depends.dependency, path=path) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\site-packages\fastapi\dependencies\utils.py", line 158, in get_sub_dependant sub_dependant = get_dependant( ^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\site-packages\fastapi\dependencies\utils.py", line 281, in get_dependant endpoint_signature = get_typed_signature(call) ^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\site-packages\fastapi\dependencies\utils.py", line 249, in get_typed_signature signature = inspect.signature(call) ^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\inspect.py", line 3278, in signature return Signature.from_callable(obj, follow_wrapped=follow_wrapped, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\inspect.py", line 3026, in from_callable return _signature_from_callable(obj, sigcls=cls, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Eugene\AppData\Local\Programs\Python\Python311\Lib\inspect.py", line 2615, in _signature_from_callable raise ValueError('callable {!r} is not supported by signature'.format(obj)) ValueError: callable <dependency_injector.providers.Factory(<class 'user.infrastructure.query.user_uow.UserUnitOfWork'>) at 0x27549c2cc40> is not supported by signature python-BaseException

When a function is decorated with @inject the wrapped function is inspected for Provide[...] arguments:

https://github.com/ets-labs/python-dependency-injector/blob/cc2304e46e054ae08dc12995428759fbfb51af10/src/dependency_injector/wiring.py#L564

But if multiple decorators are used like this

@inject
@some_other_decorator
def my_func(...):

and some_other_decorator doesn't use functools.wraps, the Provide arguments are not found.

Here is a reproducible example:

import functools
from dependency_injector import containers, providers
from dependency_injector.wiring import Provide, inject


    
def some_other_decorator(func):
    @functools.wraps(func)
    def wrapped(*args, **kwargs):
        return func(*args, **kwargs)
    return wrapped



class Service:
    ...


class Container(containers.DeclarativeContainer):
    service = providers.Factory(Service)


@inject
@some_other_decorator
def main(service: Service = Provide[Container.service]) -> None:
    print(service)


if __name__ == "__main__":
    container = Container()
    # btw, this doesn't work: container.wire(modules=[__name__])
    # copied from https://github.com/ets-labs/python-dependency-injector/blob/master/examples/wiring/example.py
    container.wire(modules=[__name__])
    main()

If we comment out the @functools.wraps(func) line, we get printed:

<dependency_injector.wiring.Provide object at 0x10ae99fd0>

Issuing a warning inside _fetch_reference_injections if the decorated function doesn't have Provide arguments would help:

1. When such a decorator is used
2. When someone is using @inject and Provide was removed as unnecessary.

A small addiction to the work of @StummeJ about issue #633 which allows to benefit of his improvement even for dependencies passed as kwargs (such as my case)