Skip to main content

Type based python dependency injection framework.

Project description

imbue

tests version python

Type based python dependency injection framework.

Why another dependency injection library?

There are many out there, including some that comes directly with (web) frameworks. I wanted one that was explicitly contextualized in the different layers of the application and that did not leak in the domain layer. In many cases, dependencies are simply parameters that are passed to constructors, and ideally, the domain layer should not know anything how they are injected.

Main objects

  • Container: a singleton containing all dependencies and the links between them
  • Dependency can be either:
    • an interface (class or function)
    • or a structured dependency linking an implementation to an interface
  • Context: contexts in which dependencies will live:
    • APPLICATION: equivalent to singletons across the app lifetime -- must be thread and async safe
    • THREAD: singleton per thread -- must be async safe
    • TASK: one instance will be provided for the entire task -- could be async safe
    • FACTORY: one instance will be provided every time it is requested
  • ContextualizedDependency: enriched dependency with context related attributes
  • ContextualizedContainer: one per context, handles their dependencies lifecycle
  • Provider: wraps injection for different interfaces (classes, functions, methods)
  • Package: provides for more dependency declaration and add grouped dependencies to the container

Usage

Summary

The container should be initialized somewhere near the entrypoint of the application.

from imbue import Container, Context, ContextualizedDependency

from myapp import APkg, ADep, BDep

container = Container(
    APkg(...),  # Packages would typically be passed config.
    ContextualizedDependency(ADep, Context.THREAD, eager=True),
)

...

# The application context should be entered through the framework initialization.
async with container.application_context() as app_container:
    a_dep = await app_container.get(ADep)
    ...
    # The task context should be entered in each request.
    async with app_container.task_context() as task_container:
        b_dep = await task_container.get(BDep)

💡 A dependency can require other dependencies, the container will resolve the graph and raise errors if circular dependencies are found.

In the details

The packages provide the dependencies for the container, there are two ways of doing that.

Complex dependencies

This might be because dependencies depend on config and other reasons. You can use the Package class:

from typing import AsyncIterator

from imbue import Package, application_context

from myapp import LoggingConfig, MyLoggerTransport, LoggerTransport, MyLogger, Logger

class LoggingPackage(Package):
    def __init__(self, config: LoggingConfig):
        self._config = config
        
    @application_context(eager=True)  # You can autoload dependencies.
    # The method should expose the interface as the type and return the implementation.
    def logger_transport(self) -> LoggerTransport:
        return MyLoggerTransport(self._config.transport)

    @application_context
    # Sub dependencies should be required through their interface.
    async def logger(self, transport: LoggerTransport) -> AsyncIterator[Logger]:
        # Package methods also allow you to handle context managers.
        async with MyLogger(transport, self._config.logger) as logger:
            # Using generators allow cleaning up resources when the context closes.
            yield logger
Declaring dependencies

It's done via methods that can be:

  • functions
  • async functions
  • generators
  • async generators

To declare the method returns a dependency, you simply choose the appropriate context and add the context decorator on the method.

💡 Eager dependencies are instantiated as soon as we enter their context. This is useful if we want to make sure a dependency will use the main thread's event loop.

Cleaning resources

When you need to close resources you can do so via a generator. The generator should yield the dependency.

💡 You should watch out for errors, if the context is closed handling an error, it will be raised in the generator.

Simple dependencies

You can directly pass them to the container:

Container(
    # Directly expose the implementation.
    ARepo,
    # Or interfaced implementations.
    Interfaced(BRepoInterface, BRepoImplementation),
    # You can also this to control context and eagerness.
    ContextualizedDependency(CDep, Context.THREAD, eager=True),
    ...,
)

or if you want to include them in a Package, add them in EXTRA_DEPENDENCIES.

💡 Whatever the method of adding dependencies, non-contextualized ones will have the context automatically set based on sub-dependencies. The lowest possible context will be used.

Integrations

Project details


Download files

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

Source Distribution

imbue-2.1.0.tar.gz (14.7 kB view details)

Uploaded Source

Built Distribution

imbue-2.1.0-py3-none-any.whl (19.9 kB view details)

Uploaded Python 3

File details

Details for the file imbue-2.1.0.tar.gz.

File metadata

  • Download URL: imbue-2.1.0.tar.gz
  • Upload date:
  • Size: 14.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.8.2 CPython/3.12.3 Linux/6.5.0-1018-azure

File hashes

Hashes for imbue-2.1.0.tar.gz
Algorithm Hash digest
SHA256 ad5ec900ca68a2e15b1ed4b5784ad3d66453f46c56a8c6cc2232311b862c78eb
MD5 625245cf8d9b0e633054efbd55ed4bfa
BLAKE2b-256 a0d294a7b2bc1731ac2635365c3fa0325a58c72aa65404cf131f19d76b0ec6b3

See more details on using hashes here.

Provenance

File details

Details for the file imbue-2.1.0-py3-none-any.whl.

File metadata

  • Download URL: imbue-2.1.0-py3-none-any.whl
  • Upload date:
  • Size: 19.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.8.2 CPython/3.12.3 Linux/6.5.0-1018-azure

File hashes

Hashes for imbue-2.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 9337057b4a1fdf24dc269886f2c15468f32a2d25d24803c0a0ac083b2b3e60b9
MD5 e45cc1d9c0b296cf4e74943403a8892b
BLAKE2b-256 9cbdd424a37944ece93f655d01ef16d7b72059022bfe25a9f4ca4ce04eb03d46

See more details on using hashes here.

Provenance

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page