NodNod

🧑‍🧑‍🧒‍🧒

about

nodnod is a dependency injection tool

nodnod strives to be effective and simple: it has many agents, but, for example, asyncio agent turns each dependency into a coroutine and yields decomposed dependency net to your loop to solve it the best way possible

nodnod is easy to integrate, be it a framework or a small project

interfaces

composing a set of dependencies is as simple as:

async def main():
    agent = EventLoopAgent.build({Username})

    async with Scope() as scope:
        scope.inject(User, User(id=1, username=Some("arseni")))
        await agent.run(scope, {})

    print(scope[Username])

two objects that define compile-time and runtime are agent and Scope.

• agent precompiles the dependency graph
• scope contains dependency roots and defines dependency visibility within one scope. many scopes can be linked in the manner of parent-child or global-local relation. dependencies are made accessible within different scopes

nodnod provides many interfaces to build different kinds of dependencies:

• polymorphic - node whose resolution may vary

  @polymorphic[float]
  class Ratio:
      @case
      def from_video(cls, video: Video) -> float:
          frame = video.frames[0]
          return frame.width / frame.length
  
      @case
      def from_photo(cls, photo: Photo) -> float:
          return photo.width / photo.length

• scalar node disguises as its result for type checkers

  @scalar_node
  class Username:
      @classmethod
      def __compose__(cls, user: User) -> str:
          return user.username.map_error(lambda _: f"#{user.id}")

• logical or

  @classmethod
  def __compose__(cls, dep: One | Another) -> ...: ...

  • can be concurrent: two nodes are racing to deliver their resolution
  • or sequential: if first resolution fails, next one is triggered

• kungfu funtional types are supported:

  Exceptions can be intercepted:

  @classmethod
  def __compose__(cls, weather: Result[Weather, ConnectionError)): ...

  Dependencies can be optional:

  @classmethod
  def __compose__(cls, ratio: Option[Ratio]): ...

your contributions are welcome!

🌺