Scoping resources

[davydovanton]

Scoping resources

[davydovanton]

Also, I get an idea to introduce data_source. I see it like this (it's just a draft to explain my idea):

module Post
  class BaseDataSource
    include Kan::DataSource

    role(:base) { |_user, _| true }

   register(:index) { |_user, repo| repo }
  end

  class AdminDataSource
    include Kan::DataSource

    role(:admin) { |user, _| user.admin? }

    register(:index) { |_user, repo| repo.exclude(draft: true) }
  end
end

app = Kan::Application.new(
  data_sources: { post: [Post::BaseDataSource.new, Post::AdminDataSource.new] }
)

KanImport = Dry::AutoInject(app)

class Service
  include KanImport[data_source: 'data_source.post.index']
  include Import['post_repo']

  def call(user, params)
    source = data_source.call(user, post_repo) #=> relation with condition

    # ...
  end
end

In this case, you can get a specific repository or data source based on your role and use it for getting or setting data. WDYT?

[ianks]

Also, I get an idea to introduce data_source. I see it like this (it's just a draft to explain my idea):

module Post
  class BaseDataSource
    include Kan::DataSource

    role(:base) { |_user, _| true }

   register(:index) { |_user, repo| repo }
  end

  class AdminDataSource
    include Kan::DataSource

    role(:admin) { |user, _| user.admin? }

    register(:index) { |_user, repo| repo.exclude(draft: true) }
  end
end

app = Kan::Application.new(
  data_sources: { post: [Post::BaseDataSource.new, Post::AdminDataSource.new] }
)

KanImport = Dry::AutoInject(app)

class Service
  include KanImport[data_source: 'data_source.post.index']
  include Import['post_repo']

  def call(user, params)
    source = data_source.call(user, post_repo) #=> relation with condition

    # ...
  end
end

In this case, you can get a specific repository or data source based on your role and use it for getting or setting data. WDYT?

I really dig this pattern.

[davydovanton]

hey! I don't have any register_scope methods right now and I'm not sure that this logic should be in the core (but I can imagine a scope plugin for kan).

If you have a simple example with the repository and just one call (user.admin?) I can suggest to use the simplest logic:

class PostRepository
  def all_for_user(user)
    if user.admin?
      self
    else
      exclude(draft: true)
    end
  end
end

But if you want to use resource idea with kan I think you can do it in this way:

module Post
  class BaseResources
    include Kan::Abilities

    role(:base) { |_user, _| true }

    register(:index) { |_user, repo| repo }
  end

  class AdminResources
    include Kan::Abilities

    role(:admin) { |user, _| user.admin? }

    register(:index) { |_user, repo| repo.exclude(draft: true) }
  end
end

resources = Kan::Application.new(
  post: [Post::BaseResources.new, Post::AdminResources.new]
)

ResourcesImport = Dry::AutoInject(resources)

class Service
  include ResourcesImport[resource: 'post.index']
  include Import['post_repo']

  def call(user, params)
    resource.call(user, post_repo) #=> relation with condition
    # ...
  end
end

[DangerDawson]

@davydovanton we are looking at using the above approach, although we are unsure of who this would be used with something like rom-rb / hanami-model. As we currently only return data back from our repository methods and do not leak the relation, e.g. we only return a boolean, single entity, or a collection of entities.

[DangerDawson]

After giving this some thought, I see 3 separate solutions if you are using the repository pattern:

• You return the repository method e.g.

register(:index) { |_user| :exclude_draft }

• you return a set of conditions in a data structure that the repository method can understand e.g.

register(:index) { |_user| { draft: false } }

Additionally wrapping this up in a Dry::Monad::Result could give you the ability to check success or failure, and on success retrieve the conditions / method e.g.

register(:index) { |user| ( Success( { draft: false } ) }