So, I was planning on replacing my typed_struct usage with the one in Ash, but I got confused on how to approach in some cases.
For example, I have this module that uses typed_struct but it also “manipulates” the data before inserting into the struct:
defmodule Core.Services.OpenCorporates.Company do
@moduledoc false
alias Core.Services.OpenCorporates.Company.{Address, Agent}
use TypedStruct
typedstruct enforce: true do
field :name, String.t()
field :number, String.t()
field :type, String.t()
field :address, Address.t()
field :jurisdiction_code, String.t()
field :agent, Agent.t()
end
def new!(company) when is_list(company), do: company |> Map.new() |> new!()
def new!(company) when is_map(company) do
struct!(__MODULE__,
name: maybe_downcase(company[:name]),
number: maybe_downcase(company[:number]),
type: maybe_downcase(company[:type]),
address: Address.new!(company[:address]),
jurisdiction_code: maybe_downcase(company[:jurisdiction_code]),
agent: Agent.new!(company[:agent])
)
end
defp maybe_downcase(nil), do: nil
defp maybe_downcase(value), do: String.downcase(value)
end
Now, with Ash’s typed_struct, the new! functions would be already implemented for me, but they would not have the maybe_downcase call I’m running here.
I can rename these functions to be something like new_normalized! and call __MODULE__.new! inside of it, but I wonder if there is a cleaner way to do it.
That made me wonder if it would be possible to actually support changes inside these structs, that way I would just create a change that would normalize the struct and keep everything inside that scope.
