The Nestru library can convert a map of any shape into a model of nested structs according to given hints. And It can convert any nested struct into a map.

The library’s primary purpose is to serialize between JSON map and an application’s model that is a struct having other structs nested in it.

It comes from the idea to have no DSL and less custom code to parse JSONs as an alternative to full-fledged Ecto with schemaless changesets.

Works good in Jason <-> Nestru pair.

There are several possible data validation options:

• with ex_json_schema applied to the map before passing to Nestru
• within Nestru gather_fields_map/3 and from_map_hint/3 callbacks
• with Ecto schemaless changesets applied to the struct
• with Domo applied to the struct to validate the struct conformance to its @type t() and associated preconditions

See typical usage in Readme.md via:

Shortly it looks like:

defmodule Order do
  @derive [
    Nestru.Encoder,
    {Nestru.Decoder, %{items: [LineItem], total: Total}}
  ]
  # Gave a hint to Nestru on how to process the items and total fields
  # others go to a struct as is.

  defstruct [:id, :items, :total]
end

defmodule LineItem do
  @derive [Nestru.Encoder, Nestru.Decoder]
  defstruct [:amount]
end

defmodule Total do
  @derive [Nestru.Encoder, Nestru.Decoder]
  defstruct [:sum]
end

map = %{
  "id" => "A548",
  "items" => [%{"amount" => 150}, %{"amount" => 350}],
  "total" => %{"sum" => 500}
}

{:ok, model} = Nestru.decode_from_map(map, Order)

returns:

{:ok,
  %OrderA{
    id: "A548",
    items: [%LineItemA{amount: 150}, %LineItemA{amount: 350}],
    total: %Total{sum: 500}
  }}

And going back to the map is as simple as that:

map = Nestru.encode_to_map(model)

returns:

%{
  id: "A548",
  items: [%{amount: 150}, %{amount: 350}],
  total: %{sum: 500}
}

More information here:

• GitHub
• Hex.pm
• Docs