Parameter

Parameter helps you shape data from external sources into Elixir internal types. Use it to deal with any external data in general, such as API integrations, parsing user input, or validating data that comes into your system.

Parameter offers the following helpers:

Schema creation and validation
Input data validation
Deserialization
Serialization

Examples

defmodule UserParam do
  use Parameter.Schema
  alias Parameter.Validators

  param do
    field :first_name, :string, key: "firstName", required: true
    field :last_name, :string, key: "lastName"
    field :email, :string, validator: &Validators.email/1
    has_one :address, Address do
      field :city, :string, required: true
      field :street, :string
      field :number, :integer
    end
  end
end

Load (deserialize) the schema against external parameters:

params = %{
  "firstName" => "John",
  "lastName" => "Doe",
  "email" => "john@email.com",
  "address" => %{"city" => "New York", "street" => "York"}
}
Parameter.load(UserParam, params)
{:ok, %{
  first_name: "John",
  last_name: "Doe",
  email: "john@email.com",
  address: %{city: "New York", street: "York"}
}}

or Dump (serialize) a populated schema to params:

schema = %{
  first_name: "John",
  last_name: "Doe",
  email: "john@email.com",
  address: %{city: "New York", street: "York"}
}

Parameter.dump(UserParam, params)
{:ok,
 %{
    "firstName" => "John",
    "lastName" => "Doe",
    "email" => "john@email.com",
    "address" => %{"city" => "New York", "street" => "York"}
}}

Parameter offers a similar Schema model from Ecto library for creating a schema and parsing it against external data. The main use case of this library is to parse response from external APIs but you may also use to validate parameters in Phoenix Controllers, when receiving requests to validate it's parameters. In general Parameter can be used to build strucutred data and deal with serialization/deserialization of data. Check the official documentation for more information.

Installation

Add parameter to your list of dependencies in mix.exs:

def deps do
  [
    {:parameter, "~> 0.14"}
  ]
end

add :parameter on .formatter.exs:

import_deps: [:parameter]

Validating parameters on Controllers

Parameter let's you define the shape of the data that it's expected to receive in Phoenix Controllers:

defmodule MyProjectWeb.UserController do
  use MyProjectWeb, :controller
  import Parameter.Schema

  alias MyProject.Accounts

  param UserParams do
    field :first_name, :string, required: true
    field :last_name, :string, required: true
  end

  def create(conn, params) do
    with {:ok, user_params} <- Parameter.load(__MODULE__.UserParams, params),
        {:ok, user} <- Accounts.create_user(user_params) do
      render(conn, "user.json", %{user: user})
    end
  end
end

We can also use parameter for both request and response:

defmodule MyProjectWeb.UserController do
  use MyProjectWeb, :controller
  import Parameter.Schema

  alias MyProject.Accounts
  alias MyProject.Accounts.User

  param UserCreateRequest do
    field :first_name, :string, required: true
    field :last_name, :string, required: true
  end

  param UserCreateResponse do
    # Returning the user ID created on request
    field :id, :integer
    field :last_name, :string
    field :last_name, :string
  end

  def create(conn, params) do
    with {:ok, user_request} <- Parameter.load(__MODULE__.UserCreateRequest, params),
          {:ok, %User{} = user} <- Accounts.create_user(user_request),
          {:ok, user_response} <- Parameter.dump(__MODULE__.UserCreateResponse, user) do

        conn
        |> put_status(:created)
        |> json(%{user: user_response})
    end
  end
end

This example also shows that Parameter can dump the user response even if it comes from a different data strucutre. The %User{} struct on this example comes from Ecto.Schema and Parameter is able to convert it to params defined in UserCreateResponse.

Runtime schemas

It's also possible to create schemas via runtime without relying on any macros. This gives great flexibility on schema creation as now Parameter schemas can be created and validated dynamically:

schema = %{
  first_name: [key: "firstName", type: :string, required: true],
  address: [type: {:map, %{street: [type: :string, required: true]}}],
  phones: [type: {:array, %{country: [type: :string, required: true]}}]
} |> Parameter.Schema.compile!()

Parameter.load(schema, %{"firstName" => "John"})
{:ok, %{first_name: "John"}}

The same API can also be evaluated on compile time by using module attributes:

defmodule UserParams do
  alias Parameter.Schema

  @schema %{
    first_name: [key: "firstName", type: :string, required: true],
    address: [required: true, type: {:map, %{street: [type: :string, required: true]}}],
    phones: [type: {:array, %{country: [type: :string, required: true]}}]
  } |> Schema.compile!()

  def load(params) do
    Parameter.load(@schema, params)
  end
end

Or dynamically creating schemas:

defmodule EnvParser do
  alias Parameter.Schema

  def fetch!(env, opts \\ []) do
    atom_env = String.to_atom(env)
    type = Keyword.get(opts, :type, :string)
    default = Keyword.get(opts, :default)

    %{
      atom_env => [key: env, type: type, default: default, required: true]
    }
    |> Schema.compile!()
    |> Parameter.load(%{env => System.get_env(env)}, ignore_nil: true)
    |> case do
      {:ok, %{^atom_env => parsed_env}} -> parsed_env
      {:error, %{^atom_env => error}} -> raise ArgumentError, message: "#{env}: #{error}"
    end
  end
end

And now with this code we can dynamically fetch environment variables with System.get_env/1, define then as required, convert it to the correct type and use on our application's runtime:

# runtime.ex
import Config

# ...

config :my_app,
  auth_enabled?: EnvParser.fetch!("AUTH_ENABLED", default: true, type: :boolean),
  api_url: EnvParser.fetch!("API_URL") # using the default type string

# ...

this will come in handy since you don't have to worry anymore when fetching environment variables, what will be the shape of the data and what type I will have to use or convert in the application, Parameter will do this automatically for you.

This small example show one of the possibilities but this can be extended depending on your use case. A common example is to use runtime schemas when you have similar schemas and you want to reuse their properties across different entities:

user_base = %{first_name: [key: "firstName", type: :string, required: true]}
admin_params = %{role: [key: "role", type: :string, required: true]}
user_admin = Map.merge(user_base, admin_params)

user_base_schema = Parameter.Schema.compile!(user_base)
user_admin_schema = Parameter.Schema.compile!(user_admin)

# Now we can use both schemas to serialize/deserialize data with `load` and `dump` parameter functions

For more info on how to create schemas, check the schema documentation

License