Type specs

Warning

Starting from dry-schema 2.0 type specs will be obligatory arguments. ie filled(:string) will work but filled without a type spec will raise an argument error.

To define what the expected type of a value is, you should use type specs. All macros support type specs as the first argument, whenever you pass a symbol that doesn't end with a question mark, or you explicitly pass in an instance of a Dry::Types::Type object, it will be set as the type.

Whenever you define a type spec, dry-schema will infer a type-check predicate. ie:

  • :string => str?
  • :integer => :int?
  • :array => :array?
  • etc.

These predicates will be prepended to the list of the predicates you specified (if any).

Using type identifiers

In most common cases you can use symbols that identify built-in types. The types are resolved from type registry which is configured for individual schemas. For example Dry::Schema::Params has its type registry configured to use Params types by default. This means that if you specify :integer as the type, then Dry::Schema::Types::Params::Integer will be used as the resolved type.

UserSchema = Dry::Schema.Params do
  # expands to: `int? & gt?(18)`
  required(:age).value(:integer, gt?: 18)
end

Using arrays with member types

To define an array with a member, you can use a shortcut method array. Here's an example of an array with :integer set as its member type:

UserSchema = Dry::Schema.Params do
  # expands to: `array? & each { int? } & size?(3)`
  required(:nums).value(array[:integer], size?: 3)
end

Using custom types

You are not limited to the built-in types. The DSL accepts any Dry::Types::Type object:

module Types
  include Dry::Types()

  StrippedString = Types::String.constructor(&:strip)
end

UserSchema = Dry::Schema.Params do
  # expands to: `str? & min_size?(10)`
  required(:login_time).value(StrippedString, min_size?: 10)
end

Learn more

octocatEdit on GitHub