Definition

Simple and composable validation and coercion of data structures. It also includes a ValueObject for convenience.

Installation

Add this line to your application's Gemfile:

gem 'definition'

And then execute:

$ bundle

Or install it yourself as:

$ gem install definition

Usage

Definitions can be used to validate data structures like for example Hashes:

schema = Definition.Keys do
  required :first_name, Definition.Type(String)
  required :last_name, Definition.Type(String)
  optional :birthday, Definition.Type(Date)
end

conform_result = schema.conform({first_name: "John", last_name: "Doe", birthday: Date.today})
conform_result.passed? # => true

conform_result = schema.conform({first_name: "John", last_name: "Doe", birthday: "2018/02/09"})
conform_result.passed? # => false
conform_result.error_message # => hash fails validation for key birthday: { Is of type String instead of Date }
conform_result.error_hash # =>
# {
#     :birthday => [
#         [0] <Definition::ConformError
#               description: "hash fails validation for key birthday: { Is of type String instead of Date }",
#               json_pointer: "/birthday">
#     ]
# }

But it can also transform those data structures at the same time. The following example shows how a Unix timestamp in milliseconds can be transformed to a Time object while validating:

milliseconds_time_definition = Definition.Lambda(:milliseconds_time) do |value|
  conform_with(Time.at(value.to_r / 1000).utc) if value.is_a?(Integer)
end

schema = Definition.Keys do
  required :title, Definition.Type(String)
  required :body, Definition.Type(String)
  optional :publication_date, milliseconds_time_definition
end

conform_result = schema.conform({title: "My first blog post", body: "Shortest one ever!", publication_date: 1546170180339})
conform_result.passed? # => true
conform_result.value # => {title: "My first blog post", body: "Shortest one ever!", publication_date: 2018-12-30 11:43:00 UTC}

Because definitions do not only validate input but also transform input, we use the term conform which stands for validation and coercion.

Handling errors

I18n translated errors

For end users you best use the translated errors that you get from definition:

schema = Definition.Keys do
  required :title, Definition.NonEmptyString
  required :body, Definition::And(
                    Definition.Type(String),
                    Definition.MinSize(100)
                  )
end

conform_result = schema.conform({title: "", body: "this is not long enough"})
conform_result.errors # => returns an array of Definition::ConformError
conform_result.errors.each do |error|
  puts "----"
  puts error.json_pointer # provides a path to the invalid value, also works with nested objects and arrays
  puts error.translated_error
end
# =>
# ----
# /title
# Value is shorter than 1
# ----
# /body
# Value is shorter than 100

The error messages are only translated into English for now, but you can add or change translations by adding a yaml file like this to your I18n load path.

Other ways of accessing errors

To get a quick error summary during debugging, you can also use conform_result.error_message

Instead of getting a flat array of all errors via conform_result.errors, you can also get a hierarchical representation:

conform_result.error_hash
# =>
# {
#     :title => [
#         [0] <Definition::ConformError 
# 	 message: "hash fails validation for key title: { Not all definitions are valid for 'non_empty_string': { Did not pass test for min_size (1) } }", 
# 	 json_pointer: "/title">
#     ],
#      :body => [
#         [0] <Definition::ConformError 
# 	 message: "hash fails validation for key body: { Not all definitions are valid for 'and': { Did not pass test for min_size (100) } }", 
# 	 json_pointer: "/body">
#     ]
# }

Conforming Hashes

Hashes can be conformed by using the Keys definition. It allows you to configure required and optional attributes. The first argument of required and optional takes either Symbols or Strings. If you use a Symbol, then the validated Hash needs to have a Symbol key with that name, otherwise a string key.

The key definition will also fail if the input value contains extra keys.

You can configure default values for optional keys, see the following example.

Definition.Keys do
  required :title, Definition.NonEmptyString
  optional :publication_date, Definition.Type(Date)
  optional :is_draft, Definition.Boolean, default: true
end

Ignoring unexpected keys

By default the Keys Definition does not conform with input hashes that contains keys that are not defined in the Definition. You can set the :ignore_extra_keys option to disable this.

schema = Definition.Keys do
  option :ignore_extra_keys

  required :title, Definition.NonEmptyString
  optional :publication_date, Definition.Type(Time)
end

conform_result = schema.conform({title: "My first blog post", body: "Shortest one ever!", publication_date: Time.new})
conform_result.passed? # => true
conform_result.value # => {title: "My first blog post", publication_date: 2018-12-30 11:43:00 UTC}

Validating types

This will validate that the value is of the specified type.

Definition.Type(String)
Definition.Type(Float)
Definition.Type(MyClass)

Definition.Type(MyClass).conform(0.1).passed? # => false
Definition.Type(MyClass).conform(MyClass.new).passed? # => true

Conforming types

This will validate that the value is of the specified type. But if its not it will try to coerce it into that type. This Definition works only with primitive types.

Definition.CoercibleType(String) # Uses String() to coerce values
Definition.CoercibleType(Float) # Uses Float() to coerce values

Definition.CoercibleType(Float).conform("0.1").passed? # => true
Definition.CoercibleType(Float).conform("0.1").value # => 0.1

Combining multiple definitions with "And"

Definition.And(definition1, definition2, ...)

This definition will only conform if all definitions conform. The definitions will be processed from left to right and the output of the previous will be the input of the next. Processing of the And-Definition stops as soon as one definition does not conform.

Combining multiple definitions with "Or"

Definition.Or(definition1, definition2, ...)

This definition will conform if at least one definition conforms. The definitions will be processed from left to right and stop as soon as a definition conforms. The output of that definition will be the output of the Or definition.

Conforming array values with "Each"

Definition.Each(item_definition)

Definition.Each(Definition.Type(Integer)).conform([1,2,3,"4"]).error_message
# => Not all items conform with each: { Item "4" did not conform to each: { Is of type String instead of Integer } }

This definition will only conform if all elements of the value conform to the item_definition.

Conforming with custom lambda functions

Definition.Lambda(:password) do |value|
  matches = Regexp.new(/^
    (?=.*[a-z]) # should contain at least one lower case letter
    (?=.*[A-Z]) # should contain at least one upper case letter
    (?=.*\d)    # should contain at least one digit
    .{6,50}     # should be between 6 and 50 characters long
    $/x).match(value.to_s)
  conform_with(value) if matches
end

This definition can be used to build any custom validation or coercion you want. The example above makes sure that a password conforms with a set of rules.

The block gets the input value as argument and you can do any transformation or validation on it that you want. If you determine that the value is valid, then you must call conform_with and pass it the value you want to return. This can either be the original value or any transformed version of it. By not calling conform_with you tell the definition to fail for the current input value.

The first argument of Definition.Lambda is a name you can give this definition. It will only be used in the error message to make it more readable.

If you want to provide detailed custom error messages you can use fail_with:

Definition.Lambda(:password) do |value|
  if !value.match(/[a-z]+/)
    fail_with("must contain at least one lower case letter")
  elsif !value.match(/[A-Z]+/)
    fail_with("must contain at least one upper case letter") 
  elsif !value.match(/\d+/)
    fail_with("must contain at least one digit") 
  elsif value.size < 6 || value.size > 50
    fail_with("must be between 6 and 50 characters long") 
  else
    conform_with(value)
  end
end

Composing Definitions

Definitions are reusable and can be easily composed:

country_code_definition = Definition.Lambda(:iso_county_code) do |value|
  if iso_code = IsoCountryCodes.find(value)
    conform_with(iso_code.alpha2)
  end
end

address_definition = Definition.Keys do
  required :street, Definition.Type(String)
  required :postal_code, Definition.Type(String)
  required :country_code, country_code_definition
end

order = Definition.Keys do
  required :user, user_definition
  required :invoice_address, address_definition
  required :shipping_address, address_definition
end

Extending Key definitions with include

Besides composing Definitions, you can also include Keys Definitions in each other. This will basically copy all required and optional keys as well as defaults into the other definition.

address_definition = Definition.Keys do
  required :street, Definition.Type(String)
  required :postal_code, Definition.Type(String)
  required :country_code, Definition.Type(String)
end

user_definition = Definition.Keys do
  required :user, user_definition

  include address_definition
end

Above Definition will equal the following:

user_definition = Definition.Keys do
  required :user, user_definition

  required :street, Definition.Type(String)
  required :postal_code, Definition.Type(String)
  required :country_code, Definition.Type(String)
end

Predefined Definitions

Strings and Arrays

Definition.MaxSize(5).conform("house") # => pass
Definition.MaxSize(5).conform([1,2,3,4,5]) # => pass

Definition.MinSize(5).conform("house") # => pass
Definition.MinSize(5).conform([1,2,3,4,5]) # => pass

Strings

Definition.NonEmptyString.conform("house") # => pass

Definition.Regex(/^\d*$/).conform("123") # => pass

Numerics

Definition.GreaterThan(5).conform(5.1) # => pass
Definition.GreaterThanEqual(5).conform(5) # => pass
Definition.LessThan(5).conform(4) # => pass
Definition.LessThanEqual(5).conform(5) # => pass

Strings, Array, Hashes

Definition.Empty.conform("") # => pass
Definition.Empty.conform([]) # => pass
Definition.Empty.conform({}) # => pass

Definition.NonEmpty.conform("Joe") # => pass
Definition.NonEmpty.conform([1]) # => pass
Definition.NonEmpty.conform({ a: 1 }) # => pass

Nil

Definition.Nil.conform(nil) # => pass

Boolean

Definition.Boolean.conform(true) # => pass

All types

Definition.Equal(5).conform(5) # => pass
Definition.Equal("foo").conform("foo") # => pass

The Nilable Definition allows a value to be nil or to conform with the definition you pass it as argument:

Definition.Nilable(Definition.Type(String)).conform(nil) # => pass
Definition.Nilable(Definition.Type(String)).conform("foo") # => pass

The Enum Definition checks if the input equals one of the values you pass it as argument. You can pass in as many arguments as you like:

Definition.Enum("foo", 1, 2.0).conform("foo") # => pass
Definition.Enum("foo", 1, 2.0).conform(1) # => pass
Definition.Enum("foo", 1, 2.0).conform("bar) # => fail

Examples

Check out the integration specs for more usage examples.

I18n translations

Every error object has a method translated_error that will give you a translated version of the error message. You can load the default English translations shipped with the gem by adding them to your I18n load path.

schema = Definition.Keys do
  required :title, Definition.Type(String)
  required :body, Definition.Type(String)
  required(:author, Definition.Keys do
    required :name, Definition.Type(String)
    required :email, Definition.Type(String)
  end)
end
schema.conform(input_hash).errors.first.translated_error # => Value is of wrong type, needs to be a String"

Helpers / useful tools

Value Objects / Models

Provides simple immutable objects that can validate and hold your data so that it can be safely passed around in your application.

class User < Definition::Model
  required :username, Definition.Type(String)
  required :password, Definition.Type(String)
  optional :age, Definition.Type(Integer)
end

user = User.new(username: "johndoe", password: "zg(2ds8x2/")
user.username # => "johndoe"
user.age # => nil
user.to_h # => { username: "johndoe", password: "zg(2ds8x2/" }
user.new(age: 21) # => new model instance with username and password from before plus the age set to 21

user.username = "Alice" # => raises NoMethodError (Models are immutable)

User.new(username: "johndoe") # => raises a Definition::InvalidModelError: hash is missing :password

You can access the conform result of InvalidModel errors via their conform_result method.

Nesting Models

Models can be nested by either using the model object itself as type definition, or by using the CoercibleModel Definition. The latter is less strict and will convert input hashes that conform with the model schema to an instance of the model.

class Address < Definition::Model
  required :street, Definition.Type(String)
  required :postal_code, Definition.Type(String)
end

class User < Definition::Model
  required :username, Definition.Type(String)
  required :address, Definition.CoercibleModel(Address)
end

# Address is converted into an Address model automatically:
user = User.new(username: "John", address: { street: "123 Fakestreet", postal_code: "2dfx4" })
user.address.street # => "123 Fakestreet"

class UserNotCoercibleAddress < Definition::Model
  required :username, Definition.Type(String)
  required :address, Address
end

# Address is not converted automatically, instead it needs to be of type Address already:
UserNotCoercibleAddress.new(username: "John", address: { street: "123 Fakestreet", postal_code: "2dfx4" }) # => raises a Definition::InvalidModelError
UserNotCoercibleAddress.new(username: "John", address: Address.new(street: "123 Fakestreet", postal_code: "2dfx4")).address.street # => "123 Fakestreet"

Intialization argument validation

Definition provides a mixin that allows you to validate keyword arguments of class initialization methods. This is meant to be used with classes that provide business logic whereas the models are meant to be used to pass data around.

The major differences to a Definition::Model are:

The values are not frozen and can be changed by the classes internal business logic
None of the getters for the attributes are public

class User
  include Definition::Initializer

  required :id, Definition.Type(Integer)
  required :name, Definition.Type(String)
  optional :phone, Definition.Type(String), default: nil

  def hello
    puts "Hello, I'm #{name}"
  end
end

user = User.new(id: 1, name: "Joe")
user.hello # => "Hello, I'm Joe"
user.name # => raises NoMethodError

User.new(id: "1", name: "Joe") # => raises a Definition::Initializer::InvalidArgumentError

Development

After checking out the repo, run bin/setup to install dependencies. Then, run rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/Goltergaul/definition.

definition

Development

Runtime

Definition

Installation

Usage

Handling errors

I18n translated errors

Other ways of accessing errors

Conforming Hashes

Ignoring unexpected keys

Validating types

Conforming types

Combining multiple definitions with "And"

Combining multiple definitions with "Or"

Conforming array values with "Each"

Conforming with custom lambda functions

Composing Definitions

Extending Key definitions with include

Predefined Definitions

Strings and Arrays

Strings

Numerics

Strings, Array, Hashes

Nil

Boolean

All types

Examples

I18n translations

Helpers / useful tools

Value Objects / Models

Nesting Models

Intialization argument validation

Development

Contributing