StrongJSON

This library allows you to test the structure of JSON objects.

This is similar to Strong Parameters, which is introduced by Rails 4, but expected to work with more complex structures. It may help you to understand what this is as: Strong Parameters is for simple structures, like HTML forms, and StrongJSON is for complex structures, like JSON objects posted to API.

Installation

Add this line to your application's Gemfile:

gem 'strong_json'

And then execute:

$ bundle

Or install it yourself as:

$ gem install strong_json

Usage

s = StrongJSON.new do
  let :item, object(id: prohibited, name: string, count: numeric)
  let :customer, object(name: string, address: string, phone: string, email: optional(string))
  let :order, object(customer: customer, items: array(item))
end

json = s.order.coerce(JSON.parse(input, symbolize_names: true))
s.order =~ JSON.parse(input, symbolize_names: true)

case JSON.parse(input2, symbolize_names: true)
when s.item
  # input2 is an item
when s.customer
  # input2 is a customer
else
  # input2 is something else
end

If the input JSON data conforms to order's structure, the json will be that value.

If the input JSON contains attributes which is not white-listed in the definition, it will raise an exception.

If an attribute has a value which does not match with given type, the coerce method call will raise an exception StrongJSON::Type::Error.

Catalogue of Types

object(f1: type1, f2: type2, ...)

  • The value must be an object
  • Fields, f1, f2, and ..., must be present and its values must be of type1, type2, ..., respectively
  • Objects with other fields will be rejected

Performance hint

Object attributes test is done in order of the keys.

slower_object = enum(
  object(id: numeric, created_at: string, updated_at: string, type: literal("person"), name: string),
  object(id: numeric, created_at: string, updated_at: string, type: literal("food"), object: any)
)

faster_object = enum(
  object(type: literal("person"), id: numeric, created_at: string, updated_at: string, name: string),
  object(type: literal("food"), id: numeric, created_at: string, updated_at: string, object: any)
)

The two enums represents same object, but testing runs faster with faster_object. Objects in faster_object have type attribute as their first keys. Testing type is done first, and it soon determines if the object is "person" or "food".

array(type)

  • The value must be an array
  • All elements in the array must be value of given type

optional(type)

  • The value can be nil (or not contained in an object)
  • If an value exists, it must be of given type

enum(type1, type2, ...)

  • The value can be one of the given types
  • First successfully coerced value will return

Base types

  • number The value must be an instance of Numeric
  • string The value must be an instance of String
  • boolean The value must be true or false
  • numeric The value must be an instance of Numeric or a string which represents a number
  • any Any value except nil is accepted
  • ignored Any value will be ignored
  • symbol The value must be an instance of String or Symbol; returns the result ot #to_sym

Literals

  • literal(lit) The value must == lit

Shortcuts

There are some alias for optional(base), where base is base types, as the following:

  • number?
  • string?
  • boolean?
  • numeric?
  • symbol?
  • literal?(lit)
  • any?

Shortcuts for complex data are also defined as the following:

  • optional(array(ty))array?(ty)
  • optional(object(fields))object?(fields)
  • optional(enum(types))enum?(types)

Type checking

StrongJSON ships with type definitions for Steep. You can type check your programs using StrongJSON by Steep.

Type definition

Define your types as the following.

class JSONSchema::Account < StrongJSON
  def account: -> StrongJSON::_Schema<{ id: Integer, name: String }>
end

Schema: JSONSchema::Account

And write your schema definition as the following.

Schema = _ = StrongJSON.new do
  # @type self: JSONSchema::Account

  let :account, object(id: number, name: string)
end

id = Schema.account.coerce(hash)[:id]       # id is Integer
name = Schema.account.coerce(hash)[:name]   # name is String

Note that you need two tricks:

  • A cast _ = StrongJSON.new ... on assignment to Schema constant
  • A @type self annotation in the block

See the example directory.

Commandline

Steep 0.8.1 supports loading type definitions from gems.

Pass -G option to type check your program.

$ steep check -G strong_json lib

When you are using bundler, it automatically detects that StrongJSON has type definitions.

$ bundle exec steep check lib

Contributing

  1. Fork it ( https://github.com/soutaro/strong_json/fork )
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create a new Pull Request