grape swagger

Add OAPI/swagger v2.0 compliant documentation to your grape API

ruby-grape
1090
470
Ruby

Gem Version
Build Status
Coverage Status
Code Climate

Table of Contents

What is grape-swagger?

The grape-swagger gem provides an autogenerated documentation for your Grape API. The generated documentation is Swagger-compliant, meaning it can easily be discovered in Swagger UI. You should be able to point the petstore demo to your API.

Demo Screenshot

This screenshot is based on the Hussars sample app.

Related Projects

Compatibility

The following versions of grape, grape-entity and grape-swagger can currently be used together.

grape-swagger swagger spec grape grape-entity representable
0.10.5 1.2 >= 0.10.0 … <= 0.14.0 < 0.5.0 n/a
0.11.0 1.2 >= 0.16.2 < 0.5.0 n/a
0.25.2 2.0 >= 0.14.0 … <= 0.18.0 <= 0.6.0 >= 2.4.1
0.26.0 2.0 >= 0.16.2 … <= 1.1.0 <= 0.6.1 >= 2.4.1
0.27.0 2.0 >= 0.16.2 … <= 1.1.0 >= 0.5.0 >= 2.4.1
0.32.0 2.0 >= 0.16.2 >= 0.5.0 >= 2.4.1
0.34.0 2.0 >= 0.16.2 … < 1.3.0 >= 0.5.0 >= 2.4.1
>= 1.0.0 2.0 >= 1.3.0 >= 0.5.0 >= 2.4.1
>= 2.0.0 2.0 >= 1.7.0 >= 0.5.0 >= 2.4.1

Swagger-Spec

Grape-swagger generates documentation per Swagger / OpenAPI Spec 2.0.

Installation

Add to your Gemfile:

gem 'grape-swagger'

Upgrade

Please see UPGRADING when upgrading from a previous version.

Usage

Mount all your different APIs (with Grape::API superclass) on a root node. In the root class definition, include add_swagger_documentation, this sets up the system and registers the documentation on ‘/swagger_doc’. See example/config.ru for a simple demo.

require 'grape-swagger'

module API
  class Root < Grape::API
    format :json
    mount API::Cats
    mount API::Dogs
    mount API::Pirates
    add_swagger_documentation
  end
end

To explore your API, either download Swagger UI and set it up yourself or go to the online swagger demo and enter your localhost url documentation root in the url field (probably something in the line of http://localhost:3000/swagger_doc).

Model Parsers

Since 0.21.0, Grape::Entity is not a part of grape-swagger, you need to add grape-swagger-entity manually to your Gemfile.
Also added support for representable via grape-swagger-representable.

# For Grape::Entity ( https://github.com/ruby-grape/grape-entity )
gem 'grape-swagger-entity', '~> 0.3'
# For representable ( https://github.com/apotonick/representable )
gem 'grape-swagger-representable', '~> 0.2'

If you are not using Rails, make sure to load the parser inside your application initialization logic, e.g., via require 'grape-swagger/entity' or require 'grape-swagger/representable'.

Custom Model Parsers

You can create your own model parser, for example for roar.

module GrapeSwagger
  module Roar
    class Parser
      attr_reader :model
      attr_reader :endpoint

      def initialize(model, endpoint)
        @model = model
        @endpoint = endpoint
      end

      def call
        # Parse your model and return hash with model schema for swagger
      end
    end
  end
end

Then you should register your custom parser.

GrapeSwagger.model_parsers.register(GrapeSwagger::Roar::Parser, Roar::Decorator)

To control model parsers sequence, you can insert your parser before or after another parser.

insert_before

GrapeSwagger.model_parsers.insert_before(GrapeSwagger::Representable::Parser, GrapeSwagger::Roar::Parser, Roar::Decorator)

insert_after

GrapeSwagger.model_parsers.insert_after(GrapeSwagger::Roar::Parser, GrapeSwagger::Representable::Parser, Representable::Decorator)

As we know, Roar::Decorator uses Representable::Decorator as a superclass, this allows to avoid a problem when Roar objects are processed by GrapeSwagger::Representable::Parser instead of GrapeSwagger::Roar::Parser.

CORS

If you use the online demo, make sure your API supports foreign requests by enabling CORS in Grape, otherwise you’ll see the API description, but requests on the API won’t return. Use rack-cors to enable CORS.

require 'rack/cors'
use Rack::Cors do
  allow do
    origins '*'
    resource '*', headers: :any, methods: [ :get, :post, :put, :delete, :options ]
  end
end

Alternatively you can set CORS headers in a Grape before block.

before do
  header['Access-Control-Allow-Origin'] = '*'
  header['Access-Control-Request-Method'] = '*'
end

Configure

You can pass a hash with optional configuration settings to add_swagger_documentation.
The examples show the default value.

The host and base_path options also accept a proc or a lambda to evaluate, which is passed a request object:

add_swagger_documentation \
  base_path: proc { |request| request.host =~ /^example/ ? '/api-example' : '/api' }

host:

Sets explicit the host, default would be taken from request.

add_swagger_documentation \
   host: 'www.example.com'

base_path:

Base path of the API that’s being exposed, default would be taken from request.

add_swagger_documentation \
   base_path: nil

host and base_path are also accepting a proc or lambda

mount_path:

The path where the API documentation is loaded, default is: /swagger_doc.

add_swagger_documentation \
   mount_path: '/swagger_doc'

add_base_path:

Add basePath key to the documented path keys, default is: false.

add_swagger_documentation \
   add_base_path: true # only if base_path given

add_root:

Add root element to all the responses, default is: false.

add_swagger_documentation \
   add_root: true

add_version:

Add version key to the documented path keys, default is: true,
here the version is the API version, specified by grape in path

add_swagger_documentation \
   add_version: true

doc_version:

Specify the version of the documentation at info section, default is: '0.0.1'

add_swagger_documentation \
   doc_version: '0.0.1'

endpoint_auth_wrapper:

Specify the middleware to use for securing endpoints.

add_swagger_documentation \
   endpoint_auth_wrapper: WineBouncer::OAuth2

swagger_endpoint_guard:

Specify the method and auth scopes, used by the middleware for securing endpoints.

add_swagger_documentation \
   swagger_endpoint_guard: 'oauth2 false'

token_owner:

Specify the token_owner method, provided by the middleware, which is typically named ‘resource_owner’.

add_swagger_documentation \
   token_owner: 'resource_owner'

security_definitions:

Specify the Security Definitions Object

NOTE: Swagger-UI is supporting only implicit flow yet

add_swagger_documentation \
  security_definitions: {
    api_key: {
      type: "apiKey",
      name: "api_key",
      in: "header"
    }
  }

security:

Specify the Security Object

add_swagger_documentation \
  security: [
    {
      api_key: []
    }
  ]

models:

A list of entities to document. Combine with the grape-entity gem.

These would be added to the definitions section of the swagger file.

add_swagger_documentation \
   models: [
     TheApi::Entities::UseResponse,
     TheApi::Entities::ApiError
   ]

tags:

A list of tags to document. By default tags are automatically generated
for endpoints based on route names.

add_swagger_documentation \
  tags: [
    { name: 'widgets', description: 'A description of widgets' }
  ]

hide_documentation_path: (default: true) 

add_swagger_documentation \
   hide_documentation_path: true

Don’t show the /swagger_doc path in the generated swagger documentation.

info: 

add_swagger_documentation \
  info: {
    title: "The API title to be displayed on the API homepage.",
    description: "A description of the API.",
    contact_name: "Contact name",
    contact_email: "[email protected]",
    contact_url: "Contact URL",
    license: "The name of the license.",
    license_url: "www.The-URL-of-the-license.org",
    terms_of_service_url: "www.The-URL-of-the-terms-and-service.com",
  }

A hash merged into the info key of the JSON documentation.

array_use_braces: 

add_swagger_documentation \
  array_use_braces: true

This setting must be true in order for params defined as an Array type to submit each element properly.

params do
 optional :metadata, type: Array[String]
end

with array_use_braces: true:

metadata[]: { "name": "Asset ID", "value": "12345" }
metadata[]: { "name": "Asset Tag", "value": "654321"}

with array_use_braces: false:

metadata: {"name": "Asset ID", "value": "123456"}
metadata: {"name": "Asset Tag", "value": "654321"}

api_documentation

Customize the Swagger API documentation route, typically contains a desc field. The default description is “Swagger compatible API description”.

add_swagger_documentation \
   api_documentation: { desc: 'Reticulated splines API swagger-compatible documentation.' }

specific_api_documentation

Customize the Swagger API specific documentation route, typically contains a desc field. The default description is “Swagger compatible API description for specific API”.

add_swagger_documentation \
   specific_api_documentation: { desc: 'Reticulated splines API swagger-compatible endpoint documentation.' }

consumes

Customize the Swagger API default global consumes field value.

add_swagger_documentation \
   consumes: ['application/json', 'application/x-www-form-urlencoded']

produces

Customize the Swagger API default global produces field value.

add_swagger_documentation \
   produces: ['text/plain']

Routes Configuration

Swagger Header Parameters

Swagger also supports the documentation of parameters passed in the header. Since grape’s params[] doesn’t return header parameters we can specify header parameters separately in a block after the description.

desc "Return super-secret information", {
  headers: {
    "XAuthToken" => {
      description: "Valdates your identity",
      required: true
    },
    "XOptionalHeader" => {
      description: "Not really needed",
      required: false
    }
  }
}

Hiding an Endpoint

You can hide an endpoint by adding hidden: true in the description of the endpoint:

desc 'Hide this endpoint', hidden: true

Or by adding hidden: true on the verb method of the endpoint, such as get, post and put:

get '/kittens', hidden: true do

Or by using a route setting:

route_setting :swagger, { hidden: true }
get '/kittens' do

Endpoints can be conditionally hidden by providing a callable object such as a lambda which evaluates to the desired
state:

desc 'Conditionally hide this endpoint', hidden: lambda { ENV['EXPERIMENTAL'] != 'true' }

Overriding Auto-Generated Nicknames

You can specify a swagger nickname to use instead of the auto generated name by adding :nickname 'string' in the description of the endpoint.

desc 'Get a full list of pets', nickname: 'getAllPets'

Specify endpoint details

To specify further details for an endpoint, use the detail option within a block passed to desc:

desc 'Get all kittens!' do
  detail 'this will expose all the kittens'
end
get '/kittens' do

Overriding the route summary

To override the summary, add summary: '[string]' after the description.

namespace 'order' do
  desc 'This will be your summary',
    summary: 'Now this is your summary!'
  get :order_id do
    ...
  end
end

Overriding the tags

Tags are used for logical grouping of operations by resources or any other qualifier. To override the
tags array, add tags: ['tag1', 'tag2'] after the description.

namespace 'order' do
  desc 'This will be your summary', tags: ['orders']
  get :order_id do
    ...
  end
end

Deprecating routes

To deprecate a route add deprecated: true after the description.

namespace 'order' do
  desc 'This is a deprecated route', deprecated: true
  get :order_id do
    ...
  end
end

Overriding the name of the body parameter

By default, body parameters have a generated name based on the operation. For
deeply nested resources, this name can get very long. To override the name of
body parameter add body_name: 'post_body' after the description.

namespace 'order' do
  desc 'Create an order', body_name: 'post_body'
  post do
    ...
  end
end

Defining an endpoint as an array

You can define an endpoint as an array by adding is_array in the description:

desc 'Get a full list of pets', is_array: true

Using an options hash

The Grape DSL supports either an options hash or a restricted block to pass settings. Passing the nickname, hidden and is_array options together with response codes is only possible when passing an options hash.
Since the syntax differs you’ll need to adjust it accordingly:

desc 'Get all kittens!', {
  hidden: true,
  is_array: true,
  nickname: 'getKittens',
  success: Entities::Kitten, # or success
  failure: [[401, 'KittenBitesError', Entities::BadKitten]] # or failure
  # also explicit as hash: [{ code: 401, message: 'KittenBitesError', model: Entities::BadKitten }]
  produces: [ "array", "of", "mime_types" ],
  consumes: [ "array", "of", "mime_types" ]
  }
get '/kittens' do

Overriding parameter type

You can override paramType, using the documentation hash. See parameter object for available types.

params do
  requires :action, type: Symbol, values: [:PAUSE, :RESUME, :STOP], documentation: { param_type: 'query' }
end
post :act do
  ...
end

Overriding data type of the parameter

You can override type, using the documentation hash.

params do
  requires :input, type: String, documentation: { type: 'integer' }
end
post :act do
  ...
end

{
  "in": "formData",
  "name": "input",
  "type": "integer",
  "format": "int32",
  "required": true
}

Multiple types

By default when you set multiple types, the first type is selected as swagger type

params do
  requires :action, types: [String, Integer]
end
post :act do
  ...
end

{
  "in": "formData",
  "name": "action",
  "type": "string",
  "required": true
}

Array of data type

Array types are also supported.

params do
  requires :action_ids, type: Array[Integer]
end
post :act do
  ...
end

{
  "in": "formData",
  "name": "action_ids",
  "type": "array",
  "items": {
      "type": "integer"
  },
  "required": true
}

Collection format of arrays

You can set the collection format of an array, using the documentation hash.

Collection format determines the format of the array if type array is used. Possible values are:

  • csv - comma separated values foo,bar.
  • ssv - space separated values foo bar.
  • tsv - tab separated values foo\tbar.
  • pipes - pipe separated values foo|bar.
  • multi - corresponds to multiple parameter instances instead of multiple values for a single instance foo=bar&foo=baz. This is valid only for parameters in “query” or “formData”.
params do
  requires :statuses, type: Array[String], documentation: { collectionFormat: 'multi' }
end
post :act do
  ...
end

{
  "in": "formData",
  "name": "statuses",
  "type": "array",
  "items": {
      "type": "string"
  },
  "collectionFormat": "multi",
  "required": true
}

Hiding parameters

Exclude single optional parameter from the documentation

not_admins = lambda { |token_owner = nil| token_owner.nil? || !token_owner.admin? }

params do
  optional :one, documentation: { hidden: true }
  optional :two, documentation: { hidden: -> { |t=nil| true } }
  optional :three, documentation: { hidden: not_admins }
end
post :act do
  ...
end

Setting a Swagger default value

Grape allows for an additional documentation hash to be passed to a parameter.

params do
  requires :id, type: Integer, desc: 'Coffee ID'
  requires :temperature, type: Integer, desc: 'Temperature of the coffee in celcius', documentation: { default: 72 }
end

Grape uses the option default to set a default value for optional parameters. This is different in that Grape will set your parameter to the provided default if the parameter is omitted, whereas the example value above will only set the value in the UI itself. This will set the Swagger defaultValue to the provided value. Note that the example value will override the Grape default value.

params do
  requires :id, type: Integer, desc: 'Coffee ID'
  optional :temperature, type: Integer, desc: 'Temperature of the coffee in celcius', default: 72
end

Setting additionalProperties for object-type parameters

Use the additional_properties option in the documentation hash for object-type parameters to set additionalProperties.

Allow any additional properties

params do
  optional :thing, type: Hash, documentation: { additional_properties: true }
end

Allow any additional properties of a particular type

params do
  optional :thing, type: Hash, documentation: { additional_properties: String }
end

Allow any additional properties matching a defined schema

class Entity < Grape::Entity
  expose :this
end

params do
  optional :thing, type: Hash, documentation: { additional_properties: Entity }
end

Example parameter value

The example parameter will populate the Swagger UI with the example value, and can be used for optional or required parameters.

params do
  requires :id, type: Integer, documentation: { example: 123 }
  optional :name, type: String, documentation: { example: 'Buddy Guy' }
end

Expose nested namespace as standalone route

Use the nested: false property in the swagger option to make nested namespaces appear as standalone resources.
This option can help to structure and keep the swagger schema simple.

namespace 'store/order', desc: 'Order operations within a store', swagger: { nested: false } do
  get :order_id do
  	...
  end
end

All routes that belong to this namespace (here: the GET /order_id) will then be assigned to the store_order route instead of the store resource route.

It is also possible to expose a namespace within another already exposed namespace:

namespace 'store/order', desc: 'Order operations within a store', swagger: { nested: false } do
  get :order_id do
  	...
  end
  namespace 'actions', desc: 'Order actions', nested: false do
    get 'evaluate' do
      ...
    end
  end
end

Here, the GET /order_id appears as operation of the store_order resource and the GET /evaluate as operation of the store_orders_actions route.

With a custom name

Auto generated names for the standalone version of complex nested resource do not have a nice look.
You can set a custom name with the name property inside the swagger option, but only if the namespace gets exposed as standalone route.
The name should not contain whitespaces or any other special characters due to further issues within swagger-ui.

namespace 'store/order', desc: 'Order operations within a store', swagger: { nested: false, name: 'Store-orders' } do
  get :order_id do
  	...
  end
end

Response documentation

You can also document the HTTP status codes with a description and a specified model, as ref in the schema to the definitions, that your API returns with one of the following syntax.

In the following cases, the schema ref would be taken from route.

desc 'thing', failure: [ { code: 400, message: 'Invalid parameter entry' } ]
get '/thing' do
  # ...
end

desc 'thing' do
  params Entities::Something.documentation
  failure [ { code: 400, message: 'Invalid parameter entry' } ]
end
get '/thing' do
  # ...
end

get '/thing', failure: [
  { code: 400, message: 'Invalid parameter entry' },
  { code: 404, message: 'Not authorized' },
] do
  # ...
end

By adding a model key, e.g. this would be taken. Setting an empty string will act like an empty body.

get '/thing', failure: [
  { code: 400, message: 'General error' },
  { code: 403, message: 'Forbidden error', model: '' },
  { code: 422, message: 'Invalid parameter entry', model: Entities::ApiError }
] do
  # ...
end

If no status code is defined defaults would be taken.

The result is then something like following:

"responses": {
  "200": {
    "description": "get Horses",
    "schema": {
      "$ref": "#/definitions/Thing"
    }
  },
  "401": {
    "description": "HorsesOutError",
    "schema": {
      "$ref": "#/definitions/ApiError"
    }
  }
},

Changing default status codes

The default status codes, one could be found (-> status codes) can be changed to your specific needs, to achieve it, you have to change it for grape itself and for the documentation.

desc 'Get a list of stuff',
    success: { code: 202, model: Entities::UseResponse, message: 'a changed status code' }
get do
  status 202
  # your code comes here
end

"responses": {
  "202": {
    "description": "ok",
    "schema": {
      "$ref": "#/definitions/UseResponse"
    }
  }
},

Multiple status codes for response

Multiple values can be provided for success and failure attributes in the response.

desc 'Attach a field to an entity through a PUT',
    success: [
      { code: 201, model: Entities::UseResponse, message: 'Successfully created' },
      { code: 204, message: 'Already exists' }
    ],
    failure: [
      { code: 400, message: 'Bad request' },
      { code: 404, message: 'Not found' }
    ]
put do
  # your code comes here
end

"responses": {
  "201": {
    "description": "Successfully created",
    "schema": {
      "$ref": "#/definitions/UseResponse"
    }
  },
  "204": {
    "description": "Already exists"
  },
  "400": {
    "description": "Bad request"
  },
  "404": {
    "description": "Not found"
  }
},

File response

Setting success to File sets a default produces of application/octet-stream.

desc 'Get a file',
    success: File
get do
  # your file response
end

"produces": [
  "application/octet-stream"
],
"responses": {
  "200": {
    "description": "Get a file",
    "schema": {
      "type": "file"
    }
  }
}

Default response

By setting the default option you can also define a default response that is the result returned for all unspecified status codes.
The definition supports the same syntax as success or failure.

In the following cases, the schema ref would be taken from route.

desc 'thing', default: { message: 'the default response' }
get '/thing' do
  # ...
end

The generated swagger section will be something like

"responses": {
  "default": {
    "description": "the default response"
  }
},

Just like with success or failure you can also provide a model parameter.

desc 'Get a list of stuff',
    default: { model: Entities::UseResponse, message: 'the default response' }
get do
  # your code comes here
end

The generated swagger document will then correctly reference the model schema.

"responses": {
  "default": {
    "description": "the default response",
    "schema": {
      "$ref": "#/definitions/UseResponse"
    }
  }
},

Extensions

Swagger spec2.0 supports extensions on different levels, for the moment,
the documentation on the root level object and the info, verb, path and definition levels are supported.

The documented key would be generated from the x + - + key of the submitted hash,
for possibilities refer to the extensions spec.
To get an overview how the extensions would be defined on grape level, see the following examples:

  • root object extension, add a x key to the root hash when calling add_swagger_documentation:
  add_swagger_documentation \
    x: {
      some: 'stuff'
    },
    info: {
    }

this would generate:

{
  "x-some": "stuff",
  "info":{
  }
}
  • info extension, add a x key to the info hash when calling add_swagger_documentation:
  add_swagger_documentation \
    info: {
      x: { some: 'stuff' }
    }

this would generate:

"info":{
  "x-some":"stuff"
}
  • verb extension, add a x key to the desc hash:
desc 'This returns something with extension on verb level',
  x: { some: 'stuff' }

this would generate:

"/path":{
  "get":{
    "…":"…",
    "x-some":"stuff"
  }
}
  • operation extension, by setting via route settings::
route_setting :x_operation, { some: 'stuff' }

this would generate:

"/path":{
  "get":{
    "…":"…",
    "x-some":"stuff"
  }
}
  • path extension, by setting via route settings:
route_setting :x_path, { some: 'stuff' }

this would generate:

"/path":{
  "x-some":"stuff",
  "get":{
    "…":"…",
  }
}
  • definition extension, again by setting via route settings,
    here the status code must be provided, for which definition the extensions should be:
route_setting :x_def, { for: 422, other: 'stuff' }

this would generate:

"/definitions":{
  "ApiError":{
    "x-other":"stuff",
    "…":"…",
  }
}

or, for more definitions:

route_setting :x_def, [{ for: 422, other: 'stuff' }, { for: 200, some: 'stuff' }]
  • params extension, add a x key to the documentation hash :
requires :foo, type: String, documentation: { x: { some: 'stuff' } }

this would generate:

{
  "in": "formData",
  "name": "foo",
  "type": "string",
  "required": true,
  "x-some": "stuff"
}

Response examples documentation

You can also add examples to your responses by using the desc DSL with block syntax.

By specifying examples to success and failure.

desc 'This returns examples' do
  success model: Thing, examples: { 'application/json' => { description: 'Names list', items: [{ id: '123', name: 'John' }] } }
  failure [[404, 'NotFound', ApiError, { 'application/json' => { code: 404, message: 'Not found' } }]]
end
get '/thing' do
  ...
end

The result will look like following:

  "responses": {
    "200": {
      "description": "This returns examples",
      "schema": {
        "$ref": "#/definitions/Thing"
      },
      "examples": {
        "application/json": {
          "description": "Names list",
          "items": [
            {
              "id": "123",
              "name": "John"
            }
          ]
        }
      }
    },
    "404": {
      "description": "NotFound",
      "schema": {
        "$ref": "#/definitions/ApiError"
      },
      "examples": {
        "application/json": {
          "code": 404,
          "message": "Not found"
        }
      }
    }
  }

Failure information can be passed as an array of arrays or an array of hashes.

Response headers documentation

You can also add header information to your responses by using the desc DSL with block syntax.

By specifying headers to success and failure.

desc 'This returns headers' do
  success model: Thing, headers: { 'Location' => { description: 'Location of resource', type: 'string' } }
  failure [[404, 'NotFound', ApiError, { 'application/json' => { code: 404, message: 'Not found' } }, { 'Date' => { description: 'Date of failure', type: 'string' } }]]
end
get '/thing' do
  ...
end

The result will look like following:

  "responses": {
    "200": {
      "description": "This returns examples",
      "schema": {
        "$ref": "#/definitions/Thing"
      },
      "headers": {
        "Location": {
          "description": "Location of resource",
          "type": "string"
        }
      }
    },
    "404": {
      "description": "NotFound",
      "schema": {
        "$ref": "#/definitions/ApiError"
      },
      "examples": {
        "application/json": {
          "code": 404,
          "message": "Not found"
        }
      },
      "headers": {
        "Date": {
          "description": "Date of failure",
          "type": "string"
        }
      }
    }
  }

Failure information can be passed as an array of arrays or an array of hashes.

Adding root element to responses

You can specify a custom root element for a successful response:

route_setting :swagger, root: 'cute_kitten'
desc 'Get a kitten' do
  http_codes [{ code: 200, model: Entities::Kitten }]
end
get '/kittens/:id' do
end

The result will look like following:

  "responses": {
    "200": {
      "description": "Get a kitten",
      "schema": {
        "type": "object",
        "properties": { "cute_kitten": { "$ref": "#/definitions/Kitten" } }
      }
    }
  }

If you specify true, the value of the root element will be deduced based on the model name.
E.g. in the following example the root element will be “kittens”:

route_setting :swagger, root: true
desc 'Get kittens' do
  is_array true
  http_codes [{ code: 200, model: Entities::Kitten }]
end
get '/kittens' do
end

The result will look like following:

  "responses": {
    "200": {
      "description": "Get kittens",
      "schema": {
        "type": "object",
        "properties": { "type": "array", "items": { "kittens": { "$ref": "#/definitions/Kitten" } } }
      }
    }
  }

Multiple present Response

You can specify a custom multiple response by using the as key:

desc 'Multiple response',
  success: [
    { model: Entities::EnumValues, as: :gender },
    { model: Entities::Something, as: :somethings }
  ]
end

get '/things' do
  ...
end

The result will look like following:

  "responses": {
    "200": {
      "description": "Multiple response",
      "schema":{
        "type":"object",
        "properties":{
          "gender":{
            "$ref":"#/definitions/EnumValues"
          },
          "somethings":{
            "$ref":"#/definitions/Something"
          }
        }
      }
    }
  }

You can also specify if the response is an array, with the is_array key:

desc 'Multiple response with array',
  success: [
    { model: Entities::EnumValues, as: :gender },
    { model: Entities::Something, as: :somethings, is_array: true, required: true }
  ]
end

get '/things' do
  ...
end

The result will look like following:

  "responses": {
    "200": {
      "description": "Multiple response with array",
      "schema":{
        "type":"object",
        "properties":{
          "gender":{
            "$ref":"#/definitions/EnumValues"
          },
          "somethings":{
            "type":"array",
            "items":{
                "$ref":"#/definitions/Something"
            }
          }
        },
        "required": ["somethings"]
      }
    }
  }

Using Grape Entities

Add the grape-entity and grape-swagger-entity gem to your Gemfile.

The following example exposes statuses. And exposes statuses documentation adding :type, :desc and :required.
The documented class/definition name could be set via #entity_name.

module API
  module Entities
    class Status < Grape::Entity
      expose :text, documentation: { type: 'string', desc: 'Status update text.', required: true }
      expose :links, using: Link, documentation: { type: 'link', is_array: true }
      expose :numbers, documentation: { type: 'integer', desc: 'favourite number', values: [1,2,3,4] }
    end

    class Link < Grape::Entity
      expose :href, documentation: { type: 'url' }
      expose :rel, documentation: { type: 'string'}

      def self.entity_name
        'LinkedStatus'
      end

    end
  end

  class Statuses < Grape::API
    version 'v1'

    desc 'Statuses index',
      entity: API::Entities::Status
    get '/statuses' do
      statuses = Status.all
      type = current_user.admin? ? :full : :default
      present statuses, with: API::Entities::Status, type: type
    end

    desc 'Creates a new status',
      entity: API::Entities::Status,
      params: API::Entities::Status.documentation
    post '/statuses' do
        ...
    end
  end
end

Relationships

You may safely omit type from relationships, as it can be inferred. However, if you need to specify or override it, use the full name of the class leaving out any modules named Entities or Entity.

1xN

module API
  module Entities
    class Client < Grape::Entity
      expose :name, documentation: { type: 'string', desc: 'Name' }
      expose :addresses, using: Entities::Address,
        documentation: { type: 'Entities::Address', desc: 'Addresses.', param_type: 'body', is_array: true }
    end

    class Address < Grape::Entity
      expose :street, documentation: { type: 'string', desc: 'Street.' }
    end
  end

  class Clients < Grape::API
    version 'v1'

    desc 'Clients index',
      params: Entities::Client.documentation,
      success: Entities::Client
    get '/clients' do
      ...
    end
  end

  add_swagger_documentation
end

1x1

Note: is_array is false by default.

module API
  module Entities
    class Client < Grape::Entity
      expose :name, documentation: { type: 'string', desc: 'Name' }
      expose :address, using: Entities::Address,
        documentation: { type: 'Entities::Address', desc: 'Addresses.', param_type: 'body', is_array: false }
    end

    class Address < Grape::Entity
      expose :street, documentation: { type: 'string', desc: 'Street' }
    end
  end

  class Clients < Grape::API
    version 'v1'

    desc 'Clients index',
      params: Entities::Client.documentation,
      success: Entities::Client
    get '/clients' do
      ...
    end
  end

  add_swagger_documentation
end

Inheritance with allOf and discriminator

module Entities
  class Pet < Grape::Entity
    expose :type, documentation: {
      type: 'string',
      is_discriminator: true,
      required: true
    }
    expose :name, documentation: {
      type: 'string',
      required: true
    }
  end

  class Cat < Pet
    expose :huntingSkill, documentation: {
      type: 'string',
      description: 'The measured skill for hunting',
      default: 'lazy',
      values: %w[
        clueless
        lazy
        adventurous
        aggressive
      ]
    }
  end
end

Should generate this definitions:

{
  "definitions": {
    "Pet": {
      "type": "object",
      "discriminator": "petType",
      "properties": {
        "name": {
          "type": "string"
        },
        "petType": {
          "type": "string"
        }
      },
      "required": [
        "name",
        "petType"
      ]
    },
    "Cat": {
      "description": "A representation of a cat",
      "allOf": [
        {
          "$ref": "#/definitions/Pet"
        },
        {
          "type": "object",
          "properties": {
            "huntingSkill": {
              "type": "string",
              "description": "The measured skill for hunting",
              "default": "lazy",
              "enum": [
                "clueless",
                "lazy",
                "adventurous",
                "aggressive"
              ]
            },
            "petType": {
              "type": "string",
              "enum": ["Cat"]
            }
          },
          "required": [
            "huntingSkill",
            "petType"
          ]
        }
      ]
    }
  }
}

Securing the Swagger UI

The Swagger UI on Grape could be secured from unauthorized access using any middleware, which provides certain methods:

  • some guard method, which could receive as argument a string or an array of authorization scopes;
  • a before method to be run in the Grape controller for authorization purpose;
  • a set of methods which will process the access token received in the HTTP request headers (usually in the
    ‘HTTP_AUTHORIZATION’ header) and try to return the owner of the token.

Below are some examples of securing the Swagger UI on Grape installed along with Ruby on Rails:

  • The WineBouncer and Doorkeeper gems are used in the examples;
  • ‘rails’ and ‘wine_bouncer’ gems should be required prior to ‘grape-swagger’ in boot.rb;
  • This works with a fresh PR to WineBouncer which is yet unmerged - WineBouncer PR.

This is how to configure the grape_swagger documentation:

  add_swagger_documentation base_path: '/',
                            title: 'My API',
                            doc_version: '0.0.1',
                            hide_documentation_path: true,
                            endpoint_auth_wrapper: WineBouncer::OAuth2, # This is the middleware for securing the Swagger UI
                            swagger_endpoint_guard: 'oauth2 false',     # this is the guard method and scope
                            token_owner: 'resource_owner'               # This is the method returning the owner of the token

The guard method should inject the Security Requirement Object into the endpoint’s route settings (see Grape::DSL::Settings.route_setting method).

The ‘oauth2 false’ added to swagger_documentation is making the main Swagger endpoint protected with OAuth, i.e. the
access_token is being retrieving from the HTTP request, but the ‘false’ scope is for skipping authorization and
showing the UI for everyone. If the scope would be set to something else, like ‘oauth2 admin’, for example, than the UI
wouldn’t be displayed at all to unauthorized users.

Further on, the guard could be used, where necessary, for endpoint access protection. Put it prior to the endpoint’s method:

  resource :users do
    oauth2 'read, write'
    get do
      render_users
    end

    oauth2 'admin'
    post do
      User.create!...
    end
  end

And, finally, if you want to not only restrict the access, but to completely hide the endpoint from unauthorized
users, you could pass a lambda to the :hidden key of a endpoint’s description:

  not_admins = lambda { |token_owner = nil| token_owner.nil? || !token_owner.admin? }

  resource :users do
    desc 'Create user', hidden: not_admins
    oauth2 'admin'
    post do
      User.create!...
    end
  end

The lambda is checking whether the user is authenticated (if not, the token_owner is nil by default), and has the admin
role - only admins can see this endpoint.

Example

Go into example directory and run it: $ bundle exec rackup
go to: http://localhost:9292/swagger_doc to get it

For request examples load the postman file

Grouping the API list using Namespace

Use namespace for grouping APIs

grape-swagger-v2-new-corrected

Example Code

class NamespaceApi < Grape::API
  namespace :hudson do
    desc 'Document root'
    get '/' do
    end

    desc 'This gets something.',
      detail: '_test_'

    get '/simple' do
      { bla: 'something' }
    end
  end

  namespace :download do
    desc 'download files',
         success: File,
         produces: ['text/csv']
    get ':id' do
      # file response
    end
  end
end
  …

Rake Tasks

Add these lines to your Rakefile, and initialize the Task class with your Api class.

require 'grape-swagger/rake/oapi_tasks'
GrapeSwagger::Rake::OapiTasks.new(::Api::Base)

You may initialize with the class name as a string if the class is not yet loaded at the time Rakefile is parsed:

require 'grape-swagger/rake/oapi_tasks'
GrapeSwagger::Rake::OapiTasks.new('::Api::Base')

OpenApi/Swagger Documentation

rake oapi:fetch
params:
- store={ true | file_name.json } – save as JSON (optional)
- resource=resource_name     – get only for this one (optional)

For multiversion API it creates several files with following naming: file_name_API_VERSION.json

OpenApi/Swagger Validation

requires: npm and swagger-cli to be installed

rake oapi:validate
params:
- resource=resource_name – get only for this one (optional)

Contributing to grape-swagger

See CONTRIBUTING.

Copyright and License

Copyright © 2012-2016 Tim Vandecasteele, ruby-grape and contributors. See LICENSE.txt for details.