ZRO: OpenApi 3 JSON-Doc Generator for Rails

Concise DSL for generating OpenAPI Specification 3 (OAS3, formerly Swagger3) JSON documentation for Rails application.

class Api::ExamplesController < ApiController
  api :update, 'POST update some thing' do
    path  :id, Integer
    query :token, String, desc: 'api token', length: 16
    form data: { phone: String }
  end
end

Hi, here is ZhanDao = ▽ =
It may be a very useful tool if you want to write API document clearly.
I'm looking forward to your issue and PR!

(Test cases are rich, like: api DSL and schema Obj)

• About OAS (OpenAPI Specification)
• Installation
• Configure
• DSL Usage
  • a.Basic DSL
    • a.1. route_base
    • a.2. doc_tag
    • a.3. components
    • a.4. api
    • a.5. api_dry
  • b. DSLs written inside api and api_dry's block
    • b.1. this_api_is_invalid!
    • b.2. desc
    • b.3. param family methods
    • b.4. request_body family methods
    • b.5. response family methods
    • b.6. callback
    • b.7. Authentication and Authorization
    • b.8. server
    • b.9. dry
  • c. DSLs written inside components's block
  • d. Schema and Type
    • d.1. (Schema) Type
    • d.2. Schema
    • d.3. Combined Schema
• Run! - Generate JSON documentation file
• Use Swagger UI(very beautiful web page) to show your Documentation
• Tricks
  • Write DSL somewhere else
  • Global DRYing
  • Auto generate description form enum
  • Skip or Use parameters define in api_dry
  • Atuo Generate index/show Actions's Responses Based on DB Schema
• Troubleshooting
• About OpenApi.docs and OpenApi.routes_index

Everything about OAS3 is on OpenAPI Specification

You can getting started from swagger.io

I suggest you should understand the basic structure of OAS3 at least. such as component (can help you reuse DSL code, when your apis are used with the same data structure).

Add this line to your Rails's Gemfile:

gem 'zero-rails_openapi'
# or
gem 'zero-rails_openapi', github: 'zhandao/zero-rails_openapi'

And then execute:

  $ bundle

Create an initializer, configure ZRO and define your OpenApi documents.

This is the simplest example:

# in config/initializers/open_api.rb
require 'open_api'

OpenApi::Config.class_eval do
  # Part 1: configs of this gem
  self.file_output_path = 'public/open_api'

  # Part 2: config (DSL) for generating OpenApi info
  open_api :doc_name, base_doc_classes: [ApiDoc]
  info version: '1.0.0', title: 'Homepage APIs'#, description: ..
  # server 'http://localhost:3000', desc: 'Internal staging server for testing'
  # bearer_auth :Authorization
end

1. file_output_path(required): The location where .json doc file will be output.
2. default_run_dry: defaults to run dry blocks even if the dry method is not called in the (Basic) DSL block. defaults to false.
3. doc_location: give regular expressions for file or folder paths. Dir[doc_location] will be require before document generates. this option is only for not writing spec in controllers.
4. rails_routes_file: give a txt's file path (which's content is the copy of rails routes's output). This will speed up document generation.
5. model_base: The parent class of models in your application. This option is for auto loading schema from database.
6. file_format

# 1. open_api
#   base_doc_classes should be an array of base classes of the classes you write the OpenApi spec in, 
#   like [ActionController::Base] (if you write spec in controllers).
open_api doc_name, base_doc_classes: []

# 2. info
info version:, title:, desc: '', **addition

# 3. server
# 4. security_scheme / base_auth / bearer_auth / api_key
# 5. global_security_require

See all the DSLs: config_dsl.rb

There are two kinds of DSL for this gem: basic and inside basic.

1. Basic DSLs are class methods which is for declaring your APIs, components, and spec code DRYing ...
2. DSLs written inside the block of Basic DSLs, is for declaring the parameters, responses (and so on) of the specified API and component.

For example:

# in app/controllers/api/api_controller.rb
class ApiController < ActionController::API
  include OpenApi::DSL
end

Here is the simplest usage:

class Api::ExamplesController < ApiController
  api :index, 'GET list' do
    query :page, Integer#, range: { ge: 1 }, default: 1
    query :rows, Integer#, desc: 'per page', range: { ge: 1 }, default: 10
  end
end

source code

# ** Method Signature
route_base path
# ** Usage
route_base 'api/v1/examples'

Usage: write the DSL somewhere else to simplify the current controller.

# ** Method Signature
doc_tag name: nil, **tag_info
# ** Usage
doc_tag name: 'ExampleTagName', description: "ExamplesController's APIs"#, externalDocs: ...

This method allows you to set the Tag (which is a node of OpenApi Object) of all the APIs in the class.

Tag's name defaults to controller_name.

# ** Method Signature
components(&block)
# ** Usage
components do
  # (block inside) DSL for defining components
  schema :DogSchema => [ { id: Integer, name: String }, dft: { id: 1, name: 'pet' } ]
  query! :UidQuery  => [ :uid, String, desc: 'uid' ]
  response :BadRqResp => [ 'bad request', :json ]
end

# to use component
api :action do
  query :doge, :DogSchema # to use a Schema component
  param_ref :UidQuery     # to use a Parameter component
  response_ref :BadRqResp # to use a Response component
end

Each RefObj is associated with components through component key.

We suggest that component keys should be camelized, and must be Symbol.

For defining API (or we could say controller action).

# ** Method Signature
api action_name, summary = '', id: nil, tag: nil, http: nil, dry: Config.default_run_dry, &block
# ** Usage
api :index, '(SUMMARY) this api blah blah ...', # block ...

Parameters explanation:

1. action_name: must be the same as controller action name
2. id: operationId
3. http: HTTP method (like: 'GET' or 'GET|POST')

This method is for DRYing. The blocks passed to api_dry will be executed to the specified APIs which are having the actions or tags in the class.

# ** Method Signature
api_dry action_or_tags = :all, &block
# ** Usage
api_dry :all, 'common response' # block ...
api_dry :index # block ...
api_dry :TagA # block ...

api_dry [:index, :show] do
  query #...
end

And then you should call dry method (detailed info) for executing the declared dry blocks:

api :index do
  dry
end

source code

These following methods in the block describe the specified API action: description, valid?, parameters, request body, responses, securities and servers.

(Here corresponds to OAS Operation Object)

this_api_is_expired!
this_api_is_unused!
this_api_is_under_repair!

# ** Method Signature
this_api_is_invalid!(*)
# ** Usage
this_api_is_invalid! 'cause old version'

After that, deprecated field of this API will be set to true.

# ** Method Signature
desc string
# ** Usage
desc "current API's description"

To define parameter for APIs.

param                              # 1. normal usage
param_ref                          # 2. links sepcified RefObjs (by component keys) to current parameters.
header,  path,  query,  cookie     # 3. passes specified parameter location (like header) to `param`
header!, path!, query!, cookie!    # 4. bang method of above methods
in_* by: { parameter_definations } # 5. batch definition, such as `in_path`, `in_query`
examples                           # 6. examples of parameters

The bang method and param_name (which's name is end of a exclamation point !) means this param is required. Without ! means optional. THE SAME BELOW.

# Part 1
# param_type:  location of parameter, like: query, path [A]
# param_name:  name of parameter, it can be Symbol or String [B]
# schema_type: type of parameter, like: String, Integer (must be a constant). see #schema-and-type
# required:    :required / :req OR :optional / :opt
# schema:      see #schema-and-type (including combined schema)
# ** Method Signature
param param_type, param_name, schema_type, required, schema = { }
# ** Usage
param :query, :page, Integer, :req,  range: { gt: 0, le: 5 }, desc: 'page number'

# Part 2
# ** Method Signature
param_ref *component_key # should pass at least 1 key
# ** Usage
param_ref :IdPath#, :NameQuery, :TokenHeader

# Part 3 & 4
# ** Method Signature
header param_name, schema_type = nil, **schema
query! param_name, schema_type = nil, **schema
# ** Usage
header :'X-Token', String
query! :readed, Boolean, default: false
# The same effect as above, but not concise
param :query, :readed, Boolean, :req, default: false

# Part 5 
# ** Method Signature
in_query **params_and_schema
# ** Usage
in_query(
  search_type: String,
   search_val: String,
      export!: { type: Boolean, desc: 'export as pdf' }
)
# The same effect as above
query  :search_type, String
query  :search_val, String
query! :export, Boolean, desc: 'export as pdf'

# Part 6
# ** Method Signature
examples exp_params = :all, examples_hash
# ** Usage
# Suppose we have three parameters: id, name, age
# * normal
examples(
  right_input: [ 1, 'user', 26 ],
  wrong_input: [ 2, 'resu', 35 ]
)
# * using exp_params
examples [:id, :name], {
  right_input: [ 1, 'user' ],
  wrong_input: [ 2, 'resu' ]
}

[A] OpenAPI 3.0 distinguishes between the following parameter types based on the parameter location: header, path, query, cookie. more info

[B] If param_type is path, for example: if the API path is /good/:id, you have to declare a path parameter named id

OpenAPI 3.0 uses the requestBody keyword to distinguish the payload from parameters.

Notice: Each API has only ONE request body object. Each request body object can has multiple media types. It means: call request_body multiple times, (schemas) will be deeply merged (let's call it fusion) into a request body object.

request_body  # 1. normal usage
body_ref      # 2. it links sepcified RefObjs (by component keys) to the body.
body, body!   # 3. alias of request_body
form, form!   # 4. to define a multipart/form-data request_body
json, json!   # 5. to define a application/json request_body
data          # 5. to define [a] property in the form-data request_body

Bang methods(!) means the specified media-type body is required.

# Part 1
# ** Method Signature
# a. `data` contains the attributes (params, or properties) and their schemas required by the request body
# b. `attr_name!` means it is required, without '!' means optional
# c. options: desc / exp_params and examples
# d. available `media_type` see: 
#   https://github.com/zhandao/zero-rails_openapi/blob/master/lib/oas_objs/media_type_obj.rb#L29
request_body required, media_type, data: { }, desc: '', **options
# ** Usage
request_body :opt, :form, data: {
  id!: Integer,
  name: { type: String, desc: 'name' }
}, desc: 'a form-data'

# Part 2
# ** Method Signature
body_ref component_key
# ** Usage
body_ref :UpdateUserBody

# Part 3
# ** Method Signature
body! media_type, data: { }, **options
# ** Usage
body :json

# Part 4
# ** method Implement
def form data:, **options # or `form!`
  body :form, data: data, **options
end
# ** Usage
form! data: {
       name!: String,
    password: { type: String, pattern: /[0-9]{6,10}/ },
}
json data: { name!: String }

# Part 5
# ** Method Signature
data name, type = nil, schema = { }
# ** Usage
data :password!, String, pattern: /[0-9]{6,10}/

How fusion works:

1. Difference media types will be merged into requestBody["content"]

form data: { }
body :json, data: { }
# will generate: "content": { "multipart/form-data": { }, "application/json": { } }

2. The same media-types will be deeply merged together, including their required array:
   (So that you can call form multiple times)

data :param_a!, String
data :param_b,  Integer
# or same as:
form data: { :param_a! => String }
form data: { :param_b  => Integer }
# will generate: { "param_a": { "type": "string" }, "param_b": { "type": "integer" } } (call it X)
# therefore:
#   "content": { "multipart/form-data":
#     { "schema": { "type": "object", "properties": { X }, "required": [ "param_a" ] }
#   }

To define the response for APIs.

response      # 1. aliases: `resp` and `error`
response_ref  # 2. it links sepcified RefObjs (by component keys) to the response.

# ** Method Signature
response code, desc, media_type = nil, headers: { }, data: { }, **options
# ** Usage
resp 200, 'success', :json, data: { name: 'test' }
response 200, 'query result', :pdf, data: File
response :success, 'succ', :json, headers: { 'X-Request-Start': String }, data: { }

# ** Method Signature
response_ref code_and_compkey_hash
# ** Usage
response_ref 700 => :AResp, 800 => :BResp

About Callbacks

In OpenAPI 3 specs, you can define callbacks – asynchronous, out-of-band requests that your service will send to some other service in response to certain events. This helps you improve the workflow your API offers to clients.
A typical example of a callback is a subscription functionality ... you can define the format of the “subscription” operation as well as the format of callback messages and expected responses to these messages.
This description will simplify communication between different servers and will help you standardize use of webhooks in your API.
Complete YAML Example

The structure of Callback Object:

callbacks:
  Event1:
    path1:
      ...
    path2:
     ...
  Event2:
    ...

callback method is for defining callbacks.

# ** Method Signature
callback event_name, http_method, callback_url, &block
# ** Usage
callback :myEvent, :post, 'localhost:3000/api/goods' do
  query :name, String
  data :token, String
  response 200, 'success', :json, data: { name: String, description: String }
end

Use runtime expressions in callback_url:

callback :myEvent, :post, '{body callback_addr}/api/goods/{query id}'
# the final URL will be: {$request.body#/callback_addr}/api/goods/{$request.query.id}
# Note: Other expressions outside "$request" are not supported yet

First of all, please make sure that you have read one of the following documents: OpenApi Auth or securitySchemeObject

Use these DSL in your initializer config or components block:

security_scheme # alias `auth_scheme`
base_auth       # will call `security_scheme`
bearer_auth     # will call `security_scheme`
api_key         # will call `security_scheme`

It's very simple to use (if you understand the above document)

# ** Method Signature
security_scheme scheme_name, other_info
# ** Usage
security_scheme :BasicAuth, { type: 'http', scheme: 'basic', desc: 'basic auth' }

# ** Method Signature
base_auth scheme_name, other_info = { }
bearer_auth scheme_name, format = 'JWT', other_info = { }
api_key scheme_name, field:, in:, **other_info
# ** Usage
base_auth :BasicAuth, desc: 'basic auth' # the same effect as above
bearer_auth :Token
api_key :ApiKeyAuth, field: 'X-API-Key', in: 'header', desc: 'pass api key to header'

# Use in initializer (Global effectiveness)
global_security_require # alias: global_security & global_auth

# Use in `api`'s block (Only valid for the current controller)
security_require # alias security & auth_with

# ** Method Signature
security_require scheme_name, scopes: [ ]
# ** Usage
global_auth :Token
auth_with   :OAuth, scopes: %w[ read_example admin ]

# ** Method Signature
server url, desc: ''
# ** Usage
server 'http://localhost:3000', desc: 'local'

You have to call dry method inside api block, or pass dry: true as parameter of api, for executing the dry blocks you declared before. Otherwise nothing will happen.

# ** Method Signature
dry only: nil, skip: nil, none: false

# ** Usage
# In general, just:
dry
# To skip some params declared in dry blocks:
dry skip: [:id, :name]
# `only` is used to specify which parameters will be taken from dry blocks
dry only: [:id]

code source (Here corresponds to OAS Components Object)

Inside components's block, you can use the same DSLs as DSLs written inside api and api_dry's block.
But notice there are two differences:

(1) Each method needs to pass one more parameter component_key (as the first parameter), it will be used as the reference name for the component.

query! :UidQuery, :uid, String， desc: 'it is a component'
#         ↑         ↑
# component_key  param_name

# You can also use "arrow writing", it may be easier to understand
query! :UidQuery => [:uid, String, desc: '']

(2) You can use schema to define a Schema Component.

# ** Method Signature
schema component_key, type = nil, **schema
# ** Usage
schema :Dog  => [ String, desc: 'doge' ]
# advance usage
schema :Dog => [
    {
         id!: Integer,
        name: { type: String, desc: 'doge name' }
    }, default: { id: 1, name: 'pet' }
]
# or flatten writing
schema :Dog, { id!: Integer, name: String }, default: { id: 1, name: 'pet' }
#
# pass a ActiveRecord class constant as `component_key`,
#   it will automatically load schema from database and then generate the component.
schema User # easy! And the component_key will be :User

To enable load schema from database, you must set model base correctly.

schema and type -- contain each other

Support all data types in OAS.

1. String / 'binary' / 'base64' / 'uri'
2. Integer / Long / 'int32' / 'int64' / Float / Double
3. File (it will be converted to { type: 'string', format: Config.file_format })
4. Date / DateTime
5. 'boolean'
6. Array / Array[<Type>] (like: Array[String], [String])
7. Nested Array (like: [[[Integer]]])
8. Object / Hash (Object with properties)
   Example: { id!: Integer, name: String }
9. Nested Hash: { id!: Integer, name: { first: String, last: String } }
10. Nested Array[Nested Hash]: [[{ id!: Integer, name: { first: String, last: String } }]]
11. Symbol Value: it will generate a Schema Reference Object link to the component correspond to ComponentKey, like: :IdPath, :NameQuery

Notice that Symbol is not allowed in all cases except 11.

OAS Schema Object and source code

Schema (Hash) is for defining properties of parameters, responses and request bodies.

The following property keys will be process slightly:

1. desc / description / d
2. enum / in / values / allowable_values
   should be Array or Range
3. range: allow value in this continuous range
   should be Range or like { gt: 0, le: 5 }
4. length / size / lth
   should be an Integer, Integer Array, Integer Range, or the following format Symbol: :gt_, :ge_, :lt_, :le_ (:ge_5 means "greater than or equal 5"; :lt_9 means "lower than 9")
5. pattern / regxp
6. additional_properties / add_prop / values_type
7. example
8. examples
9. format
10. default: default value
11. type

The other keys will be directly merged. Such as:

1. title: 'Property Title'
2. myCustomKey: 'Value'

Very easy to use:

query :combination, one_of: [ :GoodSchema, String, { type: Integer, desc: 'integer input' } ]

form data: {
    :combination_in_form => { any_of: [ Integer, String ] }
}

schema :PetSchema => [ not: [ Integer, Boolean ] ]

OAS: link1, link2

Use OpenApi.write_docs:

OpenApi.write_docs# if: !Rails.env.production?

if option is used to control whether a JSON document is generated or not.

Then the JSON files will be written to the directories you set. (Each API a file.)

Download Swagger UI (version >= 2.3.0 support the OAS3) to your project, change the default JSON file path(url) in index.html. In order to use it, you may have to enable CORS, see

Does your documentation take too many lines?
Do you want to separate documentation from controller to simplify both?
Very easy! Just follow

# config/initializers/open_api.rb
# in your configuration
base_doc_classes: [ApiDoc]

# app/api_doc/api_doc.rb
require 'open_api/dsl'

class ApiDoc < Object
  include OpenApi::DSL
end

# app/api_doc/v1/examples_doc.rb
class V1::ExamplesDoc < ApiDoc
  route_base 'api/v1/examples'

  api :index do
    # ...
  end
end

Explain: These four steps are necessary:

1. create a class, like ApiDoc, and make it include OpenApi::DSL (then it could be the base class for writing Api spec).
2. set the specified Api spec's base_doc_classes to ApiDoc.
3. let your doc class (like V1::ExamplesDoc) inherit the base_doc_classes (ApiDoc).
4. set the route_base (to route path api/v1/examples of that controller Api::V1::ExamplesController) inside V1::ExamplesDoc.

Notes: file name ends in _doc.rb by default, but you can change it by setting Config.doc_location (it should be file paths, defaults to ./app/**/*_doc.rb).

Method api_dry is for DRY but its scope is limited to the current controller.

I have no idea of best practices, But you can look at this file.
The implementation of the file is: do api_dry when inherits the base controller inside inherited method.

You can use sort to specify the order of parameters.

Just use enum!:

query :search_type, String, desc: 'search field, allows：<br/>', enum!: %w[name creator category price]
# it will generate: 
"search field, allows：<br/>1/ name<br/>2/ creator,<br/>3/ category<br/>4/ price<br/>"

Or Hash enum!:

query :view, String, desc: 'allows values<br/>', enum!: {
        'all goods (default)': :all,
                'only online': :online,
               'only offline': :offline,
            'expensive goods': :get,
                'cheap goods': :borrow,
}

• You wrote document of the current API, but not find in the generated json file?
  Check your routing settings.

• Report error when require routes.rb?*

  1. Run rails routes.
  2. Copy the output to a file, for example config/routes.txt. Ignore the file config/routes.txt.
  3. Put c.rails_routes_file = 'config/routes.txt' to your ZRO config.

After OpenApi.write_docs, the above two module variables will be generated.

OpenApi.docs: A Hash with API names as keys, and documents of each APIs as values.
documents are instances of ActiveSupport::HashWithIndifferentAccess.

OpenApi.routes_index: Inverted index of controller path to API name mappings.
Like: { 'api/v1/examples' => :homepage_api }
It's useful when you want to look up a document based on a controller and do something.

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.

The gem is available as open source under the terms of the MIT License.

Everyone interacting in the Zero-RailsOpenApi project’s codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.