A library that generates swagger specs from route files and case class reflection, no code annotation needed.
- No code pollution (e.g. annotation)
- DRY (extract as much information from the code as possible)
- When documenting an endpoint, it should be just swagger specification that you need to write. You shall not need to learn another API or spec format.
Which translates to
- Write your swagger specification in your routes files as comments (json or yml)
- Reference your case classes in your swagger spec and play-swagger will generate definitions
- Override anything in either the swagger spec in comment or the base swagger spec file (swagger.yml or swagger.json in your conf)
============================
For installation/get-started see the next section.
In a cards.routes
which is referenced in routes
as
-> /api/cards cards.Routes
You can write the following swagger spec in comment (This example is in yml, but json is also supported). The comment has to start and end with ###
.
If you don't write any comment here the endpoint is still going to be picked up by play-swagger, the parameters will be included but there will not be any response format. This allows newly added endpoints to be automatically included in swagger with some basic information.
###
# summary: create a card
# tags:
# - Card Endpoints
# responses:
# 200:
# description: success
# schema:
# $ref: '#/definitions/com.iheart.api.Protocol.CardCreated'
###
POST /users/:profileId/contexts/:contextName/cards controllers.api.Cards.createCard(profileId: Int, contextName: Option[String])
Note that everything in the comment is just standard swagger definition, and it $refs to a case class CardCreated, which is defined in a Protocol object, and it references another case class Card. Here is the source code:
package com.iheart.api
object Protocol {
case class CardCreated(card: Card)
case class Card(id: Int, name: String)
}
This will generate the path with summary, tags, parameters and a response with schema defined, which comes from the comments and case class reflection.
It also recursively generates definitions from your case class.
These schemas assumes that you are using a simple Json.format[CardCreated]
to generate the json response out of this class.
If not, you will have to write the definition yourself in the base swagger spec and reference it here at the endpoint
(give it a different name than the full package name though, play-swagger will try to generate definitions for any $ref that starts with the domain package name).
The result swagger specs will look like:
============================
In short you need to add sbt-play-swagger plugin which generates swagger.json on package time, then you just need to have a swagger UI instance to consumer that swagger spec. You can find the setup in the example project as well.
- For play 2.8, Scala 2.13.x and Scala 2.12.x please use
addSbtPlugin("io.github.play-swagger" % "sbt-play-swagger" % "1.6.1")
- For play 2.9, Scala 2.13.x please use (Code is in branch 1.7.0)
addSbtPlugin("io.github.play-swagger" % "sbt-play-swagger" % "1.7.0")
- For play 3.0, Scala 2.13.x please use (Code is in branch 2.0.0 and scala 3 is not supported)
addSbtPlugin("io.github.play-swagger" % "sbt-play-swagger" % "2.0.0")
Then enable it for your Play app - in build.sbt add SwaggerPlugin
to the root project like
lazy val root = (project in file(".")).enablePlugins(PlayScala, SwaggerPlugin) //enable plugin
Also in build.sbt add domain package names for play-swagger to auto generate swagger definitions for domain classes mentioned in your routes
swaggerDomainNameSpaces := Seq("models")
To be more specific, If you want to use the case class defined in the package something.models
in swagger (accessed via #ref:definitions/
), add the following in sbt.
swaggerDomainNameSpaces := Seq("something.models")
Additionally, if you want to use other packages (e.g. other.models
), you can do so like this.
swaggerDomainNameSpaces := Seq("something.models","other.models")
This plugin adds a sbt task swagger
, with which you can generate the swagger.json
for testing purpose.
This plugin will generate the swagger.json
and make it available under path assets/swagger.json
on sbt package
and sbt run
.
Alternatively, you can create a controller that uses play-swagger lib to generate the json and serve it, this way you can manipulate the swagger.json at runtime. See here for details.
Add a base swagger.yml
(or swagger.json
) to your resources
folder (for example, conf folder in the play application). This file needs to provide all the required fields according to swagger spec.
E.g.
---
swagger: "2.0"
info:
title: "Poweramp API"
description: "Power your music"
consumes:
- application/json
produces:
- application/json
Note that info.version
is intentionally left out, playSwagger will automatically fillin the build version of the project. However if the version is set here it will be honored. You can also dynamically generate the version string in build file using the swaggerAPIVersion
setting.
Deploy a swagger ui and point to the swagger spec end point at 'assets/swagger.json', or
Alternatively you can use swagger-ui webjar and have you play app serving the swagger ui:
The query parameter url
is disabled in 4.1.3 and later versions. (GHSA-qrmm-w75w-3wpx)
libraryDependencies += "org.webjars" % "swagger-ui" % "5.10.3"
Copy the index.html
and swagger-initializer.js
generated in target/${project}/public/lib/main/swagger-ui/
and modify the js files as follows to create Swagger-UI can be used easily.
window.onload = function() {
window.ui = SwaggerUIBundle({
// edit url
url: "/assets/swagger.json",
dom_id: '#swagger-ui',
deepLinking: true,
presets: [
SwaggerUIBundle.presets.apis,
SwaggerUIStandalonePreset
],
plugins: [
SwaggerUIBundle.plugins.DownloadUrl
],
layout: "StandaloneLayout"
});
};
For more information: installation.md
The sbt-play-swagger plugin will generate the swagger.json on sbt run
or sbt package
.
============================
If you have any questions/bug reports, please submit an issue on github. With good unit tests coverage, it's pretty easy to add/modify this library as well. Code contribution are more than welcome. Make sure that your code is tested and submit the pull request!
===========================
To override any of the automatically generated field, you just need to write the same part in your comment or base swagger spec file.
To override the type mapping used for any type, create a swagger-custom-mappings.yml or swagger-custom-mappings.json in conf and add an array of mappings. Each mapping consists of
type
a regex matching the type for which the custom specs isspecAsParameter
a list of objects to be used when this type is used by a route path parameter or query string parameter. Being a list of json object allows you to expand a single parameter into multiple ones, but in other cases you just need to provide one json object. If you leave this one an empty array, the parameter with this type will be hidden. You must provide a list here, though.specAsProperty
, a json object to be used when the type is used as a property in a definition. If you leave this one empty, play-swagger will try to use the first element inspecAsParameter
, please note that although most of the fields are common between the two types of spec, a couple of them aren't.
For example
---
- type: java\.time\.LocalDate
specAsParameter:
- type: string
format: date
- type: java\.time\.Duration
specAsParameter: [] #this hides the type from query and path parameter
specAsProperty:
$ref: "#/definitions/Duration"
The preceding example would result in output for a field with type java.time.LocalDate
like this:
"fieldName": {
"type":"string",
"format":"date"
}
To use a custom naming strategies to override your case classes field names, you need to add this to your build.sbt
:
//default is 'none', which is your camelCased case class
swaggerNamingStrategy := "snake_case" //snake_case_skip_number, kebab-case, lowercase and UpperCamelCase also available
@mosche answered this one in #114
It's a bit unfortunate the way the docker plugin redefines stage. However, the solution is pretty simple. Just add:
(stage in Docker) <<= (stage in Docker).dependsOn(swagger)
If you don't want an end point to be included, add ### NoDocs ###
in front of it
e.g.
### NoDocs ###
GET /docs/swagger-ui/*file controllers.Assets.at(path:String="/public/lib/swagger-ui", file:String)
The entire file can be skipped by adding ### SkipFileForDocs ###
at the beginning of the routes file.
Alternatively, the routes file can be split into multiple files, so that you can skip practically only a part of the file.
https://www.playframework.com/documentation/ja/2.4.x/SBTSubProjects
### SkipFileForDocs ###
GET /api/hidden/a controllers.hiddenEndPointA()
GET /api/hidden/b controllers.hiddenEndPointB()
GET /api/hidden/c controllers.hiddenEndPointC()
Body content is specified as a special parameter in swagger. So you need to create a parameter in your swagger spec comment as "body", for example
###
# parameters:
# - name: body
# schema:
# $ref: '#/definitions/com.iheart.api.Track'
###
POST /tracks controller.Api.createTrack()
Again, play-swagger will generate the definition for com.iheart.api.Track case class
Use the alternative setup. The library returns play JsObject, you can change however you want like
val spec: Try[JsObject] = ps.generate().map(_ + ("host" -> JsString(myHost)))
In build.sbt, add
swaggerRoutesFile := "my-routes"
or if you took the alternative setup
SwaggerSpecGenerator(domainPackage).generate("myRoutes.routes")
In build.sbt, add
swaggerTarget := new File("path/to/swagger/location")
In build.sbt, add
swaggerFileName := "customSwagger.json"
In build.sbt, add
swaggerPrettyJson := true
Generic types in schema definitions for request/response body is supported. Example:
package models
case class Foo[T](payload: T)
case class AnotherOne(someString: String)
One can, then, reference the schema directly with models.Foo[models.AnotherOne]
and a correct OpenAPI 3 spec will be generated (not tested with Swagger 2.0):
###
# summary: Get a message
# responses:
# 200:
# description: success
# content:
# application/json:
# schema:
# $ref: '#/components/schemas/models.Foo[models.AnotherOne]'
###
GET /message controllers.AsyncController.parametric
The generated schema name, however, cannot contain [
, ]
or ,
which appear in type argument lists in Scala.
Therefore, there's a default OutputTransformer
(ParametricTypeNamesTransformerSpec
) which normalises the name into the URL-compliant form.
The definitions output would then look like:
{
"components": {
"schemas": {
"models.AnotherOne": {
"properties": {
"someString": {
"type": "string"
}
},
"required": [
"someString"
]
},
"models.Foo-models.AnotherOne": {
"properties": {
"payload": {
"$ref": "#/components/schemas/models.AnotherOne"
}
},
"required": [
"payload"
]
}
}
},
// ...
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/models.Foo-models.AnotherOne"
}
}
}
//...
}
In the tests!
You can use markup on your swagger specs by providing OutputTransformers
classes name to the setting swaggerOutputTransformers
on your build file.
For example you can use environment variables by adding the configuration:
swaggerOutputTransformers := Seq(envOutputTransformer)
Then on your routes file or root swagger file you can use some markup like the one used below for the host field:
swagger: "2.0"
info:
title: "API"
description: "REST API"
host: ${API_HOST}
This way when the swagger file is parsed the markup ${API_HOST}
is going to be substituted by the content of the environent variable API_HOST
.
You can produce swagger files for v3 by setting the flag:
swaggerV3 := true
Make sure you also update your swagger markup to specify that you are using OpenAPI v3:
openapi: 3.0.0
info:
title: "API"
description: "REST API"
version: "1.0.0"
Also, for $ref
fields you will want to prefix paths with #/components/schemas/
instead of #/definitions/
. For example:
###
# parameters:
# - name: body
# schema:
# $ref: '#/components/schemas/com.iheart.api.Track'
###
POST /tracks controller.Api.createTrack()
By placing a json or YAML file in conf/${dir}/${file}
and referencing it with $ref
in a comment, the file can be generated embedded in swagger.json.
/conf
is specified, or if a typo is used for the filename, $ref:"${filename}"
will be output as is.
example conf/routes
file.
###
# summary: Top Page
# responses:
# 200:
# $ref: './swagger/home_200.yml'
###
GET / controllers.HomeController.index
example conf/swagger/home_200.yml
file.
description: "success"
Of course, writing schema
etc. will also be embedded.
Generated swagger.json
.
{
"paths": {
"/": {
"get": {
"operationId": "index",
"tags": [
"routes"
],
"summary": "Top Page",
"responses": {
"200": {
"description": "success"
}
}
}
}
}
// ......
}
See the following document for information on how to refer to other files by "$ref".
https://swagger.io/docs/specification/using-ref/
This feature is very useful, but OpenAPI does not allow top-level $ref
, so failing to embed it may result in an invalid swagger.json
!
###
# $ref: './swagger/home.yml'
###
GET / controllers.HomeController.index
example home.yml
file.
summary: Top Page
responses:
200:
description: "success"
It can be configured in build.sbt
.
This setting allows you to set the ${controllerName}.${methodName}
to name the operationId.
swaggerOperationIdNamingFully := true
Using runtime-scaladoc-reader, a description can be generated from Scaladoc comments written in the case class.
Add the required dependencies and Compiler Plugin to build.sbt
and configure it for use.
embedScaladoc := true
addCompilerPlugin("com.github.takezoe" %% "runtime-scaladoc-reader" % "1.0.3")
libraryDependencies += "com.github.takezoe" %% "runtime-scaladoc-reader" % "1.0.3"
For example, a case class might be written as follows.
/**
* @param name e.g. Sunday, Monday, TuesDay...
*/
case class DayOfWeek(name: String)
The generated JSON will look like this.
{
"DayOfWeek": {
"properties": {
"name": {
"type": "string",
"description": "e.g. Sunday, Monday, TuesDay..."
}
},
"required": [
"name"
]
}
}
you can generate models definition from java POJO by setting the flag:
playJava := true
The flag only suport PlayJava 2.7 and 2.8