1. Home
2. Working with Play
3. Common topics
4. Static assets

§Working with public assets

Serving a public resource in Play is the same as serving any other HTTP request. It uses the same routing as regular resources using the controller/action path to distribute CSS, JavaScript or image files to the client.

§The public/ folder

By convention public assets are stored in the public folder of your application. This folder can be organized the way that you prefer. We recommend the following organization:

public
 └ javascripts
 └ stylesheets
 └ images

If you follow this structure it will be simpler to get started, but nothing stops you to modifying it once you understand how it works.

§WebJars

WebJars provide a convenient and conventional packaging mechanism that is a part of SBT. For example you can declare that you will be using the popular Bootstrap library simply by adding the following dependency in your build file:

libraryDependencies += "org.webjars" % "bootstrap" % "3.3.6"

WebJars are automatically extracted into a lib folder relative to your public assets for convenience. For example, if you declared a dependency on RequireJs then you can reference it from a view using a line like:

<script data-main="@routes.Assets.at("javascripts/main.js")" type="text/javascript" src="@routes.Assets.at("lib/requirejs/require.js")"></script>

Note the lib/requirejs/require.js path. The lib folder denotes the extracted WebJar assets, the requirejs folder corresponds to the WebJar artifactId, and the require.js refers to the required asset at the root of the WebJar. To clarify, the requirejs webjar dependency is declared at your build file like:

libraryDependencies += "org.webjars" % "requirejs" % "2.2.0"

§How are public assets packaged?

During the build process, the contents of the public folder are processed and added to the application classpath.

When you package your application, all assets for the application, including all sub projects, are aggregated into a single jar, in target/my-first-app-1.0.0-assets.jar. This jar is included in the distribution so that your Play application can serve them. This jar can also be used to deploy the assets to a CDN or reverse proxy.

§The Assets controller

Play comes with a built-in controller to serve public assets. By default, this controller provides caching, ETag, gzip and compression support. There are two different styles that the Assets controller supports: the first is to use Play’s configuration, and the second is to use pass the assets path directly to the controller.

§Binding the Assets components

If you are using runtime dependency injection, Play already provides bindings in the AssetsModule, which is loaded by default. (If you are not using assets, you can disable this module by adding the configuration play.modules.disabled += controllers.AssetsModule.). The bindings there make Assets class injectable.

If you are using components traits to do compile-time dependency injection, you should mix in controllers.AssetsComponents. Then the controller will be available as assets: Assets. You do not need to construct the controller yourself.

§Using assets with configuration

For the most common case where you only have one place where assets are centrally located, you can use configuration to specify the location:

play.assets {
  path = "/public"
  urlPrefix = "/assets"
}

And use the Assets.at method with one parameter:

Assets.at(file: String)

Then in routes:

GET  /assets/*file        controllers.Assets.at(file)

§Passing the assets path directly

The Assets controller also defines an at action with two parameters:

Assets.at(path: String, file: String)

The path parameter must be fixed and defines the directory managed by the action. The file parameter is usually dynamically extracted from the request path.

Here is the typical mapping of the Assets controller in your conf/routes file:

GET  /assets/*file        controllers.Assets.at(path="/public", file)

Note that we define the *file dynamic part that will match the .* regular expression. So for example, if you send this request to the server:

GET /assets/javascripts/jquery.js

The router will invoke the Assets.at action with the following parameters:

controllers.Assets.at("/public", "javascripts/jquery.js")

To route to a single static file, both the path and file have to be specified:

GET  /favicon.ico        controllers.Assets.at(path="/public", file="favicon.ico")

§Reverse routing for public assets

As for any controller mapped in the routes file, a reverse controller is created in controllers.routes.Assets. You use this to reverse the URL needed to fetch a public resource. For example, from a template:

<script src="@routes.Assets.at("javascripts/jquery.js")"></script>

In DEV mode this will by default produce the following result:

<script src="/assets/javascripts/jquery.js"></script>

If your app is not running in DEV mode and a jquery.min.js or jquery-min.js file exists then by default the minified file will be used instead:

<script src="/assets/javascripts/jquery.min.js"></script>

This makes debugging of JavaScript files easier during development. Of course this not only works for JavaScript files but for any file extension.
If you don’t want Play to automatically resolve the .min.* or -min.* files, regardless of the mode your application is running in, you can set play.assets.checkForMinified = false in your application.conf (or to true to always resolve the min file, even in DEV mode).

Note that we don’t specify the first folder parameter when we reverse the route. This is because our routes file defines a single mapping for the Assets.at action, where the folder parameter is fixed. So it doesn’t need to be specified.

However, if you define two mappings for the Assets.at action, like this:

GET  /javascripts/*file        controllers.Assets.at(path="/public/javascripts", file)
GET  /images/*file             controllers.Assets.at(path="/public/images", file)

You will then need to specify both parameters when using the reverse router:

<script src="@routes.Assets.at("/public/javascripts", "jquery.js")"></script>
<img src="@routes.Assets.at("/public/images", "logo.png")" />

§Reverse routing and fingerprinting for public assets

sbt-web brings the notion of a highly configurable asset pipeline to Play e.g. in your build file:

pipelineStages := Seq(rjs, digest, gzip)

The above will order the RequireJs optimizer (sbt-rjs), the digester (sbt-digest) and then compression (sbt-gzip). Unlike many sbt tasks, these tasks will execute in the order declared, one after the other.

In essence asset fingerprinting permits your static assets to be served with aggressive caching instructions to a browser. This will result in an improved experience for your users given that subsequent visits to your site will result in less assets requiring to be downloaded. Rails also describes the benefits of asset fingerprinting.

The above declaration of pipelineStages and the requisite addSbtPlugin declarations in your plugins.sbt for the plugins you require are your start point. You must then declare to Play what assets are to be versioned.

There are two ways obtain the real path of a fingerprinted asset. The first way uses static state and supports the same style as normal reverse routing. It does so by looking up assets metadata that’s set by a running Play application. The second way is to use configuration and inject an AssetsFinder to find your asset.

§Using reverse routing and static state

If you plan to use the reverse router with static state, the following routes file entry declares that all assets are to be versioned:

GET  /assets/*file  controllers.Assets.versioned(path="/public", file: Asset)

Note: Make sure you indicate that file is an asset by writing file: Asset.

You then use the reverse router, for example within a scala.html view:

<link rel="stylesheet" href="@routes.Assets.versioned("assets/css/app.css")">

The downside of this approach is that it requires special logic that converts the Asset from the path you passed in to the final minified path with a digest. It’s also more difficult to unit test, since there’s no component you can mock to define the path.

§Using configuration and AssetsFinder

You can also define your paths in configuration, and inject an AssetsFinder into your controller to get the final path. In your configuration set up the assets path (the directory containing assets) and the urlPrefix (the prefix to the URL in your application):

play.assets {
  path = "/public"
  urlPrefix = "/assets"
}

In your routes file you can define a route as follows:

GET  /assets/*file        controllers.Assets.versioned(file)

(you should not use the : Asset type annotation here)

Then you can pass an AssetsFinder to your template and use that to get the final path:

@(assetsFinder: AssetsFinder)

<link rel="stylesheet" href="@assetsFinder.path("assets/css/app.css")">

The advantage to this approach is that it requires no static state to set up. That means you can unit test your controllers and templates without a running application by simply passing an instance of AssetsFinder. That makes it simple to mock for a unit test by simply implementing the abstract methods that return Strings.

Using the AssetsFinder approach also makes it easy to run multiple self-contained applications at once in the same classloader, since it uses no static state. This can also be helpful for testing.

The AssetsFinder interface also works in cases where fingerprinting is not used. It returns the original asset if a fingerprinted and/or minified asset cannot be found.

§Etag support

The Assets controller automatically manages ETag HTTP Headers. The ETag value is generated from the digest (if sbt-digest is being used in the asset pipeline) or otherwise the resource name and the file’s last modification date. If the resource file is embedded into a file, the JAR file’s last modification date is used.

When a web browser makes a request specifying this Etag then the server can respond with 304 NotModified.

§Gzip support

If a resource with the same name but using a .gz suffix is found then the Assets controller will also serve the latter and add the following HTTP header:

Content-Encoding: gzip

Including the sbt-gzip plugin in your build and declaring its position in the pipelineStages is all that is required to generate gzip files.

§Additional Cache-Control directive

Using Etag is usually enough for the purposes of caching. However if you want to specify a custom Cache-Control header for a particular resource, you can specify it in your application.conf file. For example:

# Assets configuration
# ~~~~~
"play.assets.cache./public/stylesheets/bootstrap.min.css"="max-age=3600"

§Managed assets

Starting with Play 2.3 managed assets are processed by sbt-web based plugins. Prior to 2.3 Play bundled managed asset processing in the form of CoffeeScript, LESS, JavaScript linting (ClosureCompiler) and RequireJS optimization. The following sections describe sbt-web and how the equivalent 2.2 functionality can be achieved. Note though that Play is not limited to this asset processing technology as many plugins should become available to sbt-web over time. Please check-in with the sbt-web project to learn more about what plugins are available.

Many plugins use sbt-web’s js-engine plugin. js-engine is able to execute plugins written to the Node API either within the JVM via the excellent Trireme project, or directly on Node.js for superior performance. Note that these tools are used during the development cycle only and have no involvement during the runtime execution of your Play application. If you have Node.js installed then you are encouraged to declare the following environment variable. For Unix, if SBT_OPTS has been defined elsewhere then you can:

export SBT_OPTS="$SBT_OPTS -Dsbt.jse.engineType=Node"

The above declaration ensures that Node.js is used when executing any sbt-web plugin.

§Range requests support

Assets controller automatically supports part of RFC 7233 which defines how range requests and partial responses works. The Assets controller will delivery a 206 Partial Content if a satisfiable Range header is present in the request. It will also returns a Accept-Ranges: bytes for all assets delivery.

Note: Besides the fact that some parsing is done to better handle multiple ranges, multipart/byteranges is not fully supported yet.

You can also return 206 Partial Content when delivering files without using the Assets controller:

§Scala version

def video(videoId: Long) = Action { implicit request =>
  val videoFile = getVideoFile(videoId)
  RangeResult.ofFile(videoFile, request.headers.get(RANGE), Some("video/mp4"))
}

§Java version

public Result video(Long videoId) {
    File videoFile = getVideoFile(videoId);
    return RangeResults.ofFile(videoFile);
}

Both examples will delivery just part of the video file, according to the requested range.

Next: Using CoffeeScript