Exposing Backbone model and collection methods to templates.
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.



Backbone.Marionette.Export is a plugin for Backbone, and specifically targeted at Marionette. It makes the methods of models and collections available to templates.

The name of the plugin has turned out to be a bit of a misnomer, though. Backbone.Marionette.Export does not depend on Marionette and works great without it. In fact, Backbone.Marionette.Export is just as useful in projects based on plain Backbone.

If you are a happy user of this project already, you can support its development by donating to it. You absolutely don't have to, of course, but perhaps it is something you might actually want to do.


Include backbone.marionette.export.js after Backbone, and after Marionette (if you use Marionette).

The stable version of Backbone.Marionette.Export is available in the dist directory (dev, prod). If you use Bower, fetch the files with bower install backbone.marionette.export. With npm, it is npm install backbone.marionette.export.

Module systems (AMD, CJS)

Because Marionette is optional, it is not defined as a dependency in Backbone.Marionette.Export. Your own code must make sure that Marionette is loaded first. With RequireJS, for instance, use a shim to express the dependency:

    shim: {
        "backbone.marionette.export": ["marionette"]

Backbone.Marionette.Export does not export a meaningful value. It solely lives in the Backbone namespace.

Underscore vs. Lodash

If you need to handle deeply nested structures recursively, swap out Underscore for a compatible Lo-dash build with _.cloneDeep support. See below.

Use case

Out of the box, templates handled by Marionette views have access to all attributes of a model, and to the array of models represented by a collection. But templates can't use the output of methods.

To work around that, you could override the toJSON method in a model or collection, but that creates its own set of problems. Specifically, anything you change in toJSON will also get written back to the server on save.

Backbone.Marionette.Export does not cause any such side effects. After dropping it in, this is what you can do:

  • Select which methods of a model, or of a collection, provide their output to templates.
  • Modify the data before it is handed to a template, by implementing an onExport handler.
  • Manipulate the model or collection state itself before it is passed to a template, using an onBeforeExport handler.
  • Clean up by implementing an onAfterExport handler.

Usage and examples

The basics

Here is how it works, in its simplest form. The examples below use the Marionette.ItemView type of Marionette 2. If you use Marionette 3, just substitute the successor, Marionette.View, for it.

<script id="item-view-template" type="text/x-handlebars-template">
    <p>Model method returns: {{foo}} </p>


var Model = Backbone.Model.extend ({
    exportable: "foo",                 // <-- this is the one line you have to add
    foo: function () { 
        return "some calculated result of calling foo"; 

var view = new Marionette.ItemView ({
    model: new Model(),
    template: "#item-view-template"


In the model definition, you declare which methods are available to a template. Just provide the method name to exportable, or an array of them. For method names, both "foo" and "this.foo" are acceptable.

That works fine for simple method signatures. But what about methods which take arguments?

<script id="item-view-template" type="text/x-handlebars-template">
    <p>Model method returns: {{foo}} </p>


var Model = Backbone.Model.extend ({     

    foo: function ( arg ) { 
        return "some calculated result of calling foo with " + arg; 

    onExport: function ( modelHash ) {
        modelHash.foo = this.foo( someArg );
        return modelHash;


var view = new Marionette.ItemView ({
    model: new Model(),
    template: "#item-view-template"


In this scenario, there is no need to declare the method as exportable. In fact, you can't: the method takes arguments, so it can't be called automatically. Instead, modify the exported model data in an onExport handler. You can do pretty much anything there. Just remember to return the data at the end.

For a collection, the process is the same as for a model.

<script id="item-view-template" type="text/x-handlebars-template">
    <p>Collection method 'foo' returns: {{items.foo}} </p>


var Collection = Backbone.Collection.extend ({
    exportable: "foo",          
    foo: function () { return "some calculated result of calling foo"; }

var view = new Marionette.ItemView ({
    collection: new Collection(),
    template: "#item-view-template"


The collection data is provided to the template as it always is: in an items array. That is the standard behaviour of a Marionette ItemView (or View in Marionette 3) and unrelated to Backbone.Marionette.Export.

Besides being an array, items is an object like any other. Arbitrary properties can be added to it. And that is exactly what happened to the exported method, foo.


There is an interactive demo you can play around with. The demo is kept simple, and is a good way to explore the features of Backbone.Cycle. Check it out at JSBin or Codepen.


Baseline support

Now, suppose a collection is passed to a template. What if it is made up of models which, in turn, have methods marked for export to the template?

<script id="item-view-template" type="text/x-handlebars-template">
    <p>Method 'foo' of the first item in the collection returns: {{items.first.foo}} </p>
    <p>Method 'foo' of the last item in the collection returns: {{items.last.foo}} </p>


var Model = Backbone.Model.extend ({
    exportable: "foo",          
    foo: function () { return "my cid is " + this.cid; }

var Collection = Backbone.Collection.extend ({
    exportable: [ "first", "last" ]           // <-- you can use it for built-in methods, too

var view = new Marionette.ItemView ({
    collection: new Collection( [ new Model(), new Model() ] ),
    template: "#item-view-template"


The message here is that the plugin handles recursion for you, no matter of what kind. You can have a collection method return another, nested collection, which in turn holds models with methods marked for export. It all gets exported, just as you would expect, without any additional measures on your part.

Enhanced recursion support with Lo-dash and _.cloneDeep

In the default Backbone setup, recursion works fine as long as Backbone objects are nested directly within one another. If the chain is broken by other, non-Backbone objects in between, recursion stops. Consider these examples:

  • A hypothetical model.getInnerModel.getInnerCollection.getYetAnotherModel will behave as expected.
  • By contrast, model.getObjectLiteral.someProperty.innerBackboneModelHere won't trigger a call to export on the inner model.

This limitation is imposed by the underlying utility library, Underscore. Underscore doesn't do deep cloning. Handling deeply nested structures with Underscore can be full of surprises, mostly unpleasant ones.

The good news is that deep cloning, and support for deep recursion, is easy to add. Just swap out Underscore for a fully compatible build of Lo-dash.

On the Lo-dash site, there are a number of builds to choose from, including one designed to replace Underscore with 100% compatibility, but it lacks support for deep cloning. You can add it yourself, though, with a few commands in a terminal.

  • Make sure you have Node.js and npm working.
  • Install Lo-dash globally with npm install -g lodash, and the CLI with npm install -g lodash-cli.
  • Change into the directory where you want to put your Underscore replacement.
  • Create the library with lodash underscore plus=clone,cloneDeep.
  • Replace the Underscore script tag on your pages with one loading the new library.

With the Lo-dash build in place, deeply nested structures are no longer a problem. In the second example above, model.getObjectLiteral.someProperty.innerBackboneModelHere indeed triggers a call to export on the inner model, as one would expect.

Maximum recursion depth

Recursion can generate huge data structures in some cases, so it makes sense to impose a limit. By default, there won't be more than four recursive calls to export for a given top-level object. That should be more than enough for almost any template requirement.

You can change the limit globally in your project by setting Backbone.Model.prototype.export.global.maxHops or Backbone.Collection.prototype.export.global.maxHops to the desired recursion depth. (Changing one of them is enough. Model and Collection prototypes share the same config object.)

Circular dependencies between your models and collections are contained by the recursion limit, too.

Complex data wrangling

In case you have to change the model state, or collection state, before the model is handed over to a template, you can do whatever you need to do with onBeforeExport. Implement it, and it will be called before the data export kicks in.

onBeforeExport is fully separated from the actual export mechanism. You can even manipulate the exportable property itself and set it dynamically each time template data is requested.

For any clean-up operations, implement onAfterExport. When it is called, the data for templates is already finalized.

Strict mode

You can make Backbone.Marionette.Export considerably more pedantic, having it throw more exceptions, if you put it in strict mode.

What strict mode does

Strict mode is off by default. When enabled, Backbone.Marionette.Export will throw an exception

  • when a method or a collection property is declared as exportable, but does not exist
  • when a model property is declared as exportable

Enabling strict mode

You enable strict mode globally in your project by setting Backbone.Model.prototype.export.global.strict or Backbone.Collection.prototype.export.global.strict to true. (Changing one of them is enough. Model and Collection prototypes share the same config object.)

When to use strict mode

By default, if you try to export a non-existent method or property, it will just get ignored. Turning on strict mode may be helpful for catching typos, and perhaps even logical errors.

Also, by default, you can declare model properties as exportable. But model properties, if they are relevant to a template, should really be model attributes. Those get passed to templates out of the box.

So in the vast majority of cases, you won't have the need to say exportable: "someProperty". By turning on strict mode, you'll get an error thrown at you for trying. That may help to catch accidental assignments.

Collections, by contrast, don't have attributes, and they don't provide a native way to have a property show up in a template. So here, exportable: "someProperty" makes sense, and indeed it will work just fine – even in strict mode.

For which Marionette view types does it work?

All of them.

But I don't use Marionette!

The export functionality in models and collections is wired up with Marionette views and works out of the box. But you can use it in your own view types, too.

When you grab model or collection data during render(), call model.export() or collection.export() instead of their toJSON counterparts. That's all there is to it.

(If you have to support old browsers which are based on ES3, don't call export as shown above, in dot notation. Because export is a reserved keyword, using it as a property name may throw an error in ES3-compliant browsers, and in fact it does in IE8 and Android 2.x. Use the bracket notation instead: model["export"]() or collection["export"]().)


Backbone is the only dependency. It makes most sense to use the enhancements with Marionette views, though. If present, Marionette is modified to respond to the export() functionality, and Marionette views pass the exported properties to the templates automatically.

Build process and tests

If you'd like to fix, customize or otherwise improve the project: here are your tools.


npm sets up the environment for you.

  • The only thing you've got to have on your machine (besides Git) is Node.js. Download the installer here.
  • Clone the project and open a command prompt in the project directory.
  • Run the setup with npm run setup.
  • Make sure the Grunt CLI is installed as a global Node module. If not, or if you are not sure, run npm install -g grunt-cli from the command prompt.

Your test and build environment is ready now. If you want to test against specific versions of Backbone or Marionette, edit bower.json first.

Running tests, creating a new build

The test tool chain: Grunt (task runner), Karma (test runner), Mocha (test framework), Chai (assertion library), Sinon (mocking framework). The good news: you don't need to worry about any of this.

A handful of commands manage everything for you:

  • Run the tests in a terminal with grunt test.
  • Run the tests in a browser interactively, live-reloading the page when the source or the tests change: grunt interactive.
  • If the live reload bothers you, you can also run the tests in a browser without it: grunt webtest.
  • Run the linter only with grunt lint or grunt hint. (The linter is part of grunt test as well.)
  • Build the dist files (also running tests and linter) with grunt build, or just grunt.
  • Build continuously on every save with grunt ci.
  • Change the version number throughout the project with grunt setver --to=1.2.3. Or just increment the revision with grunt setver --inc. (Remember to rebuild the project with grunt afterwards.)
  • grunt getver will quickly tell you which version you are at.

Finally, if need be, you can set up a quick demo page to play with the code. First, edit the files in the demo directory. Then display demo/index.html, live-reloading your changes to the code or the page, with grunt demo. Libraries needed for the demo/playground should go into the Bower dev dependencies – in the project-wide bower.json – or else be managed by the dedicated bower.json in the demo directory.

The grunt interactive and grunt demo commands spin up a web server, opening up the whole project to access via http. So please be aware of the security implications. You can restrict that access to localhost in Gruntfile.js if you just use browsers on your machine.

Changing the tool chain configuration

In case anything about the test and build process needs to be changed, have a look at the following config files:

  • karma.conf.js (changes to dependencies, additional test frameworks)
  • Gruntfile.js (changes to the whole process)
  • web-mocha/_index.html (changes to dependencies, additional test frameworks)

New test files in the spec directory are picked up automatically, no need to edit the configuration for that.

Facilitating development

To my own surprise, a kind soul wanted to donate to one of my projects, but there hadn't been a link. Now there is.

Please don't feel obliged in the slightest. The license here is MIT, and so it's free. That said, if you do want to support the maintenance and development of this component, or any of my other open-source projects, I am thankful for your contribution.

Naturally, these things don't pay for themselves – not even remotely. The components I write aim to be well tested, performant, and reliable. These qualities may not seem particularly fascinating, but I put a lot of emphasis on them because they make all the difference in production. They are also rather costly to maintain, time-wise.

That's why donations are welcome, and be it as nod of appreciation to keep spirits up. Thank you!

Donate with Paypal

Release Notes


  • Removed the separate AMD/Node builds in dist/amd. Module systems and browser globals are now supported by the same file, dist/backbone.marionette.export.js (or .min.js)


  • Added support for Marionette 3
  • Version is exposed in Backbone.Model.prototype.export.global.version and Backbone.Collection.prototype.export.global.version
  • AMD demo allows testing r.js output


  • Updated Backbone dependency
  • Updated build environment


  • Updated Backbone dependency


  • Updated component dependencies
  • Updated build environment


  • Made the component backward compatible with ES3 (fix for IE8)
  • Fixed strict mode in AMD build


  • Removed Marionette as a hard dependency in AMD/CJS build. Important: It is now up to the client code to make sure that Marionette is loaded first.
  • Made available as an npm install


Bumping the major version is necessary because of changes to the public API, even though they are minor.

  • No longer throwing exceptions by default when declaring non-existent methods or properties as exportable
  • No longer throwing an exception by default when declaring model properties as exportable
  • Added strict mode
  • Moved maxHops into global config object

v1.0.1 - 1.0.7

  • Minor bug fixes
  • Added demo code, documentation changes
  • New build environment


  • Initial public release



Copyright (c) 2014-2017 Michael Heim.