An opinionated library with simple, but unmissable additions to Backbone, Backbone.Paginator, backbone.stickit and Marionette.

At Code Yellow, we’ve been relying on these additions for over two years.

$ npm install backbone-crux --save

This section describes what functionality is added to the Backbone.Model.

Extended parse to add a new feature: ignore. If options.ignore = true, the parse function returns an empty object and effectively ignores the server response. This can be useful when you use patch where you simply want to set an attribute and not let the server result influence other attributes. If you supply an array, those keys will be omitted from the response before parsing.

isEmpty checks whether a model is still in the original, default, state, and thuss practically empty. An example:

import { Model } from 'backbone-crux';

const Bar = Model.extend({
    defaults() {
        return {
            id: null,
            type: 'foo',
        }
    }
});

const bar = new Bar();
bar.isEmpty(); // true
bar.set('type', 'bar');
bar.isEmpty(); // false

toJSON is modified to recursively flatten everything. This means that if you nest a model in a model, and then call model.save(), toJSON is also called for the nested model.

toHuman will recursively flatten everything, like toJSON. Marionette.View automatically uses toHuman and adds it to your template.

fromHuman can be used in a view to parse data from a human readable format to a server readable format. By default it does the exact same as set, but you can override it in your model.

This section describes what functionality is added to the Backbone.Collection.

The collection is always extended from backbone.paginator, and automatically parses API responses to fit into the collection.

The following format is expected from the API:

{
    "data": [{}, {}],
    "totalRecords": 2,
}

totalRecords should contain all records that were found (so not only the records that are in data!).

The collection has a model, called attributes in it. This model represents the data that will be added to each request.

import { Collection } from 'backbone-crux';

const collection = new Collection(null, {
    uri: 'api/example'
    attributes: {
        foo: 'bar',
    },
});

collection.fetch();

This will result in a fetch to api/example?foo=bar. Because the attributes are a model, it is very easy to listen to state changes and e.g. update a filter.

To add an attribute, you can do: collection.attributes.set('lorem', 'ipsum'). All new fetches will then use api/example?foo=bar&lorem=ipsum.

By default, a fetch is done with data from the attributes model. It is possible to override this, and enforce that only specific attributes are used:

import { Collection } from 'backbone-crux';

export default Collection.extend({
    fetchData() {
        return _.omit(this.attributes, 'foo');
    },
});

A XHR request for a fetch will be stored in collection.xhr. This allows you to e.g. easily abort a request when a view gets destroyed.

Marionette has the nifty @ui syntax to refer to html elements, but stickit does not. With backbone-crux, the @ui syntax is automatically enabled for stickit bindings.

Example:

export default Marionette.ItemView.extend({
    ui: {
        title: '._title',
    },
    onRender() {
        this.stickit();
    },
    bindings: {
        '@ui.title': 'title',
    },
});

For every XHR request a model or collection does, a couple of events will be fired. The name of these events has the method in it (create, read, update, or delete).

The event before:<method> is fired just before the XHR request takes place. After the XHR request, after:<method> will be fired. Depending on the outcome of the request, after:<method>:success and after:<method>:error will be fired.

An example of how you can use this:

export default Marionette.ItemView.extend({
    modelEvents: {
        'after:read:success': 'onAfterReadSuccess',
    },
    onAfterReadSuccess() {
        console.log('Look ma, I’m done fetching!');
    }
});

Using external plugins can be quite a pain with Marionette. The onRender method can get very messy fast. Also, forgetting to destroy plugins can lead to small memory leaks. Marionette plugins aims to fix these problems. An example:

export default Marionette.ItemView.extend({
    ui: {
        input: '._input',
    },
    plugins: {
        mask: {
            bind() {
                this.ui.input.mask('9999-99');
            },
            unbind() {
                this.ui.input.unmask();
            },
        },
    },
});

bind is called when rendering a view. unbind is called when a view gets destroyed.

You can also define plugins as a function:

export default Marionette.ItemView.extend({
    ui: {
        input: '._input',
    },
    plugins() {
        return {
            mask: {
                bind() {
                    this.ui.input.mask('9999-99');
                },
                unbind() {
                    this.ui.input.unmask();
                },
            },
        };
    },
});

To run the tests (and also lint the source files), run:

npm test

Distributed under ISC license.