Skip to content


Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Allows mounting of remote JSON & YAML to the Middleman data object.


With Rubygems

$ gem install middleman-data_source

With Bundler

gem 'middleman-data_source'

Then in your Middleman's config.rb

activate :data_source do |c|
  c.root  = ""
  c.files = [

And access them like any other data:

# source/index.html

= data.some.title

You can also specify data resources as a hash to map the name:

# config.rb (in data_source activation block)
c.files = {
  "/url/to/resource.json" => "my_resource"

# source/index.html
= data.my_resource

You can fetch your data in two ways:

  1. from the file system or web with Borrower
  2. with a rack app

Multiple instances of the extension are supported, so activate once for each data source you require.

With Borrower

Borrower provides a common interface for fetching files from the filesystem or through http, so it's configured the same either way. You'll specify your root, then the files you want loaded (with paths relative to root).

# config.rb

activate :data_source do |c|
  c.root = '/var/data'
  c.files = [
    'middleman.json' # will look for this file at /var/data/middleman.json
                     # and be accessible through data.middleman

activate :data_source do |c|
  c.root = ""
  c.files = [
    'middleman.json' # will look for the file at
                     # and be accessible through data.middleman

With a Rack App

We can also fetch data through a local rack app, so if you want to manipulate a database and mount the data you can. A simple example that just uses Rack::Static would look like this:

# config.rb

activate :data_source do |c|

  c.rack_app = do
                 use Rack::Static, urls: ["/"],
                                   root: "remote_data"
                 run lambda {|env| [404, {'Content-type' => 'text/plain'}, ['Not found']] }

  c.files = [
    'rack.json' # passes /rack.json to your rack app
                # and mounts the data to data.rack

Custom data types

By default we just look at the extension of a file to determine which decoder to use. By setting a source directly you can give it any data type you need. Each source is a hash with an alias, path, and type key. The alias and path would be the same as if you defined the source using files as { path => alias }, while type corresponds to the decoder you'd like to use.

There are default decoders for :yaml and :json, however you are free to override them or create your own types.

# config.rb
activate :data_source do |c|

  c.files = ['by_extension.ctype']

  c.sources = [
      alias: "foo_bar",
      path: "/foo/bar.ctype",
      type: :my_type

  c.decoders = {
    my_type: {
      extensions: ['.ctype'],
      decoder: ->(src) { CustomType.parse(src) }

In the above example, I can access w/ the file contents decoded by CustomType, because it's extension .ctype matches the one defined by the :my_type decoder. Similarly, is also run through CustomType because it's :type attribute is set to :my_type.

Creating a collection

Collections allow you to collection sources that have belong together in an array and have distinct urls. This would loosely follow the Rails index/show convension. For example, lets say we have an endpoint that tells us about some Game of Thrones characters, and then includes a single endpoint for each with expanded information:

# /got/index.json
  { "name": "Eddard Stark", "url": "/got/eddard-stark.json" },
  { "name": "Hodor", "url": "/got/hodor.json" }
# /got/eddard-stark.json
  "name": "Eddard Stark",
  "quote": "Winter is coming"

  "name": "Hodor",
  "quote": "Hodor!"

Then set up a collection to access them through Middleman. A collection requires 3 keys, an alias, path, and items. The alias & path act just like a source, except that data will be available at #{alias}.#{index}. Items should be an object that responds to #call and returns an array of sources when given the data from the collection index. A collection for our example API:

activate :data_source do |c|
  c.root = ''
  c.collection = {
    alias: 'got_chars',
    path: '/got/index.json',
    index: 'all',
    items: { |data| do |char|
          alias: char['name'].to_slug,
          path: char['url']

You'll see I've used a proc to map our index into sources. The information is then accessible via the data object:
# => ['Eddard Stark', 'Hodor']

# => Winter is coming

For index_name, you can also pass false to not generate the index data.

Adding Middleware

You may wish to alter the incoming data before it gets added to Middleman's data object, to do so you'll add middleware. Middleware should respond to #call and return your altered data. Here's an example where we add "1" to each item in an array:

# count.yaml
- 1
- 2
- 3
# config.rb
activate :data_source do |c|
  c.root = ""
  c.sources = [{
    alias: 'altered',
    path: 'count.yaml',
    middleware: { |data| { |i| i + 1 } }

# => [2, 3, 4]


$ rspec


If there is any thing you'd like to contribute or fix, please:

  • Fork the repo
  • Add tests for any new functionality
  • Make your changes
  • Verify all existing tests work properly
  • Make a pull request


The middleman-data_source gem is distributed under the MIT License.


mount remote JSON & YAML to the Middleman data object.







No packages published