HyperResource is a Ruby client library for hypermedia web services.
HyperResource makes using a hypermedia API feel like calling plain old methods on plain old objects.
It is usable with no configuration other than API root endpoint, but also allows incoming data types to be extended with Ruby code.
HyperResource supports the HAL+JSON hypermedia format, with support for Siren and other hypermedia formats planned.
For more insight into HyperResource's goals and design, read the paper!
Hypermedia APIs return a list of hyperlinks with each response. These links, each of which has a relation name or "rel", represent everything you can do to, with, or from the given response. They are URLs which can take arguments. The consumer of the hypermedia API uses these links to access the API's entire functionality.
A primary advantage to hypermedia APIs is that a client library can write itself based on the hyperlinks coming back with each response. This removes both the chore of writing a custom client library in the first place, and also the process of pushing client updates to your users.
An automatically-generated library can and should feel as comfortable as a custom client library. Hypermedia brings many promises, but this is one we can deliver on.
If you're an API user, HyperResource will help you consume a hypermedia API with short, direct, elegant code. If you're an API designer, HyperResource is a great default client, or a smart starting point for a rich SDK.
Link-driven APIs are the future, and proper tooling can make it The Jetsons instead of The Road Warrior.
Nothing special is required, just: gem install hyperresource
HyperResource works on Ruby 1.8.7 to present, and JRuby in 1.8 mode or above.
HyperResource uses the uri_template and Faraday gems.
Set up API connection:
api = HyperResource.new(root: 'https://api.example.com',
headers: {'Accept' => 'application/vnd.example.com.v1+json'},
auth: {basic: ['username', 'password']})
# => #<HyperResource:0xABCD1234 @root="https://api.example.com" @href="" @namespace=nil ... >Now we can get the API's root resource, the gateway to everything else on the API.
api.get
# => #<HyperResource:0xABCD1234 @root="https://api.example.com" @href="" @namespace=nil ... >What'd we get back?
api.body
# => { 'message' => 'Welcome to the Example.com API',
# 'version' => 1,
# '_links' => {
# 'curies' => [{
# 'name' => 'example',
# 'templated' => true,
# 'href' => 'https://api.example.com/rels/{rel}'
# }],
# 'self' => {'href' => '/'},
# 'example:users' => {'href' => '/users{?email,last_name}', 'templated' => true},
# 'example:forums' => {'href' => '/forums{?title}', 'templated' => true}
# }
# }Lovely. Let's find a user by their email.
jdoe_user = api.users(email: "[email protected]").first
# => #<HyperResource:0x12312312 ...>HyperResource has performed some behind-the-scenes expansions here.
First, the example:users link was
added as a method on the api object at the time the resource was
loaded with api.get. And since the link rel has a
CURIE prefix,
a method without the prefix, users, was created at the same time.
Then, calling first on the users link
followed the link and loaded it automatically.
Finally, calling first on the resource containing one set of
embedded objects -- like this one -- delegates the method to
.objects.first, which returns the first object in the resource.
Here are some equivalent expressions to the above. HyperResource offers a very short, expressive syntax as its primary interface, but you can always fall back to explicit syntax if you like or need to.
api.users(email: "[email protected]").first
api.get.users(email: "[email protected]").first
api.get.links.users(email: "[email protected]").first
api.get.links['users'].where(email: "[email protected]").first
api.get.links['users'].where(email: "[email protected]").get.first
api.get.links['users'].where(email: "[email protected]").get.objects.first[1][0]
If an API is returning data type information as part of the response, then we can assign those data types to ruby classes so that they can be extended.
For example, in our hypothetical Example API above, a user object is returned with a custom media type bearing a 'type=User' modifier. We will extend the User class with a few convenience methods.
class ExampleAPI < HyperResource
self.root = 'https://api.example.com'
self.headers = {'Accept' => 'application/vnd.example.com.v1+json'}
self.auth = {basic: ['username', 'password']}
class User < ExampleAPI
def full_name
first_name + ' ' + last_name
end
end
end
api = ExampleApi.new
user = api.users.where(email: '[email protected]').first
# => #<ExampleApi::User:0xffffffff ...>
user.full_name
# => "John Doe"Don't worry if your API uses some other method to indicate resource data
type; you can override the .get_data_type_from_response method and
implement your own logic.
HyperResource raises a HyperResource::ClientError on 4xx responses,
and HyperResource::ServerError on 5xx responses. Catch one or both
(HyperResource::ResponseError). The exceptions contain as much of
cause (internal exception which led to this one), response
(Faraday::Response object), and body (the decoded response
as a Hash) as is possible at the time.
Many thanks to the people who've sent pull requests and improved this code:
- Jean-Charles d'Anthenaise Sisk (@jasisk)
- Ian Asaff (@montague)
- Julien Blanchard (@julienXX)
- Andrew Rollins (@andrew311)
- Étienne Barrié (@etiennebarrie)
- James Martelletti (@jmartelletti)
- Frank Macreery (@fancyremarker)
- Chris Rollock (@Sastafer)
Copyright 2013-2014 Pete Gamache, [email protected].
Released under the MIT License. See LICENSE.txt.
If you got this far, you should probably follow me on Twitter. @gamache
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent