Awesome
<img src="https://raw.githubusercontent.com/ErvalhouS/APIcasso/master/APIcasso.png" width="300" />
Create APIs in a fast and dynamic way, without the need to develop everything from scratch. You just need to create your models and let APIcasso do the rest for you. It is the perfect candidate to make your project development go faster or for legacy Rails projects that do not have an API.
If you think it through, JSON API development can get boring and time consuming. Every time you use almost the same route structure, pointing to the same controller actions, with the same ordering, filtering and pagination features.
APIcasso is intended to be used to speed-up development, acting as a full-fledged CRUD JSON API into all your models. It is a route-based abstraction that lets you create, read, list, update or delete any ActiveRecord
object in your application. This makes it possible to make CRUD-only applications just by creating functional Rails' models. Access to your application's resources is managed by a .scope
JSON object per API key. It uses that permission scope to restrict and extend access.
You can make your own API with only 4 steps:
Step 1
Create your models
Step 2
Insert APIcasso engine into your routes and run the installation command
Step 3
Step 4
Profit! :crown: Consume your REST API
Installation
Add this line to your application's Gemfile
:
gem 'apicasso'
And then execute this to generate the required migrations:
$ bundle install && rails g apicasso:install
Requirements
- PostgreSQL with JSON columns support
- Ruby 2.3+
- Rails 5+
Usage
APIcasso is meant to be used as an engine, which means you don't need to configure any route or controller to build a working CRUD API. Sometimes you also need some customized controller actions or even a specific logic to access some of your application's resources. In that case you will use Apicasso::CrudController
class to easily build only your own logic around the API abstraction.
Mounting engine into config/routes.rb
After installing it, you can mount a full-fledged CRUD JSON API just by attaching into some route. Usually you will have it under a scoped route like /api/v1
or a subdomain. You can do that by adding this into your config/routes.rb
:
# To mount your APIcasso routes under the path scope `/api/v1`
mount Apicasso::Engine, at: "/api/v1"
# or, if you prefer subdomain scope isolation
constraints subdomain: 'apiv1' do
mount Apicasso::Engine, at: "/"
end
Your API will reflect very similarly a resources :resource
statement with the following routes:
get '/:resource/' # Index action, listing a `:resource` collection from your application
post '/:resource/' # Create action for one `:resource` from your application
get '/:resource/:id' # Show action for one `:resource` from your application
patch '/:resource/:id' # Update action for one `:resource` from your application
delete '/:resource/:id' # Destroy action for one `:resource` from your application
get '/:resource/:id/:nested/' # Index action, listing a collection of a `:nested` relation from one of your application's `:resource`
options '/:resource/' # A schema dump for the required `:resource`
options '/:resource/:id/:nested/' # A schema dump for the required `:nested` relation from one of your application's `:resource`
This means all your application's models will be exposed as :resource
and it's relations will be exposed as :nested
. It will enable you to CRUD and get schema metadata from your records.
Extending base API actions
When your application needs some kind of custom interaction that is not covered by APIcasso's CRUD approach you can make your own actions using our base classes and objects to go straight into your logic. If you have built the APIcasso's engine into a route it is important that your custom action takes precedence over the gem's ones. To do that you need to declare your custom route before the engine on you config/routes.rb
match '/:resource/:id/a-custom-action' => 'custom#not_a_crud', via: :get
mount Apicasso::Engine, at: "/api/v1"
And in your app/controllers/custom_controller.rb
you would have something like:
class CustomController < Apicasso::CrudController
def not_a_crud
render json: @object.some_operation
end
end
This way you enjoy all our object finder, authorization and authentication features, making your job more straight into your business logic.
CORS
APIcasso comes with a permissive CORS configuration out of the box. But you can make your own by editting the config/initializers/apicasso.rb
file, which is created at the installation proccess. The file comes with some descriptive comments and all configuration is based on Rack CORS options.
Authentication
But exposing my models to the internet is permissive as hell! Haven't you thought about security?
Sure! The APIcasso suite is exposing your application using authentication through Authorization: Token
HTTP header authentication. The API key objects are manageable through the Apicasso::Key
model, which gets setup at install. When a new key is created a .token
is generated using an Universally Unique Identifier(RFC 4122). An authenticated request looks like this:
curl -X GET \
https://apixample.com/v1/your_app_resource \
-H 'authorization: Token token=cda4e9f633c123ef9ddce5e6564292b3'
Each Apicasso::Key
object has a token attribute, which is used on this header to authorize access. For now, there is no plans for a login/JWT logic, you should implement this in your project's scope.
Authorization
Your Models are then exposed based on each Apicasso::Key.scope
definition, which is a way to configure how much of your application each key can access. I.E.:
Apicasso::Key.create(scope:
{ manage:
{ order: true, user: { account_id: 1 } },
read:
{ account: { manager_id: 1 } }
})
The key from this example will have full access to all orders and to users with
account_id == 1
. It will have also read-only access to accounts withid == 1
.
A scope configured like this translates directly into which kind of access each key has on all of your application's models. This kind of authorization is why one of the dependencies for this gem is CanCanCan, which abstracts the scope field into your API access control.
You can have two kind of access control:
true
- This will mean the key will have the declared clearance on ALL of this model's recordsHash
- This will build a condition to what records this key have access to. A scope as{ read: [{ account: { manager_id: 1 } }] }
will have read access into accounts withmanager_id == 1
This saves you the trouble of having to setup every controller for each model. And even if your application really needs it, just make your controllers inherit from Apicasso::CrudController
extending it and enabling the use of @object
and @resource
variables to access what is being resquested.
Features on index actions
The index actions present in the gem are already equipped with pagination, ordering, grouping, fieldset selection and filtering. This will save you a lot of trouble, adding some best-practices conventions into your application's API.
Sort
You can sort a collection query by using a URL parameter with field names preffixed with +
or -
to configure custom ordering per request.
To order a collection with ascending updated_at
and descending name
you can add the sort
parameter with those fields as options, indicating which kind of ordination you want to give to each one:
?sort=+updated_at,-name
Filtering/Search
APIcasso has ransack's search matchers on it's index actions. This means you can dynamically build search queries with any of your resource's fields, this will be done by using a ?q
parameter which groups all your filtering options on your requests. If you wanted to search all your records and return only the ones with full_name
starting with Picasso
your query would look something like this:
?q[full_name_start]=Picasso
To build complex search queries you can chain many parameter options or check ransack's wiki on how to adapt this feature into your project's needs.
Pagination
Automatic pagination is done in index actions, with the adittion of some metadata to help on the data consumption. You can pass page and per page parameters to build pagination options into your needs. And on requests that you need unpaginated collections, just pass a lower than zero per_page
. Example of a pagination query string:
?page=2&per_page=12
Your colletion will be build inside a JSON along with some metadata about it. The response structure is:
{ entries: [{Record1}, {Record2}, {Record3} ... {Record12}],
total: 1234,
total_pages: 102,
last_page: false,
previous_page: localhost:3000/my_records?page=1&per_page=12,
next_page: localhost:3000/my_records?page=3&per_page=12,
out_of_bounds: false,
offset: 12 }
Fieldset selecting
Sometimes your data can grow large in some tables and you need to consumed only a limited set of data on a given frontend application. To avoid large requests and filtering a lot of unused data with JS you can restrict which fields you need on your API's reponse. This is done adding a ?select
parameter. Just pass the field names you desire splitted by ,
Let's say you are building a user list with their name, e-mails and phones, to get only those fields your URL query would look something like:
?select=name,email,phone
This will change the response to return only the requested attributes. You need to observe that your business logic may require some fields for a valid response to be returned. This method can be used both on index and show actions
Including relations or methods on response
If there is any method or relation that you want to be inserted on the payload, you just need to pass them as a part of the URL query like this:
?include=pictures,suggestions
This will insert the contents of .pictures
and .suggestions
on the payload, along with the records' data. This means you can populate the payload both with methods or relations contents. This method can be used both on index and show actions
Grouping operations
If you need to make grouping calculations, like:
- Counting of all records, or by one optional field presence
- Maximum value of one field
- Minimum value of one field
- Average value of one field
- Value sum of one field
Grouping is done by the combination of 3 parameters
?group[by]=a_field&group[calculate]=count&group[fields]=another_field
Each of those attributes on the ?group
parameter represent an option of the query being made.
group[by]
- Represents which field will be the key for the grouping behaviorgroup[calculate]
- Which calculation will be sent in the response. Options are:count
,maximum
,minimum
,average
,sum
group[field]
- Represents which field will be the base for the response calculation.
Contributing
Bug reports and pull requests are welcome on GitHub at https://github.com/ErvalhouS/APIcasso. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct. To find good places to start contributing, try looking into our issue list and our Codeclimate profile, or if you want to participate actively on what the core team is working on checkout our todo list:
TODO
- Add support to other databases
- Add gem options like: Token rotation, Alternative authentication methods
- Refine and document auto-documentation feature
- Rate limiting
Code of conduct
Everyone interacting in the APIcasso project’s codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.
License
The gem is available as open source under the terms of the MIT License.