No release in over a year
IntrospectiveGrape provides handling for deeply nested relations according to the models' `accepts_nested_attributes_for` declarations, generating all the necessary boilerplate for flexible and consistent bulk endpoints on plural associations, and building nested routes for the same.
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Project Readme

IntrospectiveGrape

Gem Version Coverage Status Tests Lints

IntrospectiveGrape is a rails plugin for DRYing up Grape APIs by laying out simple (or, if you like, very complex) RESTful defaults based on the introspection of your database by SchemaPlusValidations.

IntrospectiveGrape provides handling for deeply nested relations according to the models' accepts_nested_attributes_for declarations, generating all the necessary boilerplate for flexible and consistent bulk endpoints on plural associations, and building nested routes for the same.

It relies on Kaminari for pagination and Pundit for authorization, both of which are semi-optional.

To facilitate idiomatic ruby and javascript, respectively, it also makes it easy to snakecase incoming parameters and camelize outputs, all the way through to your swagger docs.

Documentation

In your Gemfile:

gem 'introspective_grape'

And bundle install.

Grape Configuration

Camelization

To camelcase your JSON outputs you'll need to use IntrospectiveGrape's formatter in your API:

require 'introspective_grape/camel_snake'
class MyAPI < Grape::API
  formatter :json, IntrospectiveGrape::Formatter::CamelJson
end

This also monkey patches Grape::Swagger to camelize your API's self-documentation, while snakecasing parameters passed to your API.

To include this behavior in your test coverage you need to either access the API's params hash or you can format the response body to JSON.parse(response.body).with_snake_keys in a helper method with the using CamelSnakeKeys refinement.

If you need to disable all camel-snake transliteration set IntrospectiveGrape.config.camelize_parameters = false down in config/initializers and do not require or formatter those patches.

Authentication and authorization

Authentication and authorization are presently enforced on every endpoint. If you have named the authentication helper method in Grape something other than "authenticate!" or "authorize!" you can set it with:

IntrospectiveGrape::API.authentication_method = "whatever!"

Pundit authorization is invoked against index?, show?, update?, create?, and destroy? methods with the model instance in question (or a new instance in the case of index).

The joke goes that you may find you need to allow an unauthenticated user to attempt a log in, which can be handled with something like:

  def authorize!
    unauthorized! unless current_user || login_request?
  end

  def login_request?
    # is it the session login endpoint?
    self.method_name.start_with?('POST') && self.namespace == '/login'
  end

Generating End Points for Models

IntrospectiveGrape's parameterization of a model begins with the restful declaration, which expects an array of parameters that can be passed on to the model for update and create actions, including its nested models.

The simplest app/api/v1/my_model_api.rb with the broadest functionality would look like:

class MyModelAPI < IntrospectiveGrape::API
  filter_on :all

  restful MyModel, [:strong, :param, :fields, :and, { nested_model_attributes: [:nested,:fields, :_destroy] }]

  class <NestedModel>Entity < Grape::Entity
    expose :id, :attribute
  end

  class MyModelEntity < Grape::Entity
    expose :id, :attribute1, :attribute2
    expose :nested, using: <NestedModel>Entity>
  end
end

This would set up all the basic RESTFUL actions with nested routes for the associated model and its association, providing a good deal of flexibility for API consumers out of the box.

IntrospectiveGrape looks in the MyModelAPI class for grape-entity definitions. If you prefer to define your entities elsewhere you could inherit them here instead.

NOTE: Nested entities must be defined before their parents, inside-out, or you'll run into loading errors.

Customizing End Points

Many simple customizations are available to carve out behaviors from the expansive defaults:

class MyModelAPI < IntrospectiveGrape::API
  skip_presence_validations :attribute_with_generated_default_value

  exclude_actions Model, <:index,:show,:create,:update,:destroy>
  default_includes Model, <associations for eager loading>

  include_actions NestedModel, <:index,:show,:create,:update,:destroy>
  default_includes NestedModel, <associations for eager loading>

  paginate per_page 25, offset: 0, max_per_page: false

  filter_on :param

  restful MyModel, [:strong, :param, :fields, :and, { nested_model_attributes: [:nested,:fields, :_destroy] }] do
    # Here you can add additional end points to the model's namespace, or customize one you've recently excluded so you can implement a caching strategy.
  end

  class <NestedModel>Entity < Grape::Entity
    expose :id, :attribute
  end

  class <MyModel>Entity < Grape::Entity
    expose :id, :attribute
    expose :nested, using: <NestedModel>Entity>
  end
end

Please note, again, that the nested Grape::Entity is declared before its parent.

Skipping a Presence Validation for a Required Field

If a model has, say, a procedurally generated default for a not-null field in the database skip_presence_validations will make IntrospectiveGrape declare the parameter optional rather than required.

Excluding Endpoints

By default any association included in the strong params argument will have all RESTful (:index,:show,:create,:update, :destroy) endpoints defined. These can be excluded (or conversely included) with the exclude_actions (or include_actions) declarations in the API class.

  include_actions NestedModel, <:index,:show,:create,:update,:destroy>

You can also include or exclude :all or :none on the association as shorthand.

Eager Loading

Declaring default_includes on an activerecord class will tell IntrospectiveGrape which associations to eager load when fetching a collection or instance.

If you can diagnose an N+1 problem then this ought to fix it.

You can also eager load nested associations, if necessary:

  default_includes NestedModel, <associations for eager loading>

My suggestion would be to ignore this feature and rely instead on ArLazyPreload.

Pagination

The index action by default will not be paginated. Simply declaring paginate before the restful declaration will enable Kaminari pagination on the index results using a default 25 results per page with an offset of 0. You can pass Kaminari's options to the paginate declaration, per_page, max_per_page, etc.

This will likewise add Kaminari's "X-" pagination headers to the response.

Validating Virtual Attributes and Overriding Grape Validations

Sometimes you need to break out of IntrospectiveGrape to handle behavior that is outside its scope, sometimes you just need to add an override or virtual attribute on the represented model to handled that behavior. Consider the latter first!

To define a Grape param type for a virtual attribute or override the defaut param type from database introspection, define a class method in the model with the param types for the attributes specified in a hash, e.g.:

   def self.grape_param_types
    { "<attribute name 1>" => String,
      "<attribute name 2>" => Integer,
      "<attribute name 3>" => Grape::API::Boolean }
   end

To add additional validations on API inputs you can define a hash of hashes in the model in a class method ("grape_validations") that will be applied to that field's param declaration:

  def self.grape_validations
    { field1: { values: %w(red blue green) },
      field2: { json_array: true },
      field3: { regexp: /\w+/ }
  end

Many bespoke behaviors can be relegated to the model via virtual attributes in this way.

Validating JSON Parameters

IntrospectiveGrape provides the following custom grape validators for JSON string parameters:

json: true       # validates that the JSON string parses
json_array: true # validates that the JSON string parses and returns an Array
json_hash: true  # validates that the JSON string parses and returns a Hash

Filtering and Searching on GET :index

Simple filters on field values (and start and end values for timestamps) can be added with the filter_on declaration. Declaring filter_on :all will add filters for every attribute of the model.

class MyModelAPI < IntrospectiveGrape::API
  filter_on :my_attribute, :my_other_attribute
end

Multiple values can be queried for at once for attributes that end in "id" (i.e. conventional primary and foreign keys) by passing a comma separated list of IDs.

For timestamp attributes it will generate <name_of_timestamp>_start and <name_of_timestamp>_end range constraints.

There is also a special "filter" filter that accepts a JSON hash of attributes and values: this allows more complex filtering if one is familiar with ActiveRecord's query conventions.

Overriding Filter Queries

If, e.g., a field is some sort of complex composite rather than a simple field value you can override the default behavior (where(field: params[field])) by adding a query method on the model class:

class MyAPI < IntrospectiveGrape::API
  filter_on :my_composite_field
  restful MyModel, [my_composite_field]
end

class MyModel
  self << class
    def my_composite_field=(parameters)
      # parse the passed parameters in some way and return a query scope
    end
  end
end

Custom Filter Methods

To add a custom filter to the index action you can declare a method to be called against the model class with custom_filter. You can pass documentation and type constraints (it would default to String) and other Grape parameter options in a hash:

class MyAPI < IntrospectiveGrape::API
  custom_filter :my_filter, type: Boolean, description: "Filter on some scope"
end

class MyModel
  self << class
    def my_filter(filter=false)
      filter ? my_scope : where(nil)
    end
  end
end

Performance Tuning: IntrospectiveGrape defaults to reloading an object after any update!

By default the gem reloads the object instance before presenting the updated model, a lazy but effective workaround to updates that may not propagate in the working instance due to actions a user may take in hooks or some updates to has_many :through associations. We want to put up APIs with haste rather than digging our way out of tricky minutae that can be handled later as technical debt.

This behavior can be disabled by setting IntrospectiveGrape.config.skip_object_reload = true, when you have time for technical debt you can toggle it and work on fixing broken tests (you did take the time to write comprehensive test coverage, didn't you?).

It is presently only an application wide configuration setting, not having made much use of it myself, but would be trivial to make more atomistic.

Documenting Endpoints

If you wish to provide additional documentation for end points you can define self.<index,show,update,create,destroy>_documentation(name) class methods in the API class (or extend them from a documentation module, which would be preferable).

Grape Hooks - The Precedence of Declaration Matters

Grape only applies hooks in the order they were declared, so to hook into the default RESTful actions defined by IntrospectiveGrape you need to declare any hooks before the restful declaration, rather than inside its block, where the hook will only apply to your own subsequently declared endpoints.

Dependencies

Tool Description
Rails A web-application framework using a Model-View-Controller pattern
SchemaValidations Automatically defining validations based on the database schema
Grape An opinionated micro-framework for creating REST-like APIs in Ruby
GrapeEntity Adds Entity support to API frameworks, such as Grape.
GrapeSwagger Swagger docs.
GrapeKaminari Pagination.
Pundit Minimal authorization through OO design and pure Ruby classes