GraphQL Groups

Run group- and aggregation queries with graphql-ruby.

Installation

Add this line to your application's Gemfile and run bundle install.

gem 'graphql-groups'

$ bundle install

Usage

Suppose you want to get the number of authors, grouped by their age. Create a new group type by inheriting from GraphQL::Groups::GroupType:

class AuthorGroupType < GraphQL::Groups::Schema::GroupType
  scope { Author.all }

  by :age
end

Include the new type in your schema using the group keyword, and you are done.

class QueryType < GraphQL::Schema::Object
  include GraphQL::Groups

  group :author_group_by, AuthorGroupType
end

You can then run the following query to retrieve the number of authors per age.

query myQuery{ 
  authorGroupBy {
    age {
      key
      count
    }
 }
}

{
  "authorGroupBy":{
    "age":[
      {
        "key":"31",
        "count":1
      },
      {
        "key":"35",
        "count":3
      },
      ...
    ]
  }
}

Why?

graphql-ruby lacks a built in way to retrieve statistical data, such as counts or averages. It is possible to implement custom queries that provide this functionality by using group_by (see for example here), but this performs poorly for large amounts of data.

graphql-groups allows you to write flexible, readable queries while leveraging your database to aggreate data. It does so by performing an AST analysis on your request and executing exactly the database queries needed to fulfill it. This performs much better than grouping and aggregating in memory. See performance for a benchmark.

Advanced Usage

For a showcase of what you can do with graphql-groups check out graphql-groups-demo

Find a hosted version of the demo app on Heroku.

Grouping by Multiple Attributes

This library really shines when you want to group by multiple attributes, or otherwise retrieve complex statistical information within a single GraphQL query.

For example, to get the number of authors grouped by their name, and then also by age, you could construct a query similar to this:

query myQuery{ 
  authorGroups {
    name {
      key
      count
      groupBy {
       age {
        key
        count
       }
      }
    }
 }
}

{
  "authorGroups":{
    "name":[
      {
        "key":"Ada",
        "count":2,
        "groupBy":  {
          "age": [
            {
            "key":"30",
            "count":1
            },
            {
            "key":"35",
            "count":1
            }
          ]     
        }
      },
      ...
    ]
  }
}

graphql-groups will automatically execute the required queries and return the results in a easily parsable response.

Custom Grouping Queries

To customize which queries are executed to group items, you may specify the grouping query by creating a method of the same name in the group type.

class AuthorGroupType < GraphQL::Groups::Schema::GroupType
  scope { Author.all }

  by :age

  def age(scope:)
    scope.group("(cast(age/10 as int) * 10) || '-' || ((cast(age/10 as int) + 1) * 10)")
  end
end

You may also pass arguments to custom grouping queries. In this case, pass any arguments to your group query as keyword arguments.

class BookGroupType < GraphQL::Groups::Schema::GroupType
  scope { Book.all }

  by :published_at do
    argument :interval, String, required: false
  end

  def published_at(scope:, interval: nil)
    case interval
    when 'month'
      scope.group("strftime('%Y-%m-01 00:00:00 UTC', published_at)")
    when 'year'
      scope.group("strftime('%Y-01-01 00:00:00 UTC', published_at)")
    else
      scope.group("strftime('%Y-%m-%d 00:00:00 UTC', published_at)")
    end
  end
end

You may access the query context in custom queries. As opposed to resolver methods accessing object is not possible and will raise an error.

class BookGroupType < GraphQL::Groups::Schema::GroupType
  scope { Book.all }

  by :list_price

  def list_price(scope:)
    currency = context[:currency] || ' $'
    scope.group("list_price || ' #{currency}'")
  end
end

Custom Scopes

When defining a group type's scope you may access the parents object and context.

class QueryType < GraphQL::Schema::Object
  field :statistics, StatisticsType, null: false

  def statistics
    Book.all
  end
end

class StatisticsType < GraphQL::Schema::Object
  include GraphQL::Groups

  group :books, BookGroupType
end

class BookGroupType < GraphQL::Groups::Schema::GroupType
  # `object` refers to `Book.all`
  scope { object.where(author_id: context[:current_person]) }

  by :name
end

Custom Aggregates

Per default graphql-groups supports aggregating count out of the box. If you need to other aggregates, such as sum or average you may add them to your schema by creating a custom GroupResultType. Wire this up to your schema by specifying the result type in your group type.

class AuthorGroupResultType < GraphQL::Groups::Schema::GroupResultType
  aggregate :average do
    attribute :age
  end
end

class AuthorGroupType < GraphQL::Groups::Schema::GroupType
  scope { Author.all }

  result_type { AuthorGroupResultType }

  by :name
end

Per default, the aggregate name and attribute will be used to construct the underlying aggregation query. The example above creates

scope.average(:age)

If you need more control over how to aggregate you may define a custom query by creating a method matching the aggregate name. The method must take the keyword arguments scope and attribute.

class AuthorGroupResultType < GraphQL::Groups::Schema::GroupResultType
  aggregate :average do
    attribute :age
  end

  def average(scope:, attribute:)
    scope.average(attribute)
  end
end

For more examples see the feature spec and test schema

Performance

While it is possible to add grouping to your GraphQL schema by using group_by (see above) this performs poorly for large amounts of data. The graph below shows the number of requests per second possible with both implementations.

The benchmark queries the author count grouped by name, using an increasing number of authors. While the in-memory approach of grouping works well for a small number of records, it is outperformed quickly as that number increases.

Benchmarks can be generated by running rake benchmark. The benchmark script used to generate the report be found here

Limitations and Known Issues

Please refer to the issue tracker for a list of known issues.

Credits

graphql-groups was created at meister

Development

After checking out the repo, run bin/setup to install dependencies. Then, run rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/hschne/graphql-groups. 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.

License

The gem is available as open source under the terms of the MIT License.

Code of Conduct

Everyone interacting in the Graphql::Groups project’s codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.