Traversing relationships

PREVIOUSLY: How to rename your fetched attributes

Until now you can pluck attributes. Now we introduce an option to traverse relationships in the model, so you can pluck not only one model, but any related model, through the reflections option.

Fetching relationships

Imagine, for the previous example, you want to pluck the post information for each category.

Pluckers::Base.new(post.categories.published, {
  attributes: [:name],
  attributes_with_locale: { es: [:name] },
  renames: { name_es: :name_for_analytics},
  reflections: {
    posts: {
      attributes: [:title, :slug, :published_at],
    }
  }
}).pluck

[
  { id: 2, name: "gifs", name_for_analytics: "gifs",
     posts: [
       { id: 33, title: "Lorem Ipsum", slug: 'lorem-ipsum', published_at: "2016-04-07"},
       { id: 32, title: "Lorem Ipsum not", slug: 'lorem-ipsum-not', published_at: nil}
     ]
  },
  { id: 34, name: "shiba", name_for_analytics: "shiba",
     posts: [
       { id: 34, title: "Lorem Ipsum 3", slug: 'lorem-ipsum-3', published_at: "2016-04-09"},
       { id: 35, title: "Lorem Ipsum 4", slug: 'lorem-ipsum-4', published_at: "2016-04-12"}
     ]
  },
  { id: 35, name: "ducktales", name_for_analytics: "patoaventuras"
     posts: [
       { id: 33, title: "Lorem Ipsum", slug: 'lorem-ipsum', published_at: "2016-04-07"}
       { id: 34, title: "Lorem Ipsum 3", slug: 'lorem-ipsum-3', published_at: "2016-04-09"},
       { id: 35, title: "Lorem Ipsum 4", slug: 'lorem-ipsum-4', published_at: "2016-04-12"}
     ]
  }
]

As we just use the attributes option for the reflection we just perform an extra database query, avoiding N+1.

Each element in the reflections options has a key and a hash of options. The key is the name of the relationship as defined in the Active Record model. The value is a hash of options that takes the exact same options that are allowed for the plucker. In fact, internally we create another plucker to retrieve the posts.

This means that we can do everything in this "secondary" plucker. We can get globalize columns, we can rename... and we can also get another related models, giving us the ability to obtain a whole tree of models and objects in just one single point, with the minimum database queries required.

Polymorphic relationships

Polymorphic belongs_to has always been a headache and a source of N+1 queries given you can perform any join or includes operations.

With pluckers you can configure how each model must be plucked, being able to pluck different information that will allow you to perform different login, with the minimum number of queries.

Pluckers::Base.new(BlogPost.all, {
  attributes: [:name],
  reflections: {
    subject: {
      :Author => { attributes: [:name] }
      :Category => { attributes: [:title] }
    }
  }
}).pluck

[
  {
    id: 1, name: "Lorem Ipsum", subject_id: 1, subject_type: "Author", subject: { id: 1, name: "someone" }
  },
  {
    id: 2, name: "Lorem Ipsum 2", subject_id: 2, subject_type: "Author", subject: { id: 2, name: "another one" }
  },
  {
    id: 3, name: "Lorem Ipsum 3", subject_id: 1, subject_type: "Category", subject: { id: 1, title: "gifs" }
  },
  {
    id: 4, name: "Lorem Ipsum 4", subject_id: 2, subject_type: "Category", subject: { id: 1, name: "shiba gifs" }
  },
  {
    id: 5, name: "Lorem Ipsum 5", subject_id: 1, subject_type: "BlogPost", subject: nil
  }
]

The options for each model are standard plucker options, so you can perform the exact same operations you are about to see. This means you can filter the polymorphic relationship with a scope and, more importantly, you can do this in a recursive way.

Foreign keys and minimum data plucked

Although in the examples we only show ids involved, relationships configured with different foreign keys can be fetched too as the configuration is read by the plucker to use the proper columns in both involved tables.

In order to be able to relate the plucked data all the foreign keys must be plucked, that's why in the previous example all the posts plucked their id although in the attributes option we specied only [:title, :slug, :published_at].

Fetching only ids

For has_many and has_and_belongs_to_many relationships you could be interested in only fetching the ids of the related objects. You can get this by using the only_ids option in the relationship to fetch.

Pluckers::Base.new(post.categories.published, {
  attributes: [:name],
  attributes_with_locale: { es: [:name] },
  renames: { name_es: :name_for_analytics},
  reflections: {
    posts: { only_ids: true }
  }
}).pluck

[
  { id: 2, name: "gifs", name_for_analytics: "gifs",
     post_ids: [33, 32]
  },
  { id: 34, name: "shiba", name_for_analytics: "shiba",
     posts_ids: [34, 35]
  },
  { id: 35, name: "ducktales", name_for_analytics: "patoaventuras"
     post_ids: [33, 34, 35]
  }
]