Update

We have updated the content of our program. To access the current Software Engineering curriculum visit curriculum.turing.edu.

ActiveRecord Associations with TDD

Learning Goals

  • Write migrations in Rails.
  • Define Schema.
  • Explain what a migration is, and how it relates to our schema.
  • Create one-to-many relationships at the database level using foreign keys.
  • Use has_many and belongs_to to create one-to-many relationship at the model level.

Vocabulary

  • Migration
  • Schema
  • Relationships

Warm Up

  • In your own words, what is a migration?
  • What are some things that we can do with a migration?
  • What is the relationship between a migration and our database?

Setup

This lesson plan starts at the associations-practice branch of this SetList repo. In order to set up the app for this lesson:

  • Clone the repo
  • Checkout the associations-practice branch
  • Run bundle install
  • Run rails db:{drop,create,migrate,seed}

Models, Migrations, and Databases in Rails

In this lesson, we’ll be adding to our new SetList Rails app to demonstrate a one-to-many relationship.

We’ll add a table artists to our database, and connect them to our existing songs table. What might the relationships look like?

One-to-Many Relationships

At the Database Level: Artist

We want to create some artists with a name. Let’s add a test for that! Since this will be a model test, we need to first make a /models directory nested under /spec then create an artist_spec.rb file:

$ mkdir spec/models
$ touch spec/models/artist_spec.rb

We’re going to use the handy dandy gem shoulda-matchers to give us some streamlined syntax to use in testing our validations and relationships.

  • Add gem "shoulda-matchers" to group :development, :test in your Gemfile
  • run bundle install
  • Put the following at the bottom of rails_helper.rb:
Shoulda::Matchers.configure do |config|
  config.integrate do |with|
    with.test_framework :rspec
    with.library :rails
  end
end

spec/models/artist_spec.rb

require 'rails_helper'

describe Artist, type: :model do
  describe "validations" do
    it { should validate_presence_of :name }
  end
end

Run bundle exec rspec and we should get an error similar to this:

Failure/Error:
  describe Artist, type: :model do
    describe "validations" do
      it { should validate_presence_of :name }
    end
  end

NameError:
  uninitialized constant Artist

There are a few things that have to happen to get this test passing.

First, let’s clear this error by creating our Artist model:

$ touch app/models/artist.rb

app/models/artist.rb

class Artist < ApplicationRecord

end

Now, let’s run our test again. We should see an error like this:

ActiveRecord::StatementInvalid:
       PG::UndefinedTable: ERROR:  relation "artists" does not exist
       LINE 9:  WHERE a.attrelid = '"artists"'::regclass
                                   ^
     # ./spec/models/artist_spec.rb:5:in `block (3 levels) in <top (required)>'
     # ------------------
     # --- Caused by: ---
     # PG::UndefinedTable:
     #   ERROR:  relation "artists" does not exist
     #   LINE 9:  WHERE a.attrelid = '"artists"'::regclass
     #                               ^
     #   ./spec/models/artist_spec.rb:5:in `block (3 levels) in <top (required)>

This error is telling us that we don’t have an artists table set up in our database, so let’s create that with a migration:

$ rails g migration CreateArtists name:string

The migration generator creates a migration for us. If we follow the working convention for Rails (e.g. the name:string), the migration will be pre-populated.

Let’s look at the migration inside of db/migrate. It should look like this:

class CreateArtists < ActiveRecord::Migration[7.0]
  def change
    create_table :artists do |t|
      t.string :name

      t.timestamps
    end
  end
end

Will this fix our tests?

Take a look at our db/schema.rb; at this point, it should look something like this:

db/schema.rb

ActiveRecord::Schema[7.0].define(version: 2023_02_22_180629) do
  # These are extensions that must be enabled in order to support this database
  enable_extension "plpgsql"

  create_table "songs", force: :cascade do |t|
    t.string "title"
    t.integer "length"
    t.integer "play_count"
    t.datetime "created_at", null: false
    t.datetime "updated_at", null: false
  end

end

There are two things to focus on here. First, we only see a create_table 'songs' command, which means that our database only has a songs table. Second, take a look at the version: 2023_02_22_180629 - this version number should match the version of our latest migration file, but it doesn’t! That means that we have migrations that have not been run yet. Let’s do that now:

$ rails db:migrate

Now, if we go back to our schema, we should see a command to create_table 'artists', and when we run our tests again, we should be getting a new error!

Let’s run RSpec again.

Failures:

  1) Artist validations is expected to validate that :name cannot be empty/falsy
     Failure/Error: it { should validate_presence_of :name }

       Expected Artist to validate that :name cannot be empty/falsy, but this
       could not be proved.
         After setting :name to ‹""›, the matcher expected the Artist to be
         invalid, but it was valid instead.
     # ./spec/models/artist_spec.rb:5:in `block (3 levels) in <top (required)>'

The important part to read here is Expected Artist to validate that :name cannot be empty/falsy

Let’s add a validation to Artist!

app/models/artist.rb

class Artist < ApplicationRecord
  validates_presence_of :name
end

Run RSpec again and we get passing tests.

What about Songs?

What’s the relationship between song and artist? Draw this out in a diagram to help visualize the relationship.

Let’s create a test to help us drive this out. Add the following to your artist_spec.rb within the greater describe Artist block, but outside of the validations block.

spec/models/artist_spec.rb

require 'rails_helper'

describe Artist, type: :model do
  describe "validations" do
    it { should validate_presence_of :name }
  end

  describe 'relationships' do
    it { should have_many :songs }
  end
end

When we run this test we get an error something like this:

1) Artist relationships is expected to have many songs
     Failure/Error: it { should have_many :songs }
       Expected Artist to have a has_many association called songs (no association called songs)
     # ./spec/models/artist_spec.rb:9:in `block (3 levels) in <top (required)>'

The important part here is Expected Artist to have a has_many association called songs (no association called songs). This tells us that we are missing a relationship. We need to make one.

$ rails g migration AddArtistsToSongs artist:references

(It helps me to remember that the thing that gets put in the migration is the thing that is the one in the one to many relationship, in this case, the Artist has many Songs.)

Take a look at what this migration creates.

class AddArtistsToSongs < ActiveRecord::Migration[7.0]
  def change
    add_reference :songs, :artist, null: false, foreign_key: true
  end
end

Run rails db:migrate to run this migration. Now open schema.rb and make sure that your songs table now has the artist_id foreign key.

Run the tests again and you’ll see that we’re still getting Expected Artist to have a has_many association called songs (no association called songs). We’ve added the foreign key at the database level, but our we still haven’t set up our model to make use of that foreign key.

Associations

One-to-Many Relationships at the Model Level: Song/Artist

We have set up our database level relationships, now let’s implement some model-level associations using some handy ActiveRecord methods.

  • has_many
  • belongs_to

app/models/song.rb

class Song < ApplicationRecord
  belongs_to :artist

end

app/models/artist.rb

class Artist < ApplicationRecord
  has_many :songs

  validates_presence_of :name
end

Run the tests again and now they should be passing!

Why do we need a foreign key at the database level and the belongs_to method in the model? What do each of these things allow for?

Let’s play around in our development database by dropping in to the rails console rails console or rails c

In the console:

  • Create a artist Artist.create!(name: 'Prince')
  • Create a song Song.create!(title: 'Raspberry Beret', length: 345, play_count: 34)

Did you get an error?

  • Why are we getting this error?
  • What do we need to do to fix this error?

Now that a song belongs to an artist, a song can not exist without an artist

  • What are different ways to associate songs with artists?
artist = Artist.create(name: 'Prince')
song = Song.create(title: 'Raspberry Beret', length: 345, play_count: 34, artist: artist)

# OR

artist = Artist.create(name: 'Prince')
song = Song.create(title: 'Raspberry Beret', length: 345, play_count: 34, artist_id: artist.id)
# OR

artist = Artist.create(name: 'Prince')
song = artist.songs.create(title: 'Raspberry Beret', length: 345, play_count: 34)

Since we’ve added code into our Song model, it would also be a good idea to create a song_spec.rb (if you don’t already have one) and add a test for the relationship from Songs to Artists:

$ touch spec/models/song_spec.rb

spec/models/song_spec.rb

require 'rails_helper'

RSpec.describe Song do

  describe 'relationships' do
    it {should belong_to :artist}
  end
end

Seeds

Now that our App is getting more complex, it would be good to add some seeds. Seeding your database is when you insert a set of data into the database. It is useful to have some seed data when we are experimenting and developing. You could consider what we’ve done in the Rails Console so far as a type of seeding, but doing things manually in the Rails Console can get very tedious, so what we will do instead is write a script to seed our database that we can reuse. Rails comes with a file for us to write this script in db/seeds.rb. Open up that file and add the following:

db/seeds.rb

Song.destroy_all
Artist.destroy_all

prince = Artist.create!(name: 'Prince')
rtj = Artist.create!(name: 'Run The Jewels')
caamp = Artist.create!(name: 'Caamp')
jgb = Artist.create!(name: 'Jerry Garcia Band')
billie = Artist.create!(name: 'Billie Eilish')
lcd = Artist.create!(name: 'LCD Soundsystem')

prince.songs.create!(title: 'Raspberry Beret', length: 345, play_count: 34)
prince.songs.create!(title: 'Purple Rain', length: 524, play_count: 19)

rtj.songs.create!(title: 'Legend Has It', length: 2301, play_count: 2300000)
rtj.songs.create!(title: 'Talk to Me', length: 2301, play_count: 2300000)

caamp.songs.create!(title: '26', length: 940, play_count: 150000)
caamp.songs.create!(title: 'Vagabond', length: 240, play_count: 120000)

jgb.songs.create!(title: 'Aint No Bread In The Breadbox', length: 540, play_count: 12000)
jgb.songs.create!(title: 'The Harder They Come', length: 240, play_count: 120000)

billie.songs.create!(title: 'bury a friend', length: 340, play_count: 1200000)
billie.songs.create!(title: 'bad guy', length: 240, play_count: 100000)

lcd.songs.create!(title: 'Someone Great', length: 500, play_count: 1000000)
lcd.songs.create!(title: 'I Can Change', length: 640, play_count: 100000)

Now that we have our seeds file, we can run it with rails db:seed. Additionally, if we check this file into our version control system, other developers working on this app will be able to easily seed their local databases.

Notice that the first two lines of this seeds file will destroy all Songs and Artists from the database. The reason we want to do this is so that we know we are starting with an empty database every time we want to reseed our database. If we did not have these two lines, this script would create duplicate records every time we reran rails db:seed.

Adding Behaviors to Models

Now, we have two models that are related to each other with has_many and belongs_to, and these models can handle basic CRUD functionality through the methods that they inherit from ActiveRecord - but what if we need our models to be customized to perform behaviors related to our application? For example, what if we want our Song model to be able to tell us how many songs exist in our database, and we want an artist to be able to tell us the average length of all their songs.

First - Test!

spec/models/song_spec.rb

require 'rails_helper'

RSpec.describe Song do
  describe 'relationships' do
    it {should belong_to :artist}
  end

  describe 'class methods' do
    it '.song_count' do
      prince = Artist.create!(name: 'Prince')
      talking_heads = Artist.create!(name: 'Talking Heads')
      rasperry_beret = prince.songs.create!(title: 'Raspberry Beret', length: 234, play_count: 34)
      wild_life = talking_heads.songs.create!(title: 'Wild Wild Life', length: 456, play_count: 45)

      expect(Song.song_count).to eq(2)
    end
  end
end

Use TDD to create a class method on our Song model that returns a count of the songs in our database. As you build out this method, remember to use pry and self to help debug and guide your implementations!

And for our next method - a test!

spec/models/artist_spec.rb

require 'rails_helper'

RSpec.describe Artist do
  describe 'validations' do
    it {should validate_presence_of :name}
  end

  describe 'relationships' do
    it {should have_many :songs}
  end

  describe 'instance methods' do
    it '.average_song_length' do
      talking_heads = Artist.create!(name: 'Talking Heads')
      she_was = talking_heads.songs.create!(title: 'And She Was', length: 234, play_count: 34)
      wild_life = talking_heads.songs.create!(title: 'Wild Wild Life', length: 456, play_count: 45)

      expect(talking_heads.average_song_length).to eq(345)
    end
  end
end

Use TDD to create an instance method on our Artist model that returns the average of a single artist’s songs. As you build out this method, remember to use pry and self to help debug and guide your implementations!

Wrap Up

  • What are two different types of table relationships that you might need to implement? In what scenario would you use each?
  • What is the syntax for the following migrations in Rails?
    • Create a table
    • Add a column to a table, with or without a data type
    • Add a reference from one table to another

Completed code from this lesson plan can be found on this branch here.

Lesson Search Results

Showing top 10 results