Mocking with RSpec: Doubles and Expectations

Mocking is a powerful TDD technique which should also be used with care. This article will help you understand and apply it correctly using RSpec for Ruby.


Mocking is a technique in test-driven development (TDD) that involves using fake dependent objects or methods in order to write a test. There are a couple of reasons why you may decide to use mock objects:

  • As a replacement for objects that don't exist yet.
  • When you are working with objects which return non-deterministic values or depend on an external resource, e.g. a method that returns an RSS feed from a server.
  • To avoid setting up a complex scheme of data or dependency objects in order to write a test.
  • To avoid invoking code which would degrade the performance of the test, while at the same time being unrelated to the test you are writing.

The first reason is particularly prevalent among those who practice behavior-driven development (BDD). The dependent object needs to exist, but it's not what you're currently working on. You can take advantage of the complete freedom of defining its interface to make it as nice and clear as possible, which is usually equivalent to being the simplest thing to write in the ongoing test.

By mocking objects in advance, you can allow yourself to focus on the thing that you're working on at the moment. Let's say that you are working on a new part of the system, and you realize that the code you're currently describing and implementing will require two new collaborating objects. Using mocks, you can define their interfaces as you write a spec for the code you're currently working on.

That way, you maintain a clean environment by having all your tests pass, before moving on to implement the collaborating objects. Without mocks, you'd be required to immediately jump to writing the implementation for the collaborating objects, before having your tests pass. This can be distracting and may lead to poor code design decisions. Mocking helps us by reducing the number of things we need to keep in our head at a given moment.

Mocking with RSpec is done with the rspec-mocks gem. If you have rspec as a dependency in your Gemfile, you already have rspec-mocks available.


A test double is a simplified object which takes the place of another object in a test. Creating a double with RSpec is easy:

feed = double

# Optionally, you may give your double an identifier, which may come handy
# when debugging and inspecting objects:
feed = double("feed")

Method Stubs

A new double resembles a plain Ruby Object — it's not very useful on its own. It is usually the first step before defining some fake methods on it. This is called method stubbing, and with RSpec 3 it is done using the allow() and receive() methods:

allow(feed).to receive(:fetch).and_return("imagine I'm a JSON string")

=> "imagine I'm a JSON string"

The value provided to and_return() defines the return value of the stubbed method. The use of and_return() is optional, and if you don't have it, the stubbed method will be set to return nil.

You can also apply allow to a real object (which is not a double). When testing Rails applications, for example, it is common to mock a method which works with the database and make it return a predefined double whenever the focus of the code is not whether the database operations work or not (and the test can run faster too).

comment = double("comment")
expect(Comment).to receive(:find).and_return(comment)

We'll see how we can apply this below.

Message Expectations

Expecting messages — that is, defining expectations on test doubles that certain methods will be invoked after some code that follows runs — is a common pattern when working with doubles.

For example, let's say that we're working on a background job class whose job is to send an email via a mailer class. In BDD mocking style, we would write a spec for it by setting message expectations.

First, we know that our class needs to work with email records, available through a Rails model called Email. So, our first step is to define that our job class begins its work by finding the right email record based on an ID it receives as an argument.

describe EmailVerificationJob do
  describe "#perform" do

    it "finds the email by id" do
      expect(Email).to receive(:find).with(12)

In the code above, we set an expectation that if we run EmailVerificationJob.perform(12), its implementation will call Email.find(12). We can make that pass easily with the following implementation:

class EmailVerificationJob
  def perform(email_id)
    email = Email.find(email_id)

Next, we want to say that a mailer class should take the email record returned by Email#find and use it to send an actual email.

In order to glue the two actions together, we need to mock Email#find to return a double which we control. Then, we can use that double to set an expectation that a mailer class is using it correctly to send an email.

describe EmailVerificationJob do
  describe "#perform" do

    it "finds the email by id" { ... }

    it "sends the verification email" do
      email = double
      allow(Email).to receive(:find) { email }

      expect(UserMailer).to receive(:send_verification_email).with(email)

Note how we can compose receive with with to make sure that the send_verification_email method is being passed the right parameter. RSpec supports many ways to write argument matchers, which you can read more about in the documentation.

The final implementation of our job class should be simple, as follows:

class EmailVerificationJob
  include Sidekiq::Worker

  def perform(email_id)
    email = Email.find(email_id)

However, when we run our spec, our first example fails:

$ bundle exec rspec
    finds the email by id (FAILED - 1)
    sends the verification email


  1) EmailVerificationJob#perform finds the email by id
       undefined method `send_verification_email' for UserMailer:Class
     # ./lib/email_verification_job.rb:8:in `perform'
     # ./spec/lib/email_verification_job_spec.rb:12:in `block (3 levels)'

Finished in 0.30142 seconds (files took 4.91 seconds to load)
2 examples, 1 failure

UserMailer#send_verification_email doesn't exist yet, but we know this and it's not an issue we are concerned about right now. So, we can make the example pass by mocking the send_verification_email method by allow-ing it to be invoked:

describe "#perform" do
  it "finds the email by id" do
    allow(UserMailer).to receive(:send_verification_email)
    expect(Email).to receive(:find).with(12)

This time all of our examples pass.

$ bundle exec rspec
    finds the email by id
    sends the verification email

Finished in 0.29838 seconds (files took 4.92 seconds to load)
2 examples, 0 failures

A Word of Caution

Note that in the example above we've reached a green spec (and possibly, the whole test suite), while at the same time our implementation contains code calling an undefined method. Are we just fooling ourselves with all this mocking?

This is one of the reasons why there are people who prefer to never mock. We won't go into that argument in this tutorial. With great power (of expressiveness and flexibility of code we can write and interfaces we can design by mocking, in this case) comes great responsibility.

It is important to bear in mind that a mocking approach should always be used within a development cycle that begins with a high-level integration or acceptance test. In Ruby applications, this frequently means writing a Cucumber scenario which describes and checks for the key points of the business logic, before diving into the implementation code. There, in the lower layers of the code, mocking can sometimes be a good way to drive your code.

Message Chaining

You may need to mock chained method calls. For example, given the following code:

@comments = Comment.where(:post_id =>"created_at DESC")

A corresponding spec code can use RSpec's receive_message_chain:

allow(Comment).to receive_message_chain(:where, :order) { ... }

The equivalent of receive_message_chain in pre-RSpec 3 syntax was usually presented as one of the basic features of the library. However, it is now considered to be a code smell.

If you read the spec line of code above, you'll notice that it's not telling you anything about what the implementation is supposed to do. where and order are very generic database query methods. Frequent method chaining usually leads to violation of the Law of Demeter, making code difficult to reason about and refactor in the future.

Chaining methods, however, sometimes cannot be completely avoided and simply must end up being somewhere. For this reason, you will be better off by encapsulating the chain with a domain-defining method, in our example:

class Comment
  def self.for_post(post)
    where(:post_id =>"created_at DESC")

The spec for the code that is using the method to do higher-order work would then simply mock the high-level method:

allow(Comment).to receive(:for_post) { ... }

We could then write a full non-mocking spec for the Comment.for_post method separately.


This tutorial showed you the most frequent ways of using test doubles and set message expectations with RSpec, which you can start applying in your project now.

If mocking is a new concept for you, it may take some practice to get used to and work with effectively. Once you're comfortable with the basics, you can explore the full documentation of rspec-mocks for the complete API.

P.S. Would you like to learn how to build sustainable Rails apps and ship more often? We've recently published an ebook covering just that — "Rails Testing Handbook". Learn more and download a free copy.

Marko Anastasov

Rendered Text / Semaphore cofounder.

on this tutorial so far.
User deleted author {{comment.createdAt}}

Edited on {{comment.updatedAt}}


Sign In You must be logged in to comment.