Combining Object Mother and Fluent Builder for the Ultimate Test Data Factory

Table Of Contents

To test our business code we always need some kind of test data. This tutorial explains how to do just that with the Object Mother pattern and why we should combine it with a Fluent Builder to create test data factories that are fun to work with.

Example Code

This article is accompanied by a working code example on GitHub.

What do we Need a Test Data Factory For?

Let’s imagine that we want to create some tests around Invoice objects that are structured as shown in the figure below.

Invoice UML

An Invoice has a target Address and zero or more InvoiceItems, each containing the amount and price of a certain product that is billed with the invoice.

Now, we want to test our invoice handling business logic with a couple of test cases:

  1. a test verifying that invoices with an abroad invoice address are sent to an invoicing service specialized on foreign invoicing
  2. a test verifying that a missing house number in an invoice address leads to a validation error
  3. a test verifying that an invoice with a negative total price is forwarded to a refund service

For each of these test cases, we obviously need an Invoice object in a certain state:

  1. an invoice with an address in another country,
  2. an invoice with an address with a missing house number,
  3. and an invoice with a negative total price.

How are we going to create these Invoice instances?

Of course, we can go ahead an create the needed Invoice instance locally in each test case. But, alas, creating an Invoice requires creating some InvoiceItems and an Address, too … that seems like a lot of boiler plate code.

Apply the Object Mother Pattern to Reduce Duplication

The example classes used in this article are rather simple. In the real world, classes like Invoice, InvoiceItem or Address can easily contain 20 or more fields each.

Do we really want to have code that initializes such complex object graphs in multiple places of our test code base?

Bad test code structure hinders the development of new features just as much as bad production code, as Robert C. Martin’s Clean Architecture has once more brought to my attention (link points to ebooks.com; read my book review).

So, let’s try to keep test code duplication to a minimum by applying the Object Mother pattern.

The Object Mother pattern is essentially a special case of the Factory pattern used for creating test objects. It provides one or more factory methods that each create an object in a specific, meaningful configuration.

In a test, we can call one of those factory methods and work with the object created for us. If the pre-defined object returned by the Object Mother doesn’t fully meet our test requirements, we can go ahead and change some fields of that object locally so that it meets the requirements of our test.

In our example, the Object Mother might provide these factory methods for pre-defined Invoice objects:

  • InvoiceMother.complete(): creates a complete and valid Invoice object including sensibly configured InvoiceItems and a valid Address
  • InvoiceMother.refund(): creates a complete and valid Invoice object with a negative total price

For our three test cases, we can then use these factory methods:

  1. To create an Invoice with an abroad address, we call InvoiceMother.complete() and change the country field of the address locally
  2. To create an Invoice with a missing house number, we call InvoiceMother.complete() and remove the house number from the address locally
  3. To create an Invoice with a negative total price, we simply call InvoiceMother.refund()

The goal of the Object Mother pattern is not to provide a factory method for every single test requirement we might have but instead to provide ways to create a few functionally meaningful versions of an object that can be easily adapted within a concrete test.

Even with that goal in mind, over time, an Object Mother might degrade to the code equivalent of a termite queen, birthing new objects for each and every use case we might have. In every test case, we’d have a dependency to our Object Mother to create objects just right for the requirements at hand.

Each time we change one of our test cases, we would also have to change the factory method in our Object Mother. This violates the Single Responsibility Principle since the Object Mother must be changed for a lot of different reasons.

We stated above that we want to keep our test code base clean, so how can we reduce the risk for violating the Single Responsibility Principle?

Introduce the Fluent Builder Pattern to Promote the Single Responsibility Principle

That’s where the Builder pattern comes into play.

A Builder is an object with methods that allow us to define the parameters for creating a certain object. It also provides a factory method that creates an object from these parameters.

Instead of returning readily initialized objects, the factory methods of our Object Mother now return Builder objects that can be further modified by the client to meet the requirements of the specific use case.

The code for creating an Invoice with a modified address might look like this:

Invoice.InvoiceBuilder invoiceBuilder = InvoiceMother.complete();
Address.AddressBuilder addressBuilder = AddressMother.abroad();
invoiceBuilder.address(addressBuilder.build());
Invoice invoice = invoiceBuilder.build();

So far, we haven’t really won anything over the pure Object Mother approach described in the previous section. Our InvoiceMother now simply returns instances of InvoiceBuilder instead of directly returning Invoice objects.

Let’s introduce a fluent interface to our Builder. A fluent interface is a programming style that allows to chain multiple method calls in a single statement and is perfectly suited for the Builder pattern.

The code from above can now be changed to make use of this fluent interface:

Invoice invoice = InvoiceMother.complete()
  .address(AddressMother.abroad()
    .build())
  .build();

But why should this reduce the chance for violating the Single Responsibility Principle in an Object Mother class?

With a fluent API and an IDE that supports code completion, we can let the API guide us in creating the object we need.

Having this power at our fingertips we’ll more likely configure the specific Invoice we need in our test code and we’ll less likely create a new factory method in our Object Mother that is probably only relevant four our current test.

Thus, combining the Object Mother pattern with a fluent Builder reduces the potential of violating the Single Responsibility Principle by making it easier to do the right thing.

May a Factory Method Call Another Factory Method?

When creating an Object Mother (or actually any other kind of factory), a question that often arises is: “May I call another factory method from the factory method I’m currently coding?”.

My answer to this question is a typical “yes, but…”.

Of course, we may take advantage of other existing Object Mothers. For instance, in the code of InvoiceMother, we may happily call AddressMother and InvoiceItemMother:

class InvoiceMother {

  static Invoice.InvoiceBuilder complete() {
    return Invoice.Builder()
        .id(42L)
        .address(AddressMother.complete()
          .build())
        .items(Collections.singletonList(
          InvoiceItemMother.complete()
        	    .build()));
  }
	
}

But the same rules apply as in our client test code. We don’t want to add responsibilities to our factory method that don’t belong there.

So, before creating a custom factory method in an Object Mother we want to call from the factory method we’re currently coding, let’s think about whether we should rather use one of the pre-defined factory methods and customize the returned builder via fluent API to suit our requirements.

Conclusion

The Object Mother pattern by itself is a big help in quickly getting pre-defined objects to use in tests.

By returning Builders with a fluent API instead of directly returning object instances, we add a lot of flexibility to our test data generation, which makes creating new test objects for any given requirement a breeze. It supports the Single Responsibility Principle by making it easy to adjust created objects locally.

Further Reading

Example Code

This article is accompanied by a working code example on GitHub.
Written By:

Tom Hombergs

Written By:

Tom Hombergs

As a professional software engineer, consultant, architect, general problem solver, I've been practicing the software craft for more than fifteen years and I'm still learning something new every day. I love sharing the things I learned, so you (and future me) can get a head start. That's why I founded reflectoring.io.

Recent Posts

Guide to JUnit 5 Functional Interfaces

In this article, we will get familiar with JUnit 5 functional interfaces. JUnit 5 significantly advanced from its predecessors. Features like functional interfaces can greatly simplify our work once we grasp their functionality.

Read more

Getting Started with Spring Security and JWT

Spring Security provides a comprehensive set of security features for Java applications, covering authentication, authorization, session management, and protection against common security threats such as CSRF (Cross-Site Request Forgery).

Read more

Creating and Publishing an NPM Package with Automated Versioning and Deployment

In this step-by-step guide, we’ll create, publish, and manage an NPM package using TypeScript for better code readability and scalability. We’ll write test cases with Jest and automate our NPM package versioning and publishing process using Changesets and GitHub Actions.

Read more