Don't forget to give this project a ⭐
This repo is part of a presentation with the same name. You can find the following resources:
- Slides: https://speakerdeck.com/eliasnogueira/3-design-patterns-and-architecture-decisions-you-must-use-in-your-project
- Video: https://youtu.be/kqwPHdGn7sI?t=22438
It is a simple approach to set up a shared initialization and cleanup in your tests. We commonly categorize it as a Testing pattern.
The problem we are trying to solve here is the reuse of common behaviors across test classes avoiding coding duplication and centralizing these actions at one point.
The application of a Base test class in OO programming languages is applied using inheritance.
Without this approach, we have code duplicated related to:
- pre-condition
- post-condition
- any shared method
With this approach, we have:
- the smart use of inheritance
- an easy way to add more test
- flexible creation of test suites
without the usage of Base Test class
with the usage of Base Test class
This pattern tries to manage the construction process of an object using its constructor in fluent methods.
Without the Builder pattern we create objects like:
class BuilderExample {
@Test
void example() {
// approach using methods
Product product = new Product();
product.setName("iPod");
product.setDescription("10000 songs in your pocket!");
product.setPrice(new BigDecimal("123.99"));
// approach using constructor (not so legible)
Product product = new Product("iPod", "10000 songs in your pocket!", new BigDecimal("123.99"));
}
}The Build creation consists of a new class with the same fields, but with the following changes:
- set methods will return the builder class
- a
buildmethod consisting of the main object creation and all the possible restrictions you might add.
The usage of the Build pattern will bring clarity to the code:
class BuilderExample {
@Test
void example() {
Product product = new ProductBuilder()
.name("iPod")
.description("10000 songs in your pocket!")
.price(new BigDecimal("123.99"))
.build();
}
}This pattern is based in the Build one where we try to provide a readable Fluent API over a specific domain. The domain
can include more than one class.
This might not have the build() method to create an object.
class AmazonPage {
public AmazonPage selectDepartment(Department department) {
// some magic here
return this;
}
public AmazonPage search(String term) {
// some magic here
return this;
}
}This creational pattern enables the creation of objects without exposing the internal logic. A good test example is the creation of different browser instances.
The basic implementation is done by having specific classes that will create different objects, while the main class is responsible to understand the object type to create it.
There are many ways to generate data. The most common are:
- fakes
- static and dynamic generation
Creates an approach to generate non-sensitive data for your test without manually changing the test data in each execution. There are a lot of tools to create this type of data.
Example with java-faker
class JavaFakerExample {
void example() {
Faker faker = new Faker("pt-BR");
faker.name().fullName();
faker.address().fullAddress();
faker.internet().emailAdress();
faker.business().creditCardNumber();
faker.date().birthday();
}
}The best way to get the advantage of the fake data generation is by combining it with the Factory pattern, where we commonly refer it as Data Factory.
We can use a Fake Generation in a centralized data class that can create data in any condition.
class CreditCardDataFactory {
public CreditCard validCreditCard() {}
public CreditCard invalidCreditCardNumber() {}
public CreditCard invalidCvv() {}
}When the data cause different behaviors in your application.
A Static approach can be achieved by implementing any data approach like:
- Class
- CSV | JSON | TXT | YML
- Database
- Mock
It’s a set of common data used across the project consisting of a final class containing constants. The changeless data (constants) will be used in different classes and the advantage is the single point of change. We can have multiple classes shaping the different data requirements.
It is one of the Static data generation types. It consists of using different data in the same test, where the data changes, not the test.
Different test libraries have this approach available. Most of them consist of the data hard-coded in a test class.
The Dynamic approach can be implemented according to your context. Used to remove the maintenance of test data. Example:
- Queries in a database
- Consume data from an API
It provides a way to set different values for a running application without the necessity of re-compile or re-run it. Normally, this approach accepts dynamic values injection through environment variables or by modifying a configuration file.
We can generate logs and reports in different ways, but it’s important to have both in your test project whatever the tools you will choose for this.
By using any log strategy, saving a log file, we can understand the common errors that occurred during the test execution.
These errors can be of:
- assertion errors
- timeout exceptions
- locator exception
- an exception on your architecture
You can also log basic info to know some action you are doing in the code.
- log4j2.properties
- CreditCardDataFactory
3-design-patterns-arch-decision.logfile generated in your user folder
Do not matter if you will generate the test report in any style (below): the most important thing is to have one to satisfy your requirements.
We can have it reported as:
- Gherkin style (readable by humans)
- xUnit style (readable by CI/CD tools)
- HTML (readable by humans)
- any other style like XML, JSON, etc…
- allure.properties
- Allure configuration placed on pom.xml
- auto-generated xUnit XML reports at
target/surefire-reports