Frontend

End-to-end testing

At GoodRequest, our goal is to create a meaningful and responsible digital world. For this reason, we have long been thinking how to improve testing processes, increase testing efficiency in the case of significant application changes, ensure the functionality of critical application parts, and deliver a robust and reliable software product to our clients.

Roman Haluška28 Nov 2023

What are the reasons and causes why developers do not implement end-to-end tests? 

There are many reasons why developers don't implement end-to-end tests, but it's important to remember that testing is a crucial part of software development. End-to-end tests are usually costly to implement and maintain, and sometimes, writing them to be both reliable and effective can be challenging.

Some of the reasons and causes why developers do not implement E2E tests are:

  1. Lack of time: Developers often face time constraints during development and focus on implementing functionality rather than testing.
  2. Lack of resources: Teams may not always have a enough number of testing resources, such as human resources or financial means for test automation.
  3. Test complexity: End-to-end tests can be challenging to implement and maintain as they require testing of the integrated system, not just individual components. Setting up the testing environment and executing and evaluating tests can also be problematic.
  4. Focus on other test types: Developers can focus on other types of tests, such as unit tests or integration tests, which are simpler to implement and provide faster feedback.
  5. Insufficient motivation: If there is not enough emphasis on software quality or a lack of awareness of testing benefits, it can lead to a lack of motivation to implement end-to-end tests.

Despite these reasons, it is important to understand that testing is key to ensuring software quality and minimizing errors. Implementing end-to-end tests can lead to more robust and reliable software products, which should be the goal of every developer.

Why did we choose Cypress? 

Within our team, we had our first experience with the Selenium tool, often compared to Cypress. The setup and implementation of some scenarios using Selenium were relatively complicated, so we decided to give Cypress a try. When choosing between Selenium and Cypress, it is important to consider project needs and team skills. Selenium is versatile and scalable, while Cypress is optimized for modern web applications, offering simpler and faster test development. Therefore, we chose Cypress as it better suited our needs.

For more information about Cypress, you can visit their blog. Further details on Cypress architecture and key features can be found in the article: Key differences.

Testing organization

After successfully installing Cypress into the project, the following file structure is created. More information about the structure and individual components can be found in this article: Writing and organizing tests.

Cypress configuration

Our recommendations for configuring individual options are as follows:

  • If you observe high memory consumption during test execution, consider setting the option numTestsKeptInMemory to 0 to optimize it.
  • For optimization and resource saving during test execution, we found it useful to turn off video compression. → videoCompression: false.
  • Sometimes a test may fail for unknown reasons during execution. In such cases, using the retries option for test reruns has proven helpful. An optimal value for this problem is 3. More information can be found in the article: Test retries. 
  • Some tests may fail because Cypress "misses" an element within the time limit due to animation or asynchronous operations blocking the element's display. In this case, it is necessary to increase the defaultCommandTimeout to 6000.

For more information about the basic and other options, you can refer to the article: Configuration.

Our configuration file

import { defineConfig } from 'cypress'

export default defineConfig({
  projectId: '<projectID>',
  scrollBehavior: 'center',
  viewportWidth: 1920,
  viewportHeight: 1080,
  retries: 3,
  numTestsKeptInMemory: 10,
  videoCompression: false,
  screenshotOnRunFailure: false,
  e2e: {
    setupNodeEvents: (on, config) => {
		require('cypress-localstorage-commands/plugin')(on, config)
		require('./cypress/plugins/index.ts').default(on, config)
		require('@cypress/code-coverage/task')(on, config)
		on('before:browser:launch', (browser, launchOptions) => {
			// Chrome is used by default for test:CI script
			if (browser.name === 'chrome') {
				launchOptions.args.push('--disable-dev-shm-usage')
			} else if (browser.name === 'electron') {
				launchOptions.args['disable-dev-shm-usage'] = true
			}

			return launchOptions
		})
		return config
    },
	env: {
		auth_email: process.env.AUTH_EMAIL,
		auth_password: process.env.AUTH_PASSWORD,
		sign_in_url: process.env.SIGN_IN_URL
	},
	experimentalRunAllSpecs: true,
	baseUrl: 'http://localhost:3001',
	defaultCommandTimeout: 6000
  },
})

Examples of suite test and time test

TestSuite → describe, context | TestCase → it, specify

More information can be found in the article: Writing tests.

describe('Users crud operations', () => {
    it('Create user', () => {
      ...
    })

		it('Show user', () => {
      ...
    })

    it('Update user', () => {
      ...
    })

		it('Delete user', () => {
      ...
    })
})

// alebo

context('Users crud operations', () => {
    specify('Create user', () => {
      ...
    })

		specify('Show user', () => {
      ...
    })

    specify('Update user', () => {
      ...
    })

		specify('Delete user', () => {
      ...
    })
})

Naming convention of folders

This naming convention and folder/test hierarchy model has proven effective for larger projects requiring the implementation of a significant number of tests.

Hooks

Within our team, we use hooks for authorization, token storage, and token retrieval to avoid logging in before each test, as tests are isolated.

before(() => {
	loginViaApi(email, password)
})

beforeEach(() => {
	// restore local storage with tokens and salon id from snapshot
	cy.restoreLocalStorage()
})

afterEach(() => {
	// take snapshot of local storage with new refresh and access token
	cy.saveLocalStorage()
})

Hooks defined in the support folder and the e2e.ts file automatically run before each test suite. You can use this method when logging in.

describe('Hooks', () => {
	it('loginViaApi', () => {
		cy.log(`sign_in_url is ${Cypress.env('sign_in_url')}`)
		cy.log(`auth_email is ${Cypress.env('auth_email')}`)
		cy.log(`auth_password is ${Cypress.env('auth_password')}`)
		cy.apiAuth(Cypress.env('auth_email'), Cypress.env('auth_password'), Cypress.env('sign_in_url'))
	})
})

An example function we use to obtain tokens needed for authorization during test execution.

export const loginViaApi = (user?: string, password?: string) => {
	cy.apiAuth(user || Cypress.env('auth_email'), password || Cypress.env('auth_password'), Cypress.env('sign_in_url'))
}

Hooks can also run commands or functions to prepare data expected during tests. For more information, read the following article: Hooks.

Excluding and Including Tests

More about excluding and including tests can be found in this article: Excluding and Including tests.

Debugging

For more information about debugging, refer to the article: Debugging.

Testing/data strategies

Before executing each test, it is essential to have test data stored, for example, in a database. There are many ways to prepare this data before testing. 

More information can be found in the article: Testing strategies.

Seeding

Options for preparing data needed for individual tests.

describe('Users CRUD operations', () => {
  beforeEach(() => {
    // pred každým testom môžete spustiť príkaz, ktorý premaže databázu
		// a vytvorí nové záznamy
    cy.exec('npm run db:reset && npm run db:seed')

    // vytvorenie používateľa
    cy.request('POST', '/user', {
      body: {
				// ... testovacie data
			},
    })
  })

	...
})

 For more information see the article: Database initialization & seeding.

Stubbing

An example of intercepting a post request.

{
	...
	cy.intercept('POST', '/user/*', {
		statusCode: 200,
		body: {
			name: 'John Garfield'
		}
	}).as('createUser')
	...
}

For more information, read here.

Best practices, tips and tricks

Selecting elements - a.k.a. selectors

Example of a correct component identifier.

...
<div className={cx('input-inner-wrapper', { 'to-check-changes': toCheck })}>
				<Input
					{...input}
					id={formFieldID(form, input.name)}
					{/* alebo */}
					data-cy={formFieldID(form, input.name)}
					className={cx('input', { 'input-filter': fieldMode === FIELD_MODE.FILTER })}
					onChange={onChange}
					onBlur={onBlur}
					addonBefore={addonBefore}
					size={size || 'middle'}
					onFocus={onFocus}
					value={input.value}
				/>
</div>
...

Example of a command to set the value for a "pin" component. As you can see below, a unique ID is used to retrieve the element, which in our case consists of a unique form name and a field.

Cypress.Commands.add('setValuesForPinField', (form: string, key: string, value: string) => {
	const elementId: string = form ? `#${form}-${key}` : `#${key}`
	const nthInput = (n: number) => `${elementId} > :nth-child(${n})`
	const pin = [...value]
	pin.forEach((char: string, index) =>
		cy
			.get(nthInput(index + 1))
			.type(char)
			.should('have.value', char)
	)
})

Creating custom commands

Example of how we use “custom commands” to select and set values for individual elements. 

Cypress.Commands.add('selectOptionDropdownCustom', (form?: string, key?: string, value?: string, force?: boolean) => {
	const elementId: string = form ? `#${form}-${key}` : `#${key}`
	cy.get(elementId).click({ force })
	if (value) {
		// check for specific value in dropdown
		cy.get('.ant-select-dropdown :not(.ant-select-dropdown-hidden)', { timeout: 10000 })
			.should('be.visible')
			.find('.ant-select-item-option')
			.each((el: any) => {
				if (el.text() === value) {
					cy.wrap(el).click({ force })
				}
			})
	} else {
		// default select first option in list
		cy.get('.ant-select-dropdown :not(.ant-select-dropdown-hidden)', { timeout: 10000 }).should('be.visible').find('.ant-select-item-option').first().click({ force: true })
	}
})

Cypress.Commands.add('clickDropdownItem', (triggerId: string, dropdownItemId?: string, force?: boolean) => {
	cy.get(triggerId).click({ force })
	if (dropdownItemId) {
		// check for specific value in dropdown
		cy.get('.ant-dropdown :not(.ant-dropdown-hidden)', { timeout: 10000 })
			.should('be.visible')
			.find('.ant-dropdown-menu-item')
			.each((el: any) => {
				if (el.has(dropdownItemId)) {
					cy.wrap(el).click({ force })
				}
			})
	} else {
		// default select first item in list
		cy.get('.ant-dropdown :not(.ant-dropdown-hidden)', { timeout: 10000 }).should('be.visible').find('.ant-dropdown-menu-item').first().click({ force: true })
	}
})

More information can be found in the article: Custom commands.

Assigning Return Values

Instead of defining constants and storing the selected element, it is necessary to use so-called aliases.

// DONT DO THIS. IT DOES NOT WORK
// THE WAY YOU THINK IT DOES.
const a = cy.get('a')

cy.visit('https://example.cypress.io')

// nope, fails
a.first().click()

// Instead, do this.
cy.get('a').as('links')
cy.get('@links').first().click()

More information can be found in the article: Variables and aliases.

Testing email notifications 

What if we need to test registration, including email confirmation or activation codes? In such cases, the "task plugin" is useful for loading emails from the mailbox. 

In Cypress, it is possible to use the "task plugin" to perform "custom" tasks and modify the testing environment's behavior. These tasks can include various operations, such as accessing files, data manipulation, API calls, setting various variables, and more. 

In our case, we use a "task" to load emails and retrieve activation codes. Below is an example of defining a task and its use in a test.

  • example of a “task” → ./plugins/index.ts
import axios from 'axios'

/**
 * @type {Cypress.PluginConfig}
 */
export default (on: any, config: any) => {
	on('task', {
		getEmail(email: string) {
			return new Promise((resolve, reject) => {
				axios
					.get(`http://localhost:1085/api/emails?to=${email}`)
					.then((response) => {
						if (response) {
							resolve(response.data)
						}
					})
					.catch((err) => {
						reject(err)
					})
			})
		}
	})

	return on
}
  • example test
it('Sign up', () => {
		cy.intercept({
			method: 'POST',
			url: '/api/sign-up'
		}).as('registration')
		cy.visit('/')
		cy.wait('@getConfig').then((interceptionGetConfig: any) => {
			// check status code of login request
			expect(interceptionGetConfig.response.statusCode).to.equal(200)
			cy.clickButton(SIGNUP_BUTTON_ID, FORM.LOGIN)
			// check redirect to signup page
			cy.location('pathname').should('eq', '/signup')
			cy.setInputValue(FORM.REGISTRATION, 'email', userEmail)
			cy.setInputValue(FORM.REGISTRATION, 'password', user.create.password)
			cy.setInputValue(FORM.REGISTRATION, 'phone', user.create.phone)
			cy.clickButton('agreeGDPR', FORM.REGISTRATION, true)
			cy.clickButton('marketing', FORM.REGISTRATION, true)
			cy.clickButton(SUBMIT_BUTTON_ID, FORM.REGISTRATION)
			cy.wait('@registration').then((interceptionRegistration: any) => {
				// check status code of registration request
				expect(interceptionRegistration.response.statusCode).to.equal(200)
				// take local storage snapshot
				cy.saveLocalStorage()
			})
			// check redirect to activation page
			cy.location('pathname').should('eq', '/confirmation')

			cy.task('getEmail', userEmail).then((email) => {
				if (email && email.length > 0) {
					const emailHtml = parse(email[0].html)
					const htmlTag = emailHtml.querySelector('#confirmation-code')
					if (htmlTag) {
						cy.log('Confirmation code: ', htmlTag.text)
						cy.visit('/confirmation')
						cy.intercept({
							method: 'POST',
							url: '/api/confirmation'
						}).as('activation')
						cy.setValuesForPinField(FORM.ACTIVATION, 'code', htmlTag.text)
						cy.clickButton(SUBMIT_BUTTON_ID, FORM.ACTIVATION)
						cy.wait('@activation').then((interception: any) => {
							// check status code of registration request
							expect(interception.response.statusCode).to.equal(200)
							// take local storage snapshot
							cy.saveLocalStorage()
						})
					}
				}
			})
		})
	})

For more information, read the article: Task.

Code coverage

We use the babel-plugin-istanbul and cypress/instrument-cra for code instrumentation. The cypress/instrument-cra library allows activating instrumented code even in "dev mode" or when running the local dev server.

{
  "scripts": {
		...
    "start": "react-scripts -r @cypress/instrument-cra start",
		...
  }
}

The script we use for an instrumented build of the application for E2E tests.

{
  "scripts": {
		...
		"build:coverage": "cross-env CYPRESS_INSTRUMENT_PRODUCTION=true NODE_ENV=production SKIP_PREFLIGHT_CHECK=true REACT_APP_VERSION=$npm_package_version react-scripts -r @cypress/instrument-cra build",
		...
	}
}

In the package.json file, you can define which files within the codebase will be included in the “code coverage” report.

}
	...
	"nyc": {
	  "include": [
	    "src/pages/*"
	  ],
	  "exclude": [
	    "src/pages/Calendar/**/*"
	  ]
	}
}

The @cypress/code-coverage library is used for integration (processing, updating, and storing) of code coverage results during test execution. More information can be found here.

Test Automation and CI/CD Integration

We recommend choosing a CI provider that can scale hardware resources (e.g., Github Actions and so on…) since executing a large number of tests is a demanding operation.

If scaling hardware resources or migrating to another CI provider is not possible, another option is to run tests gradually. This allows running specific tests for different roles.

  • Run specific test: Create a script for each run in the package.json file.
{
	/* spustenie konrétneho test suitu */
	"test:CI:auth": "AUTH_EMAIL=admin@test.com  AUTH_PASSWORD=test SIGN_IN_URL=http://localhost:3000/api/login cypress run --spec cypress/e2e/01-users/auth.cy.ts",
	/* spustenie všetkých test suitov, ktoré sa nachádzajú v priečinku "01-users" */
	"test:CI:users": "AUTH_EMAIL=manager@test.com AUTH_PASSWORD=test SIGN_IN_URL=http://localhost:3000/api/login cypress run --spec cypress/e2e/01-users/*.ts",
}
  • Cypress settings - env. variables: Add all necessary variables to the Cypress configuration file, which can then be used in tests.
...
env: {
		auth_email: process.env.AUTH_EMAIL,
		auth_password: process.env.AUTH_PASSWORD,
		sign_in_url: process.env.SIGN_IN_URL
},
...

The integration can be extended with the Cypress cloud option. The main advantage is the ability to parallelize test execution and manage test results.

More information about integration can be found in the article: Running Cypress in continuous integration.

Roman Haluška28 Nov 2023