Udemy Playwright: Web Automation Testing From Zero to Hero: Difference between revisions

Revision as of 03:51, 10 June 2024

Link to Udemy: Udemy Playwright: Web Automation Testing From Zero to Hero

Link to GitHub: my Playwright Udemy Course repo

Section 1: Preparation

Playwright vs Cypress

Playwright Pros.	Cypress Pros.
Faster test execution OOTB free parallel execution Multiple languages (JS/TS, Python, Java, C# Multiple Tabs Better iFrames similar to Selenium	Less code - fast to write Better auto-wait mechanism Better documentation Better testrunner (time machine) Dashboard service

Development Environment Configuration

node.js => updated => done
Git => updated => done
VS Code => updated => done
Playwright extn for VS Code => installed

Clone Test App

From https://github.com/bondar-artem/pw-practice-app
Cloned it in VS Code
npm install --force

--force needed to accept various warnings

npm start
http://localhost:4200/

Section 2: JavaScript Fundamentals

I'm familiar with JavaScript - I'm fast forwarding through this without keeping notes

Section 3: Playwright Hands-On Overview

create new folder & npm init playwright@latest

Ways to Run & Debug

CLI Test Executions

npx playwright test
npx playwright test example.spec.ts --project=chromium --headed
npx playwright test -g "has title"
npx playwright show-report
CI=true npx playwright test

Test Execution with UI - OMG this debug UI is cool!

npx playwright test --ui

Test Execution with trace on

npx playwright test --project=chromium --trace on
npx playwright show-report

=> you can now open the trace from the report (which looks similar to the ui tool above)

trace can be generated from CI/CD pipeline too, and then you can view the results saved in a zip file with a trace viewer

Test Execution with debug

npx playwright test --project=chromium --debug

this opens the Playwright inspector showing the code, debugging controls, and console information

and the browser window

Test execution with VS Code Extension => Test Explorer

Navigate to the test you want to debug

set any breakpoint(s)

VS Code shows the code, debugging controls, and console information

Tests Structure

First Test

In VS Code
Open PW-PRACTICE-APP
run npm init playwright@latest --force

force is needed to avoid errors
package.json updated with new dev dependencies
playwright.config.ts is created as well as other files

delete test-examples folder - it's not needed
delete test/example.spec.ts file - it's not needed
create file firsTest.spec.ts

import {test} from '@playwright/test'
test('the first test', async ({page}) => {
    await page.goto('http://localhost:4200/')
    await page.getByText('Forms').click()
    await page.getByText('Form Layouts').click()
})

Notice the page fixture, it has a lot of useful methods, eg. page.goto('url') and page.getByText('label')

Hooks & Control Flow

tes.describe(' a test suite'...)
test.beforeEach() and test.beforeAll()

can be used outside as well as inside a suite

.only(..) can be used on tests as well as suites
test.afterEach() and test.afterAll()

try to avoid using the after... hooks, better to do it in the before... hooks

Section 4: Interaction with Web Elements

Understanding DOM and Terminology - Review HTML terms

<parent>
   <html_tag_name html-attribute="a value" class="class1 class2" id="unique">
      <child>
        ... html text value
      </child>
   </html_tag_name>
   <sibling></sibling>
</parent>

Locator Syntax Rules

    page.locator('input') //finds all of them
    page.locator('#inputEmail1') // by id
    page.locator('.shape-rectangle')     //by class value
    page.locator('[placeholder="Email"]')     // by attribute
    page.locator('[class="input-full-width size-medium status-basic shape-rectangle nb-transition cdk-focused cdk-mouse-focused"]')     // by class value (full)
    page.locator('input[placeholder="Email"][nbinput].shape-rectangle')     // combine selectors
    page.locator('//*[@id="inputEmail1"]')     // XPath (NOT Recommended because it's testing implementation rather than user visible aspects)
    page.locator(':text("Using")')  // by partial text match
    page.locator(':text-is("Using the Grid")')     // by exact text match

Note:

If you had previously run the test, and the associated browser window is still open, then when the cursor is on a code line with page.locator(...) it highlights the elements selected by the locator (very cool!)
It's blue when a single element is selected, and orange when multiple elements match the locator.

page.locator(...) will always return all matching elements, can use .first() to refine to first element to perform an action.

use npx playwright test --ui and click the watch icon so that test auto-re-runs when you edit the code (I love this)

User Facing Locators

Playwright Best Practices

Test user-visible behaviour
- tests should typically only see/interact with rendered output
- mimic user behaviour

page.getByRole(...)
- See ARIA roles and attributes (google it?)

  await page.getByRole("textbox", { name: "Email" }).first().click()
  await page.getByRole("button", { name: "Sign in" }).first().click()
  await page.getByLabel("Email").first().click()
  await page.getByPlaceholder('Jane Doe').click()
  await page.getByText('Using the Grid').click()
  await page.getByTestId('SignIn').click() // this expects html attribute=>  data-testid="SignIn"
  await page.getByTitle('IoT Dashboard').click()

Child Elements

test("Locating child elements", async ({ page }) => {
  await page.locator('nb-card nb-radio :text-is("Option 1")').click()
  await page.locator('nb-card').locator('nb-radio').locator(':text-is("Option 2")').click() // this is nicer than line above
  await page.locator('nb-card').getByRole('button', {name: "Sign In"}).first().click() // avoid first last, and nth because the lists enveriably change
  await page.locator('nb-card').nth(3).getByRole('button').click()
})

Parent Elements

test("Locating parent elements", async ({ page }) => {
  await page
    .locator("nb-card", { hasText: "Using the Grid" })
    .getByRole("textbox", { name: "Email" })
    .click()
  await page
    .locator("nb-card", { has: page.locator("#inputEmail1") })
    .getByRole("textbox", { name: "Email" })
    .click()
  await page
    .locator("nb-card")
    .filter({ hasText: "Basic Form" })
    .getByRole("textbox", { name: "Email" })
    .click()
  await page
    .locator("nb-card")
    .filter({ has: page.locator("nb-checkbox") })
    .filter({ hasText: "Sign in" })
    .getByRole("textbox", { name: "Email" })
    .click()
  await page
    .locator(':text-is("Using the Grid")')
    .locator("..")
    .getByRole("textbox", { name: "Email" })
    .click()

  await page
    .getByText("Using the Grid")
    .locator("..")
    .getByRole("textbox", { name: "Email" })
    .click()
})

I think I like best the last one, where you select something that looks like a heading to the user, and then go to the parent that contains the selected element, and the element you want to locate.

Reusing Locators

Stop copying and pasting code ....

  const testEmailAddress = "test@test.com"
  const basicForm = page.locator("nb-card").filter({ hasText: "Basic Form" })
  const emailField = basicForm.getByRole("textbox", { name: "Email" })


  await emailField.fill(testEmailAddress)
  await basicForm.getByRole("textbox", { name: "Password" }).fill("Welcome123")
  await basicForm.getByRole("button").click()

  await expect(emailField).toHaveValue(testEmailAddress)

Extracting Values

  // Single test value
  const basicForm = page.locator("nb-card").filter({ hasText: "Basic Form" })
  const buttonText = await basicForm.locator('button').textContent()
  expect(buttonText).toEqual("Submit")

  // Array of text values
  const allRadioButtonLabels = await page.locator('nb-radio').allTextContents()
  expect(allRadioButtonLabels).toContain('Option 1')

   // input value
   const emailField = basicForm.getByRole('textbox', {name: 'Email'})
   await emailField.fill('test@test.com')
   const emailValue = await emailField.inputValue()
   expect(emailValue).toEqual('test@test.com')

   // attribute
   const placeholderValue = await emailField.getAttribute('placeholder')
   expect(placeholderValue).toEqual('Email')

Assertions

  // General assertions
  const value = 5
  expect(value).toEqual(5)

  const basicFormButton = page
    .locator("nb-card")
    .filter({ hasText: "Basic Form" })
    .locator("button")

  const text = await basicFormButton.textContent()
  expect(text).toEqual("Submit")

  // locator assertion
  await expect(basicFormButton).toHaveText("Submit")

  // soft assertion (continues even if it fails)
  await expect.soft(basicFormButton).toHaveText("Submit")
  await basicFormButton.click()

Auto-waiting

Playwright docs

There are some different wait topologies

some auto-wait for 30s by default (but configurable

expect() waits for only 5s

some do not wait at all

and there's a bunch of alternative waitFor methods, eg. wait for a response to an API call

test.beforeEach(async ({ page }) => {
  await page.goto("http://uitestingplayground.com/ajax")
  await page.getByText("Button Triggering AJAX Request").click()
})

test("Auto Waiting", async ({ page }) => {
  const successButton = page.locator(".bg-success")

  //   const text = await successButton.textContent()  // waits automatically (upto default 30s)
  //   expect(text).toEqual('Data loaded with AJAX get request.')

  //   await successButton.waitFor({state: 'attached'}) // waits upto default 30s
  //   const text2 = await successButton.allTextContents() // fails doesn't wait by itself (must use waitFor(...))
  //   expect(text2).toContain('Data loaded with AJAX get request.')

  await expect(successButton).toHaveText("Data loaded with AJAX get request.", {
    timeout: 20000,
  }) // waits default  upto 5s, unless overridden
})

test("alternative waits", async ({ page }) => {
  const successButton = page.locator(".bg-success")

  // wait for element
  //await page.waitForSelector(".bg-success")

  // wait for particular response
  //await page.waitForResponse('http://uitestingplayground.com/ajaxdata')

  // wait for network calls to be completed (NOT RECOMMENDED)
  await page.waitForLoadState("networkidle")

  const text2 = await successButton.allTextContents()
  expect(text2).toContain("Data loaded with AJAX get request.")
})

Timeouts

test("timeouts", async ({ page }) => {
  //test.setTimeout(10000)
  test.slow()
  const successButton = page.locator(".bg-success")
  await successButton.click()
})

There are three layers to timeouts

1. Global Timeout

time limit for whole test suite to run

default: no limit

2. Test Timeout

Within the global timeout, it is the time limit for a single test

default: 30000ms

3. Action, Navigation, Expect

Within the global and test timeout there are the following timeouts

Action = click(), fill(), textContent() etc

default: no limit

Navigation = page.goto(url...)

default: no limit

Expect = locator assertions

default: 5000ms

Note: regular expect assertions execute immediately, only locator exceptions will wait.

You can override the expect timeout await expect(element).toHaveText("some text", { timeout: 20000 })

Overrides inside test case

Use test.slow() to extend timeout to 3x configured value, or test.setTimeout(10000).

Project Timeout settings in `playwright.config.ts`

export default defineConfig({
  timeout: 10000,   // test case max run time
  globalTimeout: 60000,  // entire test run max run time
  expect:{
    timeout: 2000, // sets locator assertion timeout
  },
  /* Shared settings for all the projects below. See https://playwright.dev/docs/api/class-testoptions. */
  use: {
    actionTimeout: 5000,
    navigationTimeout: 5000,
  },
  ...
});

Timeout settings for a whole spec.ts file

test.beforeEach(async ({ page }, testInfo) => {
  await page.goto("http://uitestingplayground.com/ajax")
  await page.getByText("Button Triggering AJAX Request").click()
  testInfo.setTimeout(testInfo.timeout + 2000)
})

Section 5: UI Components

Input Fields

  await usingTheGridEmailInput.fill('test@test.com')
  await usingTheGridEmailInput.clear()
  await usingTheGridEmailInput.pressSequentially('test2@test.com', {delay:500})

  // generic assertion
  const inputValue = await usingTheGridEmailInput.inputValue()
  expect(inputValue).toEqual('test2@test.com')

  // locator assertion
  await expect(usingTheGridEmailInput).toHaveValue('test2@test.com')

Radio Buttons

//await usingTheGridForm.getByLabel('Option 1').check({force:true})
await usingTheGridForm
    .getByRole("radio", { name: "Option 1" })
    .check({ force: true })

const radioStatus = await usingTheGridForm
    .getByRole("radio", { name: "Option 1" })
    .isChecked()
expect(radioStatus).toBeTruthy()

await expect(
    usingTheGridForm.getByRole("radio", { name: "Option 1" })
).toBeChecked()

await usingTheGridForm
    .getByRole("radio", { name: "Option 2" })
    .check({ force: true })

expect(
    await usingTheGridForm
    .getByRole("radio", { name: "Option 1" })
    .isChecked()
).toBeFalsy()
expect(
    await usingTheGridForm
        .getByRole("radio", { name: "Option 2" })
        .isChecked()
).toBeTruthy()

Note: force needed because class="native-input visually-hidden" ie. the element is hidden, and force suspends Playwright's the actionability checks (ie. Visible, Stable, Receives Events, Enabled)

Checkboxes

await page.getByText("Modal & Overlays").click()
await page.locator(".menu-item").getByText("Toastr").click()
await page
  .getByRole("checkbox", { name: "Hide on click" })
  .uncheck({ force: true })
await page
  .getByRole("checkbox", { name: "Prevent arising of duplicate toast" })
  .check({ force: true })

const checkBoxes = page.getByRole("checkbox")
for (const checkbox of await checkBoxes.all()) {
  await checkbox.uncheck({ force: true })
  expect(await checkbox.isChecked()).toBeFalsy()
}

Note: force needed because class="native-input visually-hidden" ie. the element is hidden, and force suspends Playwright's the actionability checks (ie. Visible, Stable, Receives Events, Enabled)

Also I couldn't get const checkBoxes = await page.getByRole("checkbox").all() and then use array forEach(...) to work, not sure why ... but using the for loop is documented in the method tooltip ...

Lists & Dropdowns

test("lists and dropdowns", async ({ page }) => {
  const dropDownMenu = page.locator("ngx-header nb-select")
  await dropDownMenu.click()

  page.getByRole("list") // for <ul> => parent list container
  page.getByRole("listitem") // for <li> => not always used

  // const optionList = page.getByRole('list').locator('nb-option')

  const optionList = page.locator("nb-option-list nb-option")

  await expect(optionList).toHaveText(["Light", "Dark", "Cosmic", "Corporate"])

  await optionList.filter({ hasText: "Cosmic" }).click()

  const header = page.locator("nb-layout-header")
  await expect(header).toHaveCSS("background-color", "rgb(50, 50, 89)")

  const colors = {
    Light: "rgb(255, 255, 255)",
    Dark: "rgb(34, 43, 69)",
    Cosmic: "rgb(50, 50, 89)",
    Corporate: "rgb(255, 255, 255)",
  }

  for (const color in colors) {
    await dropDownMenu.click()
    await optionList.filter({ hasText: color }).click()
    await expect(header).toHaveCSS("background-color", colors[color])
  }
})

Tooltips

To see the tooltip you may need to go to the dev tools, sources tab, and press F8 (on windows) to pause the debugger whilst showing the tooltip. This will now allow you to investigate the frozen DOM.

test("tooltip", async ({ page }) => {
  await page.getByText("Modal & Overlays").click()
  await page.locator(".menu-item").getByText("Tooltip").click()
  const toolTipCard = page.locator("nb-card", { hasText: "Tooltip Placements" })
  await toolTipCard.locator("button", { hasText: "TOP" }).hover()

  page.getByRole('tooltip')  // only works if the tooltip role was added to the element
  const tooltip = await page.locator('nb-tooltip').textContent()
  expect(tooltip).toEqual('This is a tooltip')
})

Dailog Boxes

test("Dialog box", async ({ page }) => {
  await page.getByText("Tables & Data").click()
  await page.locator(".menu-item").getByText("Smart Table").click()

  page.on("dialog", (dialog) => {
    expect(dialog.message()).toEqual("Are you sure you want to delete?")
    dialog.accept()
  })

  const email = "mdo@gmail.com"
  await page
    .getByRole("table")
    .locator("tr", { hasText: email })
    .locator(".nb-trash")
    .click()

  await expect(page.locator("table tr").first()).not.toHaveText(email)
})

Web Tables

test("Web Table", async ({ page }) => {
  await page.getByText("Tables & Data").click()
  await page.locator(".menu-item").getByText("Smart Table").click()

  // 1 locate a row by a unique value
  const targetRow = page.getByRole("row", { name: "twitter@outlook.com" })
  await targetRow.locator(".nb-edit").click()
  await page.locator("input-editor").getByPlaceholder("Age").clear()
  await page.locator("input-editor").getByPlaceholder("Age").fill("35")
  await page.locator(".nb-checkmark").click()

  // 2 get a row by a value in specific column
  await page.locator(".ng2-smart-pagination-nav").getByText("2").click()
  const targetRowById = page.getByRole('row', {name:'11'}).filter({has: page.locator('td').nth(1).getByText('11')})
  await targetRowById.locator('.nb-edit').click()
  await page.locator("input-editor").getByPlaceholder("E-mail").clear()
  await page.locator("input-editor").getByPlaceholder("E-Mail").fill("test@test.com")
  await page.locator(".nb-checkmark").click()
  await expect(targetRowById.locator('td').nth(5)).toHaveText('test@test.com')

  // 3 test filter of the  table
  const ages = ["20", "30", "40", "200"]

  for (let age of ages) {
    await page.locator("input-filter").getByPlaceholder("Age").clear()
    await page.locator("input-filter").getByPlaceholder("Age").fill(age)
    await page.waitForTimeout(500) // it takes a moment to refresh the list

    const ageRows = page.locator("tbody tr")

    for (let row of await ageRows.all()) {
      const cellValue = await row.locator("td").last().textContent()
      if (age == "200") {
        expect(await page.getByRole("table").textContent()).toContain(
          "No data found"
        )
      } else {
        expect(cellValue).toEqual(age)
      }
    }
  }
})

Not sure I like how one scenario is testing multiple things, and particularly // 3 where the last filter scenario expects to find no data, compared to the first three that do have rows...

Date Picker

test("Date Picker", async ({ page }) => {
  await page.getByText("Forms").click()
  await page.locator(".menu-item").getByText("Datepicker").click()

  const calendarInputfield = page.getByPlaceholder("Form Picker")
  await calendarInputfield.click()

  let date = new Date()
  date.setDate(date.getDate() + 500) 
  const expectedDate = date.getDate().toString()
  const expectedMonthShort = date.toLocaleString("En-US", { month: "short" })
  const expectedMonthLong = date.toLocaleString("En-US", { month: "long" })
  const expectedYear = date.getFullYear()
  const formattedDate = `${expectedMonthShort} ${expectedDate}, ${expectedYear}`

  let calendarMonthAndYear = await page
    .locator("nb-calendar-view-mode")
    .textContent()
  const expectedMonthAndYear = ` ${expectedMonthLong} ${expectedYear} `

  while (!calendarMonthAndYear.includes(expectedMonthAndYear)) {
    await page.locator(".next-month").click()
    //    await page.locator('nb-calendar-pageable-navigation [data-name="chevron-right"]').click()
    calendarMonthAndYear = await page
      .locator("nb-calendar-view-mode")
      .textContent()
  }

  await page
    .locator('[class="day-cell ng-star-inserted"]')
    .getByText(expectedDate, { exact: true })
    .click()

  await expect(calendarInputfield).toHaveValue(formattedDate)
})

Notes:

be careful when adding to a date, it may wrap to the next month
things I don't quite like here
- all the date string variables, and using the expected prefixes
- this check "sometimes" checks the next month button. but not always - these should be two separate checks (test cases)
- the code duplication for calendarMonthAndYear = ...
- the course uses 'nb-calendar-pageable-navigation [data-name="chevron-right"]' when ".next-month" is cleaner and more readable
- the while loop is dangerous, it assumes that calendarMonthAndYear.includes(expectedMonthAndYear) will be true at some point.

Consider what would happen if the format of the displayed text is changed (eg. the developer changes it to use a short month...)

I think a for loop with conditional exit break would be better,

and using two separate checks, for short month, and then year, would be more resilient.

With suggested changes

test("Date Picker", async ({ page }) => {
  await page.getByText("Forms").click()
  await page.locator(".menu-item").getByText("Datepicker").click()

  const calendarInputfield = page.getByPlaceholder("Form Picker")
  await calendarInputfield.click()

  let selectDate = new Date()
  const addDays = 500
  selectDate.setDate(selectDate.getDate() + addDays)
  const selectDayOfMonth = selectDate.getDate().toString() // value of "1" to "31"
  const selectMonth = selectDate.toLocaleString("En-US", { month: "short" })
  const selectYear = selectDate.getFullYear().toString()
  const formattedSelectDate = `${selectMonth} ${selectDayOfMonth}, ${selectYear}`

  for (let i = 0; i < addDays / 28; i++) { // divide by shortest month
    let calendarMonthAndYear = await page
      .locator("nb-calendar-view-mode")
      .textContent()

    if (
      calendarMonthAndYear.includes(selectMonth) &&
      calendarMonthAndYear.includes(selectYear)
    ) {
      break
    } else {
      await page.locator(".next-month").click()
    }
  }

  await page
    .locator('[class="day-cell ng-star-inserted"]') 
    .getByText(selectDayOfMonth, { exact: true })
    .click()
    // locator has to be exact match, partial matches would include days from prev and next months

  await expect(calendarInputfield).toHaveValue(formattedSelectDate)
})

Sliders

test("Sliders", async ({ page }) => {
  // Update attribute
  // const tempGaugeDraggerHandle = page.locator('[tabtitle="Temperature"] ngx-temperature-dragger circle')
  // await tempGaugeDraggerHandle.evaluate(node => {
  //     node.setAttribute('cx',"232.103")
  //     node.setAttribute('cy',"232.103")
  // })
  // await tempGaugeDraggerHandle.click()

  // mouse movement
  const tempGauge = page.locator(
    '[tabtitle="Temperature"] ngx-temperature-dragger'
  )

  tempGauge.scrollIntoViewIfNeeded()
  // probably also need to make sure that the bowser window is big enough to show the whole UI control

  const box = await tempGauge.boundingBox()

  const c = {
    x: box.x + box.width / 2,
    y: box.y + box.height / 2,
  }

  await page.mouse.move(c.x, c.y)
  await page.mouse.down()
  await page.mouse.move(c.x + 100, c.y)
  await page.mouse.move(c.x + 100, c.y + 100)
  await page.mouse.up()

  await page.mouse.click(c.x + 100, c.y + 100)
  // this was needed to get the scenario to run in Playwright UI mode viewer

  await expect(tempGauge).toContainText("30")
})

Notes:

Had some issues with browser window being too small and interfering with scrolling and mouse actions
also the UI Mode view doesn't show the temperature UI component correctly, and didn't seem to update the value shown. I added a page.mouse.click(...location...) at the end to fix this.

Drag & Drop with iFrames

test("drag and drop with iframe", async ({ page }) => {
  await page.goto("https://www.globalsqa.com/demo-site/draganddrop/")
  const frame = page.frameLocator('[rel-title="Photo Manager"] iframe')

  await frame
    .locator("li", { hasText: "High Tatras 2" })
    .dragTo(frame.locator("#trash"))

  // more precise control
  await frame.locator("li", { hasText: "High Tatras 4" }).hover()
  await page.mouse.down()
  await frame.locator("#trash").hover()
  await page.mouse.up()

  await expect(frame.locator("#trash li h5")).toHaveText([
    "High Tatras 2",
    "High Tatras 4",
  ])
})

Section 6: Page Object Model

Intro

design pattern to improve maintainability and reusability
there's no industry standard way to do it though
Core concepts
- each page has a class
- with methods for operations
Two Important Principles
- DRY - don't repeat yourself
- KISS - keep it simple stupid
Two Good Practices
- Descriptive naming
- ~~Avoid tiny methods~~ => I don't agree, especially if this leads to code repetition.

Note:

For me I like POM's to encapsulate the page (or component) so that the object knows

how to find things it contains
how to do things that it implements

with the end result being that code consuming the object interacts with it in a way that feels and reads like a real user might.

First Page Object

new file page-objects\navigationPage.ts

import { Page } from "@playwright/test"

export class NavigationPage {
  readonly page: Page

  constructor(page: Page) {
    this.page = page
  }

  async formLayoutsPage() {
    await this.page.getByText("Forms").click()
    await this.page.getByText("Form Layouts").click()
  }
}

new file tests\usePageObjects.spec.ts

import { expect, test } from "@playwright/test"
import { NavigationPage } from "../page-objects/navigationPage"

test.beforeEach(async ({ page }, testInfo) => {
  await page.goto("http://localhost:4200/")
})

test("navigate to form page", async ({ page }) => {
  const navigateTo = new NavigationPage(page)
  await navigateTo.formLayoutsPage()

})

Notes

I don't like a lot of the naming here,

navigationPage -> navigationMenu
formLayoutsPage -> clickFormLayouts
navigateTo -> navigationMenu

Navigation Page Object

As per above section, but adding/modifying

  async datePickerPage() {
    await this.selectGroupMenuItem("Forms")
    await this.page.locator(".menu-item").getByText("Datepicker").click()
  }

  /* other menu items */

  private async selectGroupMenuItem(groupItemtitle: string) {
    const groupMenuItem = this.page.getByTitle(groupItemtitle)
    const expandedState = await groupMenuItem.getAttribute("aria-expanded")

    if (expandedState == "false") {
      await groupMenuItem.click()
    }
  }

Locators in Page Objects

Modifying the previous example


  readonly formLayoutsMenuItem: Locator
  readonly datePickerMenuItem: Locator
  ...

  constructor(page: Page) {
    this.page = page
    this.formLayoutsMenuItem = page.getByText("Form Layouts")
    this.datePickerMenuItem = page.locator(".menu-item").getByText("Datepicker")
    ...
  }

  async formLayoutsPage() {
    await this.selectGroupMenuItem("Forms")
    await this.formLayoutsMenuItem.click()
  }
  
  ...

Presenter pointed out that you can end up with long lists of fields, and constructor lines, resulting in hard to review and maintain code. You're probably better off just leaving them hard coded in the action methods.

Parameterised Methods

File: tests\usepageObjects.spec.ts

import { expect, test } from "@playwright/test"
import { NavigationPage } from "../page-objects/navigationPage"
import { FormLayoutsPage } from "../page-objects/formLayoutsPage"

test.beforeEach(async ({ page }, testInfo) => {
  await page.goto("http://localhost:4200/")
})

test("navigate to all pages", async ({ page }) => {
  const navigateTo = new NavigationPage(page)
  await navigateTo.formLayoutsPage()
  await navigateTo.datePickerPage()
  await navigateTo.smartTablepage()
  await navigateTo.toastrPage()
  await navigateTo.tooltipPage()
})

test("parameterised methods", async ({ page }) => {
  const navigateTo = new NavigationPage(page)
  const onFormLayoutsPage = new FormLayoutsPage(page)

  await navigateTo.formLayoutsPage()
  await onFormLayoutsPage.submitUsingTheGridFormWithCredentialsAndSelectOption(
    "test@test.com",
    "Welcome1",
    "Option 2"
  )
  await onFormLayoutsPage.submitInLineFormWithNameEmailAndCheckbox(
    "John Smith",
    "john@test.com",
    true
  )
})

File: page-objects\formLayoutsPage.ts

import { Page } from "@playwright/test"

export class FormLayoutsPage {
  private readonly page: Page
  constructor(page: Page) {
    this.page = page
  }

  async submitUsingTheGridFormWithCredentialsAndSelectOption(
    email: string,
    password: string,
    optionText: string
  ) {
    const usingTheGridForm = this.page.locator("nb-card", {
      hasText: "Using the Grid",
    })
    await usingTheGridForm.getByRole("textbox", { name: "Email" }).fill(email)
    await usingTheGridForm
      .getByRole("textbox", { name: "Password" })
      .fill(password)
    await usingTheGridForm
      .getByRole("radio", { name: optionText })
      .check({ force: true })

    await usingTheGridForm.getByRole("button").click()
  }

  /**
   * This method fills out the inline form with user details
   * 
   * @param name - first and last name
   * @param email - valid email for the test user
   * @param rememberMe - if the user session is to be saved
   */
  async submitInLineFormWithNameEmailAndCheckbox(
    name: string,
    email: string,
    rememberMe: boolean
  ) {
    const inlineForm = this.page.locator("nb-card", {
      hasText: "Inline form",
    })
    await inlineForm.getByRole("textbox", { name: "Jane Doe" }).fill(name)
    await inlineForm.getByRole("textbox", { name: "Email" }).fill(email)

    rememberMe &&
      (await inlineForm.getByRole("checkbox").check({ force: true }))

    await inlineForm.getByRole("button").click()
  }
}

Date Picker Page Object

test("DatePickerPage Object", async ({page}) => {
  const navigateTo = new NavigationPage(page)
  const onDatepickerPage = new DatePickerPage(page)

  await navigateTo.datePickerPage()
  await onDatepickerPage.selectCommonDatePickerDateFromToday(10)
  await onDatepickerPage.selectDatePickerWithRangeFromToday(6 ,15)
})

File: page-objects\datePickerPage.ts

import { Page, expect } from "@playwright/test"

export class DatePickerPage {
  private readonly page: Page

  constructor(page: Page) {
    this.page = page
  }

  async selectCommonDatePickerDateFromToday(numberOfDaysFromToday: number) {
    const calendarInputfield = this.page.getByPlaceholder("Form Picker")
    await calendarInputfield.click()

    const formattedSelectDate = await this.selectDateInTheCalendar(
      numberOfDaysFromToday
    )

    await expect(calendarInputfield).toHaveValue(formattedSelectDate)
  }

  async selectDatePickerWithRangeFromToday(
    startDayFromToday: number,
    endDayFromToday: number
  ) {
    if(startDayFromToday > endDayFromToday) throw new Error('startDayFromToday must not be after endDayFromToday')

    const calendarInputfield = this.page.getByPlaceholder("Range Picker")
    await calendarInputfield.click()

    const formattedStartDate = await this.selectDateInTheCalendar(
      startDayFromToday
    )
    const formattedEndDate = await this.selectDateInTheCalendar(endDayFromToday)

    await expect(calendarInputfield).toHaveValue(`${formattedStartDate} - ${formattedEndDate}`)
  }

  private async selectDateInTheCalendar(numberOfDaysFromToday: number) {
    let selectDate = new Date()
    selectDate.setDate(selectDate.getDate() + numberOfDaysFromToday)
    const selectDayOfMonth = selectDate.getDate().toString() // value of "1" to "31"
    const selectMonth = selectDate.toLocaleString("En-US", { month: "short" })
    const selectYear = selectDate.getFullYear().toString()
    const formattedSelectDate = `${selectMonth} ${selectDayOfMonth}, ${selectYear}`

    // check if selectDate is in this month, or whether we need to click next to a future month
    for (let i = 0; i < numberOfDaysFromToday / 28; i++) {
      // divide by shortest month
      let calendarMonthAndYear = await this.page
        .locator("nb-calendar-view-mode")
        .textContent()

      if (
        calendarMonthAndYear.includes(selectMonth) &&
        calendarMonthAndYear.includes(selectYear)
      ) {
        break
      } else {
        await this.page.locator(".next-month").click()
      }
    }

    await this.page
      .locator('.day-cell.ng-star-inserted:not(.bounding-month)')
    .getByText(selectDayOfMonth, { exact: true })
      .click()
    // locator has to be exact match, partial matches would include days from prev and next months

    return formattedSelectDate
  }
}

Notes:

I feel that the page object should not have expect(...) in it, IMHO these ought to be in the tests themselves, with possibly helper methods in the page object to obtain the values to assert on.
I added a little input validation to check startDayFromToday is not be after endDayFromToday
Not sure I like a few of the names used, e.g. method name selectDatePickerWithRangeFromToday(...) is not entirely self evident what it is, and I'd probably just want to use Date object for start and end input parameters, rather than the number of days from today.

Page Objects Manager

File: page-objects\pageManager.ts

import { Page, expect } from "@playwright/test"
import { NavigationPage } from "../page-objects/navigationPage"
import { FormLayoutsPage } from "../page-objects/formLayoutsPage"
import { DatePickerPage } from "../page-objects/datePickerPage"

export class PageManager {
  private readonly page: Page
  private readonly navigationPage: NavigationPage
  private readonly formLayoutsPage: FormLayoutsPage
  private readonly datePickerPage: DatePickerPage

  constructor(page: Page) {
    this.page = page
    this.navigationPage = new NavigationPage(this.page)
    this.formLayoutsPage = new FormLayoutsPage(this.page)
    this.datePickerPage = new DatePickerPage(this.page)
  }

  navigateTo = () => this.navigationPage
  onFormLayoutsPage = () => this.formLayoutsPage
  onDatePickerPage = () => this.datePickerPage
}

Notes:

I swapped to using arrow notation for specifying the PageManager's methods. (Make it a lot more concise and has less verbiage)

test("Page Manger", async ({ page }) => {
  const pm = new PageManager(page)

  await pm.navigateTo().formLayoutsPage()
  await pm.onFormLayoutsPage().submitUsingTheGridFormWithCredentialsAndSelectOption(
    "test@test.com",
    "Welcome1",
    "Option 2"
  )
  await pm.onFormLayoutsPage().submitInLineFormWithNameEmailAndCheckbox(
    "John Smith",
    "john@test.com",
    true
  )

  await pm.navigateTo().datePickerPage()
  await pm.onDatePickerPage().selectCommonDatePickerDateFromToday(10)
  await pm.onDatePickerPage().selectDatePickerWithRangeFromToday(6 ,15)
})

Page Objects Helper Base

File page-objects\helperBase.ts

import { Page } from "@playwright/test"

export class HelperBase {
  readonly page: Page

  constructor(page: Page) {
    this.page = page
  }

  async waitForNumberOfSeconds(timeInSeconds: number) {
    await this.page.waitForTimeout(timeInSeconds * 1000)
  }
}

File page-objects\navigationPage.ts

...
async formLayoutsPage() {
    await this.selectGroupMenuItem("Forms")
    await this.page.getByText("Form Layouts").click()
    await this.waitForNumberOfSeconds(2)
  }
...

Which is used in tests\usePageObjects.spec.ts

test("Page Manger", async ({ page }) => {
  const pm = new PageManager(page)

  await pm.navigateTo().formLayoutsPage()
  await pm
    .onFormLayoutsPage()
    .submitUsingTheGridFormWithCredentialsAndSelectOption(
      "test@test.com",
      "Welcome1",
      "Option 2"
    )
  await pm
    .onFormLayoutsPage()
    .submitInLineFormWithNameEmailAndCheckbox(
      "John Smith",
      "john@test.com",
      true
    )

  await pm.navigateTo().datePickerPage()
  await pm.onDatePickerPage().selectCommonDatePickerDateFromToday(10)
  await pm.onDatePickerPage().selectDatePickerWithRangeFromToday(6, 15)
})

Section 7: Working With APIs

What is an API?

I'm reasonably familiar with API's. Just making some cursory notes for this lesson.

Methods

GET
POST
PUT
DELETE

Parts

URL
Headers
Method
Body

Response Status Codes

200's - Success
300's - Redirection
400's - Client Error
500's - Server Error

Playwright API Mocking

powerful
fast
risky when API contracts are changing

Setup New Project

goto https://conduit.bondaracademy.com/
- open dev tools to observe network traffic
- create account
- create an article
- delete the article
download & install Postman

VS Code
- Create new project folder ~/Repositories/PlaywrightUdemyCourse/pw-APITest-app
- open new terminal and run npm init playwright@latest
- delete test-examples folder
- rename tests\example.spec.ts to tests\workingWithAPI.spec.ts

Mocking an API Response

Goto https://conduit.bondaracademy.com/
check network tab in dev tools
mocking the tags api call

file tests\workingWithAPI.spec.ts

import { test, expect } from "@playwright/test"
import ressponseBody from "../test-data/tags.json"

test.beforeEach(async ({ page }) => {
  await page.route(
    "*/**/api/tags",
    async (route) => {
      await route.fulfill({
        body: JSON.stringify(ressponseBody),
      })
    }
  )

  await page.goto("https://conduit.bondaracademy.com/")
})

test("has title", async ({ page }) => {
  await expect(page.locator(".navbar-brand")).toHaveText("conduit")
  await expect(
    page.locator(".tag-default.tag-pill", { hasText: "Playwright" })
  ).toBeVisible()
})

file test-data\tags.json

{
  "tags": ["Automation", "Playwright"]
}

Modifying an API Response

Add the following code the the previous test.beforeEach(...) call

  await page.route("*/**/api/articles*", async (route) => {
    const response = await route.fetch()
    const responseBody = await response.json()
    responseBody.articles[0].title = "This is a test title"
    responseBody.articles[0].description = "This is a description"

    await route.fulfill({ body: JSON.stringify(responseBody) })
  })

Performing API Requests

To test the delete article function we need to first use the API to

Login to obtain access token
create an article

And then we can use web ui to

Login
go to Global Feed tab
find & click the article we created
delete the article
refresh the Global Feed tab
Confirm the article is no longer listed

test("Delete Article", async ({ page, request }) => {
  const response = await request.post(
    "https://conduit-api.bondaracademy.com/api/users/login",
    {
      data: {
        user: {
          email: "conduit@dirksonline.net",
          password: "qB85R86#ZMKME$jVEVq#vJMDr*A!cJk",
        },
      },
    }
  )
  const responseBody = await response.json()
  const accessToken = responseBody.user.token

  const articleResponse = await request.post(
    "https://conduit-api.bondaracademy.com/api/articles/",
    {
      data: {
        article: {
          title: "Test Title",
          description: "Test description",
          body: "Test body",
          tagList: [],
        },
      },
      headers: {
        Authorization: `Token ${accessToken}`,
      },
    }
  )
  expect(articleResponse.status()).toEqual(201)

  await page.getByText('Global Feed').click()
  await page.getByText('Test Title').click()
  await page.getByRole('button',{name:'Delete Article'}).first().click()
  await page.getByText('Global Feed').click()
  await expect(page.locator("app-article-list h1").first()).not.toContainText(
    "Test Title"
  )

})

Intercept Browser API Response

test("create article", async ({ page, request }) => {
  // Use the web ui to create the article
  await page.getByText("New Article").click()
  await page
    .getByRole("textbox", { name: "Article Title" })
    .fill("Playwright is awesome")
  await page
    .getByRole("textbox", { name: "What's this article about?" })
    .fill("About Playwright")
  await page
    .getByRole("textbox", { name: "Write your article (in markdown)" })
    .fill("We like to use Playwright for automation")
  await page.getByRole("button", { name: "Publish Article" }).click()

  // intercept the API response to extract the slug which is needed for clean up later
  const articleResponse = await page.waitForResponse(
    "https://conduit-api.bondaracademy.com/api/articles/"
  )
  const articleresponseBody = await articleResponse.json()
  const slugID = articleresponseBody.article.slug

  // Assert that article was created
  await expect(page.locator(".article-page h1")).toContainText(
    "Playwright is awesome"
  )

  // Go to global feed
  await page.getByText("Home").click()
  await page.getByText("Global Feed").click()

  // Assert new article is listed
  await expect(page.locator("app-article-list h1").first()).toContainText(
    "Playwright is awesome"
  )

  // Clean up
  // Obtain access token for API call
  const response = await request.post(
    "https://conduit-api.bondaracademy.com/api/users/login",
    {
      data: {
        user: {
          email: "conduit@dirksonline.net",
          password: "qB85R86#ZMKME$jVEVq#vJMDr*A!cJk",
        },
      },
    }
  )
  const responseBody = await response.json()
  const accessToken = responseBody.user.token

  // delete the article using the slug extracted earlier
  const articleDeleteResponse = await request.delete(
    `https://conduit-api.bondaracademy.com/api/articles/${slugID}`,
    {
      headers: {
        Authorization: `Token ${accessToken}`,
      },
    }
  )
  expect(articleDeleteResponse.status()).toEqual(204)
})

Notes:

should really only login once and re-use the access token
should be refactored using page objects to improve readability

Sharing Authentication State

Create new folder .auth

Create new file tests\auth.setup.ts

import { test as setup } from "@playwright/test"

const authfile = ".auth/user.json"

setup("authentication", async ({ page }) => {
  await page.goto("https://conduit.bondaracademy.com/")
  await page.getByText("Sign in").click()
  await page
    .getByRole("textbox", { name: "Email" })
    .fill("conduit@dirksonline.net")
  await page
    .getByRole("textbox", { name: "Password" })
    .fill("qB85R86#ZMKME$jVEVq#vJMDr*A!cJk")
  await page.getByRole("button").click()

  await page.waitForResponse('https://conduit-api.bondaracademy.com/api/tags')
                              
  await page.context().storageState({ path: authfile })
})

In fileplaywright.config.ts update projects

  /* Configure projects for major browsers */
  projects: [
    { name: "setup", testMatch: "auth.setup.ts" },
    {
      name: "chromium",
      use: { ...devices["Desktop Chrome"], storageState: ".auth/user.json" },
      dependencies: ["setup"],
    },
    {
      name: "firefox",
      use: { ...devices["Desktop Firefox"], storageState: ".auth/user.json" },
      dependencies: ["setup"],
    },

    {
      name: "webkit",
      use: { ...devices["Desktop Safari"], storageState: ".auth/user.json" },
      dependencies: ["setup"],
    },

Delete the login steps from test.beforeEach(...), but leave the page.goto(...)

Notes:

this all seems a little spooky, I'd prefer something in a beforeXXX(...) hook that saves something that is recalled .... will need to look a bit deeper into the page.context().storageState() stuff
something is broken in v1.44.0 and needed to down grade to 1.43.0 (
- update package.json "@playwright/test": "1.43.0"
- delete package-lock.json
- run npm install
- run npx playwright install

API Authentication

File tests/auth.setup.ts change to use API and save to file .auth\user.json env variable ACCESS_TOKEN

setup("authentication", async ({ request }) => {
  const response = await request.post(
    "https://conduit-api.bondaracademy.com/api/users/login",
    {
      data: {
        user: {
          email: "conduit@dirksonline.net",
          password: "qB85R86#ZMKME$jVEVq#vJMDr*A!cJk",
        },
      },
    }
  )
  const responseBody = await response.json()
  const accessToken = responseBody.user.token
  user.origins[0].localStorage[0].value = accessToken
  fs.writeFileSync(authfile, JSON.stringify(user))
  process.env['ACCESS_TOKEN'] = accessToken

File tests/workingWithAPI.spec.ts remove all code for obtaining access token, and for specifying the access token header

File playwright.config.ts add extraHTTPHeaders node

export default defineConfig({
  ...  
  use: {
    ...
    extraHTTPHeaders: {
      'Authorization': `Token ${process.env.ACCESS_TOKEN}`
    }

Notes:

tried playwright@1.44.1 which still gives issues with the setup settings/file etc. continued using 1.43.0
This relies on .auth\user.json existing with the correct format, but it is in .gitignore so a clean clone of repo will not work
I do like how this now removes a lot of boiler code from test cases.
I wonder, given that the auth header is now always included via the env variable, is the .auth\user.json still needed? (Might also need removing storageState values from the playwright.config.ts

Section 8: Advanced

npm Scripts and CLI Commands

npx playwright test usePageObjects.spec.ts --project=chromium
npx playwright show-report
npx playwright test usePageObjects.spec.ts --project=firefox

In file: package.json add these to scripts node

"pageObjects-chrome": "npx playwright test usePageObjects.spec.ts --project=chromium",
"pageObjects-firefox": "npx playwright test usePageObjects.spec.ts --project=firefox",
"pageObjects-all": "npm run pageObjects-chrome & npm run pageObjects-firefox"

npm run pageObjects-chrome
npm run pageObjects-firefox
npm run pageObjects-all

Notes:

&& runs sequentially
& runs parallely
On windows npm scripts are executed in cmd.exe by default, use this to change it to bash npm config set script-shell "C:/Program Files/Git/bin/bash.exe"

Test Data Generator

Install faker with npm i @faker-js/faker --save-dev --force (force because it has several CVE's)

Update test in tests\usePageObjects.spec.ts

test("Page Manger", async ({ page }) => {
  const pm = new PageManager(page)
  const randomFullName = faker.person.fullName({
    // sex: "male",       // specify sex (optional)
    // lastName: "Johns", // specify last name (optional)
  })
  const randomEmail = `${randomFullName.replace(/ /g, "")}${faker.number.int(
    1000
  )}@test.com`

  await pm.navigateTo().formLayoutsPage()
  await pm
    .onFormLayoutsPage()
    .submitUsingTheGridFormWithCredentialsAndSelectOption(
      "test@test.com",
      "Welcome1",
      "Option 2"
    )
  await pm
    .onFormLayoutsPage()
    .submitInLineFormWithNameEmailAndCheckbox(randomFullName, randomEmail, true)

  await pm.navigateTo().datePickerPage()
  await pm.onDatePickerPage().selectCommonDatePickerDateFromToday(10)
  await pm.onDatePickerPage().selectDatePickerWithRangeFromToday(6, 15)
})

Notes:

The faker fullName() method can sometimes add title prefixes, suffixes, etc., which look very odd in email addresses ....

Test Retries

workers are clean incognito browser windows
worker is re-used after completing a passing test
a new worker is started after a failed test
with retries on the failed test is retried in the new worker
with retries off the next test is started in the new worker

In file playwright.config.ts

export default defineConfig({
  ...
  /* Retry on CI only */
  retries: process.env.CI ? 2 : 1,
  use: {
    ...
    /* Collect trace when retrying the failed test. See https://playwright.dev/docs/trace-viewer */
    trace: "on-first-retry", //['on-first-retry', 'on-all-retries', 'off', 'on', 'retain-on-failure']
  },

In file tests\uiComponents.spec.ts

test.describe.only("Forms Layouts page", () => {
  test.describe.configure({ retries: 2,  })

  test.beforeEach(async ({ page }, testInfo) => {
    if(testInfo.retry){
      // can clean up from previous failed attempt
    }

    ...
  })
...

@@ Line 1,472: / Line 1,472: @@
 === Test Retries ===
+* workers are clean incognito browser windows
+* worker is re-used after completing a passing test
+* a new worker is started after a failed test
+* with retries on the failed test is retried in the new worker
+* with retries off the next test is started in the new worker
+In file <code>playwright.config.ts</code>
+ <nowiki>
+export default defineConfig({
+  ...
+  /* Retry on CI only */
+  retries: process.env.CI ? 2 : 1,
+  use: {
+    ...
+    /* Collect trace when retrying the failed test. See https://playwright.dev/docs/trace-viewer */
+    trace: "on-first-retry", //['on-first-retry', 'on-all-retries', 'off', 'on', 'retain-on-failure']
+  },
+</nowiki>
+In file <code>tests\uiComponents.spec.ts</code>
+ <nowiki>
+test.describe.only("Forms Layouts page", () => {
+  test.describe.configure({ retries: 2,  })
+  test.beforeEach(async ({ page }, testInfo) => {
+    if(testInfo.retry){
+      // can clean up from previous failed attempt
+    }
+    ...
+  })
+...
+</nowiki>
+Notes:
+* Having trace on-first-retry is very cool
 === Parallel Execution ===
 === Screenshots and Videos ===

Udemy Playwright: Web Automation Testing From Zero to Hero: Difference between revisions

Revision as of 03:51, 10 June 2024

Section 1: Preparation

Playwright vs Cypress

Development Environment Configuration

Clone Test App

Section 2: JavaScript Fundamentals

Section 3: Playwright Hands-On Overview

Ways to Run & Debug

Tests Structure

First Test

Hooks & Control Flow

Section 4: Interaction with Web Elements

Understanding DOM and Terminology - Review HTML terms

Locator Syntax Rules

User Facing Locators

Child Elements

Parent Elements

Reusing Locators

Extracting Values

Assertions

Auto-waiting

Timeouts

1. Global Timeout

2. Test Timeout

3. Action, Navigation, Expect

Overrides inside test case

Project Timeout settings in playwright.config.ts

Timeout settings for a whole spec.ts file

Section 5: UI Components

Input Fields

Radio Buttons

Checkboxes

Lists & Dropdowns

Tooltips

Dailog Boxes

Web Tables

Date Picker

Sliders

Drag & Drop with iFrames

Section 6: Page Object Model

Intro

First Page Object

Navigation Page Object

Locators in Page Objects

Parameterised Methods

Date Picker Page Object

Page Objects Manager

Page Objects Helper Base

Section 7: Working With APIs

What is an API?

Setup New Project

Mocking an API Response

Modifying an API Response

Performing API Requests

Intercept Browser API Response

Sharing Authentication State

API Authentication

Section 8: Advanced

npm Scripts and CLI Commands

Test Data Generator

Test Retries

Parallel Execution

Screenshots and Videos

Environment Variables

Configuration File

Fixtures

Project Setup and Teardown

Test Tags

Mobile Device Emulator

Reporting

Visual Testing

Playwright with Docker Container

Github Actions and Argos CI

Navigation menu

Search

Project Timeout settings in `playwright.config.ts`