Skip to content

10 · Project — Automate a Login Flow

This project ties together every Module 1–9 concept into one small but real test file: project setup, first script structure, locators, actions, expect assertions, waiting, forms, and headed/headless execution. It uses https://www.saucedemo.com/, a public demo store built specifically for automation practice, with a known set of test accounts.

Goal

Write a pytest test file that:

  1. Logs in with valid credentials and asserts success.
  2. Logs in with invalid credentials and asserts the specific error shown.
  3. Logs in as a known "locked out" user and asserts that specific error.
  4. Adds an item to the cart after logging in and asserts the cart badge.
  5. Logs out and confirms you're returned to the login page.

Project setup

playwright-practice/
├── pytest.ini
└── tests/
    └── test_login_flow.py

pytest.ini:

[pytest]
testpaths = tests
addopts = --browser chromium

The test file

# tests/test_login_flow.py
from playwright.sync_api import Page, expect

BASE_URL = "https://www.saucedemo.com/"


def login(page: Page, username: str, password: str) -> None:
    page.goto(BASE_URL)
    page.get_by_placeholder("Username").fill(username)
    page.get_by_placeholder("Password").fill(password)
    page.get_by_role("button", name="Login").click()


def test_valid_login_reaches_inventory_page(page: Page):
    login(page, "standard_user", "secret_sauce")

    expect(page).to_have_url(BASE_URL + "inventory.html")
    expect(page.get_by_text("Products")).to_be_visible()


def test_invalid_password_shows_error(page: Page):
    login(page, "standard_user", "wrong_password")

    error = page.locator("[data-test='error']")
    expect(error).to_be_visible()
    expect(error).to_contain_text("Username and password do not match")
    expect(page).to_have_url(BASE_URL)  # never navigated away


def test_locked_out_user_shows_locked_error(page: Page):
    login(page, "locked_out_user", "secret_sauce")

    error = page.locator("[data-test='error']")
    expect(error).to_contain_text("Sorry, this user has been locked out")


def test_login_then_add_item_updates_cart_badge(page: Page):
    login(page, "standard_user", "secret_sauce")

    first_product = page.locator(".inventory_item").first
    product_name = first_product.locator(".inventory_item_name").text_content()
    first_product.get_by_role("button", name="Add to cart").click()

    badge = page.locator(".shopping_cart_badge")
    expect(badge).to_have_text("1")

    page.locator(".shopping_cart_link").click()
    expect(page.locator(".inventory_item_name")).to_have_text(product_name)


def test_logout_returns_to_login_page(page: Page):
    login(page, "standard_user", "secret_sauce")
    expect(page).to_have_url(BASE_URL + "inventory.html")

    page.locator("#react-burger-menu-btn").click()
    page.get_by_role("link", name="Logout").click()

    expect(page).to_have_url(BASE_URL)
    expect(page.get_by_role("button", name="Login")).to_be_visible()

Running it

pytest tests/test_login_flow.py -v

tests/test_login_flow.py::test_valid_login_reaches_inventory_page PASSED  [ 20%]
tests/test_login_flow.py::test_invalid_password_shows_error PASSED       [ 40%]
tests/test_login_flow.py::test_locked_out_user_shows_locked_error PASSED [ 60%]
tests/test_login_flow.py::test_login_then_add_item_updates_cart_badge PASSED [ 80%]
tests/test_login_flow.py::test_logout_returns_to_login_page PASSED       [100%]

============================== 5 passed in 4.87s ===============================

pytest tests/test_login_flow.py --headed --slowmo 200

tests/test_login_flow.py::test_valid_login_reaches_inventory_page PASSED  [ 20%]
tests/test_login_flow.py::test_invalid_password_shows_error PASSED       [ 40%]
tests/test_login_flow.py::test_locked_out_user_shows_locked_error PASSED [ 60%]
tests/test_login_flow.py::test_login_then_add_item_updates_cart_badge PASSED [ 80%]
tests/test_login_flow.py::test_logout_returns_to_login_page PASSED       [100%]

============================== 5 passed in 12.41s ===============================

What this project deliberately practices

  • A helper function (login) instead of repeating the same three lines in every test — a small preview of the Page Object Model formalized in Level 2 Module 1.
  • Testing the negative path (wrong password, locked-out account) with the same rigor as the happy path — a login flow that's only ever tested with correct credentials hasn't actually verified the error handling a real user will eventually hit.
  • Asserting on state that changed, not just "no crash happened" — the cart badge count and the product name carried across pages are concrete, checkable facts, not vague "did anything look wrong" checks.
  • expect(...) everywhere, never a plain assert on a live-fetched DOM value, so the suite tolerates normal small rendering delays without being flaky.

Extending it further (optional)

  • Parametrize the invalid-login test over multiple bad usernames/passwords using @pytest.mark.parametrize, asserting the same error path each time.
  • Add a test for the problem_user account (another built-in saucedemo test account) and see if you can spot the intentionally broken behavior it exhibits — a good exercise in noticing an app defect through E2E testing rather than being told about it in advance.
  • Run the whole file with --browser firefox and --browser webkit and confirm all five tests still pass identically across engines.

How It Actually Works

Nothing in this file introduces new API surface — its value for understanding internals is in seeing the CDP round trips stack up across one realistic flow. login() alone triggers: a Page.navigate + lifecycle-event wait for goto, two fill() calls that each resolve a locator via an accessibility-tree query and dispatch input/change events directly, and a click() that runs the full actionability retry loop before dispatching real Input.dispatchMouseEvent mouse-down/up events on the Login button — after which the driver watches for the resulting Page.frameNavigated to know the SPA/page transition occurred, which is exactly what the subsequent expect(page).to_have_url(...) is polling for.

The pytest-playwright page fixture used throughout this file (formalized in Level 2 Module 2) is what actually owns the Browser/BrowserContext lifecycle: it launches one browser process per test session (reused across tests for speed) but creates a fresh BrowserContext per test function, which is why five independent tests in this file never leak cookies, localStorage, or the saucedemo session between each other — each gets a brand-new isolated CDP browser context with an empty cookie jar, torn down with Target.disposeBrowserContext after the test completes.

Exercise

  1. Build the project exactly as shown and get all 5 tests passing headless.
  2. Re-run headed with --slowmo 200 and watch each test execute — confirm your mental model of each step matches what you see on screen.
  3. Add a parametrized version of the invalid-login test covering at least three different bad credential combinations from a single test function.
  4. Investigate the problem_user account: log in with it, add an item to the cart, and see if the product image or some other element looks wrong compared to standard_user. Write one sentence describing the bug you found — this is the exact kind of defect E2E testing exists to catch.
  5. Run the full file against all three engines (--browser chromium, --browser firefox, --browser webkit) and confirm identical results, completing Level 1.