[Python] Using PyTest and Playwright, run one method before loading all the test files

This question seems to be asked a lot, but I'm not finding any answers that work, or at least that work for what I'm looking for.

The premise is simple: I have a website I need to automate testing on (using Playwright with PyTest). I can use browser.new_context(storage_state=auth_file), which puts me in a logged in state, and the testing runs as expected.

But for security reasons, that session times out after a few days, making the auth_file invalid. Or when someone pulls in the repo, there is no auth_file and when they try to tun pytest it's 100% errors because they're stuck on the login screen.

I added a fixture that all the other fixtures depend on that looks to see if the auth_file exists, or if it's modified date is too far back, and if so, run the authentication_login() function (creates a browser window, not-headless, so that I can go through the login process). But all that did was make every test open to the same login page.

So how do I create something that runs even before pytest looks for tests to run?

Here is what I have so far:

auth_file: Path = Path(".auth/storage_state.json")

@pytest.fixture(scope="session", autouse=True)
def authentication():
TIME_OUT: int = 60 * 60

if not auth_file.is_file() or (time.time() - auth_file.stat().st_mtime) > TIME_OUT:
authentication_login()

yield

@pytest.fixture
def playwright(authentication):
with sync_playwright() as p:
yield p


@pytest.fixture
def browser_context(playwright):
browser: Browser = playwright.chromium.launch(**options)
context: BrowserContext = browser.new_context(
storage_state=auth_file
)
yield context
context.close()
browser.close()


@pytest.fixture
def page(browser_context, request):
page: Page = browser_context.new_page()
page.goto("http://localhost/")
wait_for_loaders_to_complete(page)
yield page
page.close()

def authentication_login():
with sync_playwright() as playwright:
browser = playwright.chromium.launch(headless=False, slow_mo=500)

context = browser.new_context()
page = context.new_page()
page.goto("http://localhost/")
page.pause()

login_button: Locator = page.get_by_role("button", name="Login")
if login_button.is_visible():
login_button.click()
...
# Save authentication state
context.storage_state(path=".auth/storage_state.json")
print("New authentication saved.")
context.close()
browser.close()



Part of the problem is that the login routine includes using a OTP, so it really needs to be a human going through the login process ONCE, and after that is done, then all the other tests can launch their own windows (headless or not). Eventually, it will be a dummy account that doesn't need the authenticator, but I would still have the issue of 10+ sessions all trying to login. After the first succeeds, all new sessions will have the valid auth_file.

Continue reading...