Automatic screenshots on test completion no longer working with Playwright 1.55.0

[kyleoperix]

I noticed in my tests, we are no longer getting screenshots when tests complete using the --screenshot cli arg with on or only-on-failure after upgrading to Playwright 1.55.0.

This PR in the playwright-python package I think introduced this: microsoft/playwright-python#2951

This introduced this if statement to determine the file type based on path if a type isn't provided.

if "type" not in params:
                params["type"] = determine_screenshot_type(params["path"])

The on_will_close_browser_context function in playwright-pytest does not provide a type arg to page.screenshot, nor is there a file extension in the screenshot_path. This function then catches all Errors and passes, so this silently fails.

playwright-pytest/pytest-playwright/pytest_playwright/pytest_playwright.py

Line 581 in 48e68b7

def on_will_close_browser_context(self, context: BrowserContext) -> None:

I can submit a pull request to fix this, but I can see a couple of ways to fix this and I don't know which would be preferred by the maintainers:

1. Add a ".png" to the end of the screenshot_path variable.
2. Add a default "type" arg to the screenshot method in the playwright-python package

[kmcgon]

I am the original author of this (wrote it up on my corporate github account, now on personal).

I thought about this, and I think the fix should be that determine_screenshot_type should return "png" if no mimetype is determined (and still throw an Error if an invalid type is provided).

This is in line with the documentation and docstrings for page.screenshot, which state it should default to png.

I'm going to recreate this bug over in playwright-python and open a PR.

EDIT: Here's a link to the bug I opened in playwright-python. microsoft/playwright-python#2965
I have submitted a pull request with a potential fix as well.

[mxschmitt]

Should be fixed by #290 and released as v0.7.1.