Cannot get Main window of app

[bernd-wechner]

I'm playing with WInAppDriver in Python following Python examples I find on-line but am stuck on the very basic issue of finding my app's main window. To illustrate, I can start my app easily with:

from appium import webdriver

driver = webdriver.Remote(
    command_executor='http://127.0.0.1:4723',
    desired_capabilities={
        "debugConnectToRunningApp": 'false',
        "app": PathToExe
    })

and I can see driver is: <appium.webdriver.webdriver.WebDriver (session="F33897D8-7C32-4071-BF7F-FC316B21FA77")>

But this is not a web element it's a driver. I want the main window. I can get it's handle using driver.current_window_handle but that isn't an element and I can't find a way to get or find an element from a window handle.

It eventuates that this does return an element:

window = driver.find_element_by_xpath('*')
print(f"Provider: {window.get_attribute('ProviderDescription')}")

and window is <appium.webdriver.webelement.WebElement (session="F33897D8-7C32-4071-BF7F-FC316B21FA77", element="42.1444382")>

So we have an element now. And I can get attributes from it. ProviderDescription is an interesting attribute and is:

Provider: [pid:8680,providerId:0x160A1E Main:Nested [pid:37148,providerId:0x160A1E Annotation(parent link):Microsoft: Annotation Proxy (unmanaged:uiautomationcore.dll); Main:Microsoft: MSAA Proxy (unmanaged:uiautomationcore.dll)]; Hwnd(parent link):Microsoft: HWND Proxy (unmanaged:uiautomationcore.dll)]

When I check the PIDs here I see 8680 is the WinAppDriver.exe PID and 37148 is my app's PID. It seems to be nested as Main.

I can look at window.id and check in inspect.exe this not the ID of my apps window. Alas inspect.exe provides no means for me to search on that id to find what it is and I've not been able to locate it manually, but I suspect it's related to WinAppDriver.exe (i.e PID 8680 above, not PID 37148).

What remains a mystery is:

1. How can one get the Main window element?
2. How can one get all its child elements

On 2, I find (using window above):

children = window.find_elements_by_xpath('*/*')

produces an empty list in children. No surprise as window seems to be related to WinAppDriver and not the main window of my app!

[bernd-wechner]

I have found that a similar reported issue: #359 describes what I was experiencing.

It happens that driver.find_element_by_xpath('*') returns the main window of the application, though I cannot see this documented anywhere, don't find it intuitive at all and can find no method that does not use an xpath to achieve this result (reliance on Xpaths is not a plus here)

Problem is this only works sometimes and intermittently and the issue above explains why. Namely if the app takes longer to render its main window than WinAppDriver does to return a response, then driver.find_element_by_xpath('*') does not returna WIndow at all, but an Application object and that has no children.

The workaround I found is exemplified with this pro forma:

window = driver.find_element_by_xpath('*')
if windowget_attribute('ClassName') == "TApplication":
    main_window = driver.window_handles[0]
    driver.switch_to_window(main_window)

    window = driver.find_element_by_xpath('*')

    if not window.get_attribute('ClassName') == "MyWindowCLass":
        print(f"Failed to find the Main Window")
        exit(1)

This works every time. I have no need for a sleep or to retry with the app I'm using though you could generalise this for very slow-to-start apps with a retry loop of course.

After this find_element_by_name works to find elements on the form. And all is good.

This is a very interesting feature of WinAppDriver, which isn't particularly inspiring. Indeed I think it fair to say WinAppDriver should wait and return the top window. Is there any application for it, in which we don't need the window? It's a UI automation tool, not?