TestImageGrab.test_grab_no_xcb fails when gnome-screenshot is installed #6710
Description
What did you do?
Ran Pillow tests with gnome-screenshot installed.
What did you expect to happen?
The tests would all pass!
What actually happened?
The test TestImageGrab.test_grab_no_xcb fails, with the following output:
__________________________________________________ TestImageGrab.test_grab_no_xcb __________________________________________________
self = <Tests.test_imagegrab.TestImageGrab object at 0x7f8aaa979a80>
@pytest.mark.skipif(Image.core.HAVE_XCB, reason="tests missing XCB")
def test_grab_no_xcb(self):
if sys.platform not in ("win32", "darwin"):
with pytest.raises(OSError) as e:
ImageGrab.grab()
> assert str(e.value).startswith("Pillow was built without XCB support")
E assert False
E + where False = <built-in method startswith of str object at 0x7f8aaa62a730>('Pillow was built without XCB support')
E + where <built-in method startswith of str object at 0x7f8aaa62a730> = "cannot identify image file '/var/tmp/paludis
/build/dev-python-Pillow-9.3.0/temp/tmpz2e7yvdk.png'".startswith
E + where "cannot identify image file '/var/tmp/paludis/build/dev-python-Pillow-9.3.0/temp/tmpz2e7yvdk.png'" = str(U
nidentifiedImageError("cannot identify image file '/var/tmp/paludis/build/dev-python-Pillow-9.3.0/temp/tmpz2e7yvdk.png'"))
E + where UnidentifiedImageError("cannot identify image file '/var/tmp/paludis/build/dev-python-Pillow-9.3.0/temp/
tmpz2e7yvdk.png'") = <ExceptionInfo UnidentifiedImageError("cannot identify image file '/var/tmp/paludis/build/dev-python-Pillow-9.3
.0/temp/tmpz2e7yvdk.png'") tblen=3>.value
Tests/test_imagegrab.py:39: AssertionError
------------------------------------------------------- Captured stderr call -------------------------------------------------------
(gnome-screenshot:1481061): Gtk-WARNING **: 13:55:04.539: cannot open display:
It looks like what is happening here is that the ImageGrab.grab() method has a fallback which runs gnome-screenshot (if installed) when XCB is disabled. In my case, the tests are running in a sandboxed package build environment, so gnome-screenshot fails to grab an image, resulting in an error message that is different from the expected error.
Of course, if gnome-screenshot had succeeded, the test would still fail, since it's expecting an error to be raised, not a successful screenshot to be taken.
This issue appears to have been introduced by #6361
What are your OS, Python and Pillow versions?
- OS: Exherbo Linux, amd64
- Python: 3.10.8
- Pillow: 9.3.0