|  | # Ridiculously simple test of the os.startfile function for Windows. | 
|  | # | 
|  | # empty.vbs is an empty file (except for a comment), which does | 
|  | # nothing when run with cscript or wscript. | 
|  | # | 
|  | # A possible improvement would be to have empty.vbs do something that | 
|  | # we can detect here, to make sure that not only the os.startfile() | 
|  | # call succeeded, but also the script actually has run. | 
|  |  | 
|  | import unittest | 
|  | from test import support | 
|  | from test.support import os_helper | 
|  | import os | 
|  | import platform | 
|  | import sys | 
|  | from os import path | 
|  |  | 
|  | startfile = support.get_attribute(os, 'startfile') | 
|  |  | 
|  |  | 
|  | @unittest.skipIf(platform.win32_is_iot(), "starting files is not supported on Windows IoT Core or nanoserver") | 
|  | class TestCase(unittest.TestCase): | 
|  | def test_nonexisting(self): | 
|  | self.assertRaises(OSError, startfile, "nonexisting.vbs") | 
|  |  | 
|  | def test_empty(self): | 
|  | # We need to make sure the child process starts in a directory | 
|  | # we're not about to delete. If we're running under -j, that | 
|  | # means the test harness provided directory isn't a safe option. | 
|  | # See http://bugs.python.org/issue15526 for more details | 
|  | with os_helper.change_cwd(path.dirname(sys.executable)): | 
|  | empty = path.join(path.dirname(__file__), "empty.vbs") | 
|  | startfile(empty) | 
|  | startfile(empty, "open") | 
|  | startfile(empty, cwd=path.dirname(sys.executable)) | 
|  |  | 
|  | def test_python(self): | 
|  | # Passing "-V" ensures that it closes quickly, though still not | 
|  | # quickly enough that we can run in the test directory | 
|  | cwd, name = path.split(sys.executable) | 
|  | startfile(name, arguments="-V", cwd=cwd) | 
|  | startfile(name, arguments="-V", cwd=cwd, show_cmd=0) | 
|  |  | 
|  | if __name__ == "__main__": | 
|  | unittest.main() |