Method test_invalid_utf8_arg

Lib/test/test_cmd_line.py:266–309

(self)

    @unittest.skipIf(sys.platform == 'win32',
                     'Windows has a native unicode API')
    def test_invalid_utf8_arg(self):
        # bpo-35883: Py_DecodeLocale() must escape b'\xfd\xbf\xbf\xbb\xba\xba'
        # byte sequence with surrogateescape rather than decoding it as the
        # U+7fffbeba character which is outside the [U+0000; U+10ffff] range of
        # Python Unicode characters.
        #
        # Test with default config, in the C locale, in the Python UTF-8 Mode.
        code = 'import sys, os; s=os.fsencode(sys.argv[1]); print(ascii(s))'

        # TODO: RUSTPYTHON
        def run_default(arg):
            cmd = [sys.executable, '-c', code, arg]
            return subprocess.run(cmd, stdout=subprocess.PIPE, text=True)

        # TODO: RUSTPYTHON
        def run_c_locale(arg):
            cmd = [sys.executable, '-c', code, arg]
            env = dict(os.environ)
            env['LC_ALL'] = 'C'
            return subprocess.run(cmd, stdout=subprocess.PIPE,
                                  text=True, env=env)

        # TODO: RUSTPYTHON
        def run_utf8_mode(arg):
            cmd = [sys.executable, '-X', 'utf8', '-c', code, arg]
            return subprocess.run(cmd, stdout=subprocess.PIPE, text=True)

        valid_utf8 = 'e:\xe9, euro:\u20ac, non-bmp:\U0010ffff'.encode('utf-8')
        # invalid UTF-8 byte sequences with a valid UTF-8 sequence
        # in the middle.
        invalid_utf8 = (
            b'\xff'                      # invalid byte
            b'\xc3\xff'                  # invalid byte sequence
            b'\xc3\xa9'                  # valid utf-8: U+00E9 character
            b'\xed\xa0\x80'              # lone surrogate character (invalid)
            b'\xfd\xbf\xbf\xbb\xba\xba'  # character outside [U+0000; U+10ffff]
        )
        test_args = [valid_utf8, invalid_utf8]

        for run_cmd in (run_default, run_c_locale, run_utf8_mode):
            with self.subTest(run_cmd=run_cmd):
                for arg in test_args:
                    proc = run_cmd(arg)
                    self.assertEqual(proc.stdout.rstrip(), ascii(arg))

    @unittest.skipUnless((sys.platform == 'darwin' or
                support.is_android), 'test specific to Mac OS X and Android')