# Test essentials of command line Unicode handling


# Data sources:
#   -c content
#   TODO: stdin
#   Command line argument
#   Environment variable
#   Directory listing

# Data sinks:
#   stdout
#   stderr

# Stream configurations
#   Subprocess default
#   TODO: Copied from current process (via PYTHONIOENCODING)
#   FAILS: Misconfigured via LANG=C

import unittest
import sys
import os.path
from test.support import temp_dir, verbose
from test.support.script_helper import assert_python_ok

# Test pattern from Ned Batchelder's Pragmatic Unicode
TEST_TEXT = "ℙƴ☂ℌøἤ"

class CmdLineUnicodeTest(unittest.TestCase):

    def adjust_env(self, env_vars):
        pass

    def get_expected_stdout(self):
        return sys.stdout.encoding, TEST_TEXT

    def get_expected_stderr(self):
        return sys.stderr.encoding, TEST_TEXT

    def _check_output(self, *run_args, **env_vars):
        self.adjust_env(env_vars)
        rc, out, err = assert_python_ok("-c", *run_args, **env_vars,
                                        __isolated=False)
        if verbose > 1:
            print("Output:", repr(out))
            print("Error output:", repr(err))
        stdout_encoding, expected_stdout = self.get_expected_stdout()
        stderr_encoding, expected_stderr = self.get_expected_stderr()
        self.assertEqual(out.decode(stdout_encoding).strip(), expected_stdout)
        self.assertEqual(err.decode(stderr_encoding).strip(), expected_stderr)

    def test_dash_c_unicode(self):
        cmd = (f"import sys;"
               f"print('{TEST_TEXT}');"
               f"print('{TEST_TEXT}', file=sys.stderr)")
        self._check_output(cmd)

    def test_cmdline_argument_unicode(self):
        cmd = ("import sys;"
               "print(sys.argv[1]);"
               "print(sys.argv[1], file=sys.stderr)")
        self._check_output(cmd, TEST_TEXT)

    def test_envvar_unicode(self):
        cmd = ("import sys, os;"
               "print(os.environ['PYTHON_TEST_TEXT']);"
               "print(os.environ['PYTHON_TEST_TEXT'], file=sys.stderr)")
        self._check_output(cmd, PYTHON_TEST_TEXT=TEST_TEXT)

    def test_direntry_unicode(self):

        cmd = ("import sys, os;"
               "listing = os.listdir(sys.argv[1]);"
               "print(listing[0]);"
               "print(listing[0], file=sys.stderr)")
        with temp_dir() as dirname:
            with open(os.path.join(dirname, TEST_TEXT), "wb"):
                pass
            self._check_output(cmd, dirname)


class CmdLineMisconfiguredUnicodeTest(CmdLineUnicodeTest):
    def adjust_env(self, env_vars):
        env_vars["LANG"] = "C"

    def get_expected_stderr(self):
        encoded_in = os.fsencode(TEST_TEXT)
        decoded_in = encoded_in.decode("ascii", errors="surrogateescape")
        encoded_out = decoded_in.encode("ascii", errors="backslashreplace")
        return "ascii", encoded_out.decode("ascii")