Fix bad filenames for Python3.

In Python3, if there are filenames encoded with 8-bit encodings, there
might be an issues with converting them into unicode objects. This is a
workaround on this subject. Python2 is not affected.

Other than that, there was tests added to cover this case, appropriate
Makefile which automate creating venvs for both: Python 2 and 3, and
also there is a check against pep8 rules using flake8.

test_adbfs.py

@@ -0,0 +1,71 @@
+# -*- coding: utf-8 -*-
+import os
+import six
+import unittest
+try:
+    from unittest import mock
+except ImportError:
+    import mock
+
+FILE = os.path.join(os.path.abspath(os.path.dirname(__file__)), 'adbfs')
+
+try:
+    from importlib.util import spec_from_loader, module_from_spec
+    from importlib.machinery import SourceFileLoader
+    spec = spec_from_loader("adbfs", SourceFileLoader("adbfs", FILE))
+    adbfs = module_from_spec(spec)
+    spec.loader.exec_module(adbfs)
+except ImportError:
+    # py27
+    import imp
+    adbfs = imp.load_source('adbfs', FILE)
+
+LISTING = '''\
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/Grüß Gott
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/\x80
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/Γεια σας
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/Здравствуйте
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/שָׁלוֹם
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/السَّلامُ عَلَيْكُمْ
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/გამარჯობა
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/こんにちは。
+-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 /storage/emulated/0/안녕하십니까
+'''  # noqa
+
+
+class TestCheckOutput(unittest.TestCase):
+
+    @mock.patch('subprocess.check_output')
+    def test_check_output(self, out):
+        """
+        As for Python2 (and its last version: 2.7), subprocess.check_output
+        always return string like objects, contrary to bytes - no conversion
+        to string is needed.
+        Python3 treats string as unicode objects, but subprocess.check_output
+        returns bytes object, which is equvalend for py2 string… annoying.
+        """
+        if six.PY3:
+            out.return_value = bytes(LISTING, 'utf-8')
+        else:
+            out.return_value = LISTING
+        result = adbfs.check_output(None)
+        self.assertEqual(result, LISTING)
+
+    @mock.patch('subprocess.check_output')
+    def test_check_output_py3_invalid_char(self, out):
+        """
+        Special case for py3. We have bytes with some weird character - like
+        some system write something with codepage, instead of utf8.
+        """
+        if six.PY2:
+            # doesn't affect Python 2
+            return
+        line = (b'-rw-rw----  1 0  1015  0 01/01/2010 22:11:01 '
+                b'/storage/emulated/0/\xe2\n')  # Latin 1 char â
+        out.return_value = bytes(line)
+        result = adbfs.check_output(None)
+        self.assertEqual(result, line.decode('iso-8859-1'))
+
+
+if __name__ == "__main__":
+    unittest.main()