Fix issues with GCS path parsing.

Author: cfezequiel

tfrecorder/converter.py

@@ -83,7 +83,7 @@ def _path_split(filepath: str) -> Tuple[str, str]:
 
   if filepath.startswith(constants.GCS_PREFIX):
     _, path = filepath.split(constants.GCS_PREFIX)
-    head, tail = os.path.split(path)
+    head, tail = os.path.split(os.path.normpath(path))
     return constants.GCS_PREFIX + head, tail
 
   return os.path.split(filepath)
@@ -138,7 +138,9 @@ def _read_image_directory(image_dir: str) -> pd.DataFrame:
 def _is_directory(input_data) -> bool:
   """Returns True if `input_data` is a directory; False otherwise."""
 
-  return tf.io.gfile.isdir(input_data)
+  # Note: First check will flag if user has the necessary credentials
+  # to access the directory (if it is in GCS)
+  return tf.io.gfile.exists(input_data) and tf.io.gfile.isdir(input_data)
 
 
 def _get_job_name(job_label: str = None) -> str:

tfrecorder/converter_test.py

@@ -38,7 +38,24 @@
 # pylint: disable=protected-access
 
 
-class ClientTest(unittest.TestCase):
+class IsDirectoryTest(unittest.TestCase):
+  """Tests `_is_directory`."""
+
+  def test_local_ok(self):
+    """Test function returns True on local directory."""
+
+    with tempfile.TemporaryDirectory() as dirname:
+      self.assertTrue(converter._is_directory(dirname))
+
+  def test_local_exists_but_not_dir(self):
+    """Test function returns False on local (non-directory) file."""
+
+    with tempfile.NamedTemporaryFile(prefix='test_', dir='/tmp') as f:
+      self.assertFalse(converter._is_directory(f.name))
+
+
+# TODO(cezequiel): Refactor to per-function test case classes
+class MiscTest(unittest.TestCase):
   """Misc tests for `client` module."""
 
   def setUp(self):
@@ -88,11 +105,11 @@ def test_path_split(self):
     """Tests `_path_split`."""
 
     filename = 'image_file.jpg'
-    dirpaths = ['/path/to/image/dir', 'gs://path/to/image/dir']
+    dirpaths = ['/path/to/image/dir/', 'gs://path/to/image/dir/']
     for dir_ in dirpaths:
       filepath = os.path.join(dir_, filename)
       act_dirpath, act_filename = converter._path_split(filepath)
-      self.assertEqual(act_dirpath, dir_)
+      self.assertEqual(act_dirpath, dir_.rsplit('/', 1)[0])
       self.assertEqual(act_filename, filename)