Dataset.copy() to preserve encoding (#1590)

* Dataset.copy() to preserve encoding

* Fix sphinx hook

* Add unit test

doc/whats-new.rst

@@ -209,6 +209,10 @@ Bug fixes
   when objects other than  ``Dataset`` are provided (:issue:`1555`).
   By `Joe Hamman <https://github.com/jhamman>`_.
 
+- :py:func:`xarray.Dataset.copy` would not preserve the encoding property
+  (:issue:`1586`).
+  By `Guido Imperiale <https://github.com/crusaderky>`_.
+
 - :py:func:`xarray.concat` would eagerly load dask variables into memory if
   the first argument was a numpy variable (:issue:`1588`).
   By `Guido Imperiale <https://github.com/crusaderky>`_.

xarray/core/dataset.py

@@ -634,7 +634,8 @@ def copy(self, deep=False):
                                 for k, v in iteritems(self._variables))
         # skip __init__ to avoid costly validation
         return self._construct_direct(variables, self._coord_names.copy(),
-                                      self._dims.copy(), self._attrs_copy())
+                                      self._dims.copy(), self._attrs_copy(),
+                                      encoding=self.encoding)
 
     def _subset_with_all_valid_coords(self, variables, coord_names, attrs):
         needed_dims = set()

xarray/tests/test_dataset.py

@@ -50,6 +50,7 @@ def create_test_data(seed=None):
         obj[v] = (dims, data, {'foo': 'variable'})
     obj.coords['numbers'] = ('dim3', np.array([0, 1, 2, 0, 0, 1, 1, 2, 2, 3],
                                               dtype='int64'))
+    obj.encoding = {'foo': 'bar'}
     assert all(obj.data.flags.writeable for obj in obj.values())
     return obj
 
@@ -1434,6 +1435,7 @@ def test_copy(self):
 
         for copied in [data.copy(deep=False), copy(data)]:
             self.assertDatasetIdentical(data, copied)
+            self.assertEqual(data.encoding, copied.encoding)
             # Note: IndexVariable objects with string dtype are always
             # copied because of xarray.core.util.safe_cast_to_index.
             # Limiting the test to data variables.