preserve boolean dtype in encoding (#7720)

* preserve boolean-dtype within encoding, adapt test

* add comment as per review

* add whats-new.rst entry

doc/whats-new.rst

@@ -54,7 +54,8 @@ Bug fixes
   (:issue:`7420`, :pull:`7441`).
   By `Justus Magin <https://github.com/keewis>`_ and
   `Spencer Clark  <https://github.com/spencerkclark>`_.
-
+- Preserve boolean dtype within encoding (:issue:`7652`, :pull:`7720`).
+  By `Kai Mühlbauer <https://github.com/kmuehlbauer>`_
 
 Documentation
 ~~~~~~~~~~~~~

xarray/coding/variables.py

@@ -458,7 +458,9 @@ def encode(self, variable: Variable, name: T_Name = None) -> Variable:
     def decode(self, variable: Variable, name: T_Name = None) -> Variable:
         if variable.attrs.get("dtype", False) == "bool":
             dims, data, attrs, encoding = unpack_for_decoding(variable)
-            del attrs["dtype"]
+            # overwrite (!) dtype in encoding, and remove from attrs
+            # needed for correct subsequent encoding
+            encoding["dtype"] = attrs.pop("dtype")
             data = BoolTypeArray(data)
             return Variable(dims, data, attrs, encoding, fastpath=True)
         else:

xarray/tests/test_backends.py

@@ -630,6 +630,11 @@ def test_roundtrip_boolean_dtype(self) -> None:
         with self.roundtrip(original) as actual:
             assert_identical(original, actual)
             assert actual["x"].dtype == "bool"
+            # this checks for preserving dtype during second roundtrip
+            # see https://github.com/pydata/xarray/issues/7652#issuecomment-1476956975
+            with self.roundtrip(actual) as actual2:
+                assert_identical(original, actual2)
+                assert actual2["x"].dtype == "bool"
 
     def test_orthogonal_indexing(self) -> None:
         in_memory = create_test_data()