transform numpy object-dtype strings (vlen) to numpy unicode strings

pydata · Mar 21, 2023 · 4b65ffd · 4b65ffd
1 parent 2041f6e
commit 4b65ffd
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 0 deletions.
diff --git a/xarray/conventions.py b/xarray/conventions.py
@@ -389,6 +389,15 @@ def decode_cf_variable(
     if decode_times:
         var = times.CFDatetimeCoder(use_cftime=use_cftime).decode(var, name=name)
 
+    # transform numpy object-dtype strings to numpy unicode strings
+    if (
+        "dtype" in var.encoding
+        and var.encoding["dtype"] == str
+        and original_dtype == object
+    ):
+        original_dtype = var.encoding["dtype"]
+        var = var.astype(var.encoding["dtype"])
+
     dimensions, data, attributes, encoding = variables.unpack_for_decoding(var)
     # TODO(shoyer): convert everything below to use coders
 

diff --git a/xarray/tests/test_backends.py b/xarray/tests/test_backends.py
@@ -1360,6 +1360,7 @@ def test_encoding_kwarg_vlen_string(self) -> None:
             with self.roundtrip(original, save_kwargs=kwargs) as actual:
                 assert actual["x"].encoding["dtype"] is str
                 assert_identical(actual, expected)
+                assert actual["x"].dtype == "<U3"
 
     def test_roundtrip_string_with_fill_value_vlen(self) -> None:
         values = np.array(["ab", "cdef", np.nan], dtype=object)