single-item tuple for compressors in v2

Author: normanrz

src/zarr/core/array.py

@@ -3816,10 +3816,12 @@ def _parse_chunk_encoding_v2(
     _filters: tuple[numcodecs.abc.Codec, ...] | None
     _compressor: numcodecs.abc.Codec | None
 
-    if compressor is None:
+    if compressor is None or compressor == ():
         _compressor = None
     elif compressor == "auto":
         _compressor = default_compressor
+    elif isinstance(compressor, tuple | list) and len(compressor) == 1:
+        _compressor = parse_compressor(compressor[0])
     else:
         if isinstance(compressor, Iterable) and not isinstance(compressor, dict):
             msg = f"For Zarr v2 arrays, the `compressor` must be a single codec. Got an iterable with type {type(compressor)} instead."

tests/test_array.py

@@ -985,7 +985,9 @@ async def test_create_array_no_filters_compressors(
         compressors=empty_value,
         filters=empty_value,
     )
+    # The v2 metadata stores None and () separately
     assert arr.metadata.filters == empty_value  # type: ignore[union-attr]
+    # The v2 metadata does not allow tuple for compressor, therefore it is turned into None
     assert arr.metadata.compressor is None  # type: ignore[union-attr]
 
     # v3
@@ -1081,6 +1083,8 @@ async def test_create_array_v3_chunk_encoding(
         "auto",
         None,
         numcodecs.Zstd(level=3),
+        (),
+        (numcodecs.Zstd(level=3),),
     ],
 )
 @pytest.mark.parametrize(