if data type has endianness, then codecs must specify endian attribute

Author: brokkoli71

src/zarr/core/metadata/v3.py

@@ -64,7 +64,10 @@ def parse_node_type_array(data: object) -> Literal["array"]:
     raise NodeTypeValidationError("node_type", "array", data)
 
 
-def parse_codecs(data: object) -> tuple[Codec, ...]:
+def parse_codecs(data: object, data_type_has_endianness: bool) -> tuple[Codec, ...]:
+    """
+    if data type has endianness, then codecs must specify endian attribute
+    """
     out: tuple[Codec, ...] = ()
 
     if not isinstance(data, Iterable):
@@ -79,8 +82,18 @@ def parse_codecs(data: object) -> tuple[Codec, ...]:
         else:
             if isinstance(c, str):
                 c = {"name": c}
-            name_parsed, _ = parse_named_configuration(c, require_configuration=False)
-            out += (get_codec_class(name_parsed).from_dict(c),)
+            name_parsed, config_parsed = parse_named_configuration(c, require_configuration=False)
+            codec = get_codec_class(name_parsed).from_dict(c)
+
+            if (
+                hasattr(codec, "endian")
+                and data_type_has_endianness
+                and (config_parsed is None or "endian" not in config_parsed)
+            ):
+                raise ValueError(
+                    f"Expected {name_parsed} codec to specify argument endian for data types for which endianness is applicable."
+                )
+            out += (codec,)
 
     return out
 
@@ -282,7 +295,9 @@ def __init__(
             fill_value, dtype=cast(ALL_DTYPES, data_type_parsed.value)
         )
         attributes_parsed = parse_attributes(attributes)
-        codecs_parsed_partial = parse_codecs(codecs)
+        codecs_parsed_partial = parse_codecs(
+            codecs, data_type_has_endianness=data_type_parsed.has_endianness
+        )
         storage_transformers_parsed = parse_storage_transformers(storage_transformers)
 
         array_spec = ArraySpec(

tests/test_metadata/test_v3.py

@@ -451,8 +451,63 @@ def test_string_codecs() -> None:
     assert result1.codecs == expected.codecs
     result2 = ArrayV3Metadata.from_dict(default_metadata_dict(data_type="bool", codecs="bytes"))
     assert result2.codecs == expected.codecs
-    # TODO
-    # with pytest.raises(
-    #     ValueError, match="Expected bytes codec to specify argument endian for data_type=int16"
-    # ):
-    #     ArrayV3Metadata.from_dict(default_metadata_dict(data_type="int16", codecs=["bytes"]))
+
+
+def test_codec_requires_endian() -> None:
+    with pytest.raises(
+        ValueError,
+        match="Expected bytes codec to specify argument endian for data types for which endianness is applicable.",
+    ):
+        ArrayV3Metadata.from_dict(
+            default_metadata_dict(data_type="int16", codecs=[{"name": "bytes"}])
+        )
+
+    with pytest.raises(
+        ValueError,
+        match="Expected bytes codec to specify argument endian for data types for which endianness is applicable.",
+    ):
+        ArrayV3Metadata.from_dict(
+            default_metadata_dict(
+                data_type="int16", codecs=[{"name": "bytes", "configuration": {}}]
+            )
+        )
+
+    ArrayV3Metadata.from_dict(
+        default_metadata_dict(
+            data_type="int16", codecs=[{"name": "bytes", "configuration": {"endian": "little"}}]
+        )
+    )
+
+    ArrayV3Metadata.from_dict(
+        default_metadata_dict(
+            data_type="int16",
+            codecs=[
+                {
+                    "name": "sharding_indexed",
+                    "configuration": {
+                        "chunk_shape": (1,),
+                        "codecs": [{"name": "bytes", "configuration": {"endian": "little"}}]
+                    },
+                }
+            ],
+        )
+    )
+
+    #TODO
+    with pytest.raises(
+        ValueError,
+        match="Expected bytes codec to specify argument endian for data types for which endianness is applicable.",
+    ):
+        ArrayV3Metadata.from_dict(
+            default_metadata_dict(
+                data_type="int16",
+                codecs=[
+                    {
+                        "name": "sharding_indexed",
+                        "configuration": {
+                            "chunk_shape": (1,),
+                        },
+                    }
+                ],
+            )
+        )