Ensure compressor=None results in no compression for V2 (#2709)

* Ensure compressor=None results in no compression for V2

* rename argumnent

* Update tests/test_v2.py

Co-authored-by: Davis Bennett <[email protected]>

* fix

* coverage

* add release note

* Update release note

---------

Co-authored-by: Davis Bennett <[email protected]>
Co-authored-by: David Stansby <[email protected]>

Author: 3 people

docs/release-notes.rst

@@ -13,6 +13,8 @@ Bug fixes
 
 * Fixes a bug that prevented reading Zarr format 2 data with consolidated metadata written using ``zarr-python`` version 2 (:issue:`2694`).
 
+* Ensure that compressor=None results in no compression when writing Zarr format 2 data (:issue:`2708`)
+
 Behaviour changes
 ~~~~~~~~~~~~~~~~~
 

src/zarr/core/array.py

@@ -4131,15 +4131,22 @@ def _parse_chunk_encoding_v3(
 
 
 def _parse_deprecated_compressor(
-    compressor: CompressorLike | None, compressors: CompressorsLike
+    compressor: CompressorLike | None, compressors: CompressorsLike, zarr_format: int = 3
 ) -> CompressorsLike | None:
-    if compressor:
+    if compressor != "auto":
         if compressors != "auto":
             raise ValueError("Cannot specify both `compressor` and `compressors`.")
-        warn(
-            "The `compressor` argument is deprecated. Use `compressors` instead.",
-            category=UserWarning,
-            stacklevel=2,
-        )
-        compressors = (compressor,)
+        if zarr_format == 3:
+            warn(
+                "The `compressor` argument is deprecated. Use `compressors` instead.",
+                category=UserWarning,
+                stacklevel=2,
+            )
+        if compressor is None:
+            # "no compression"
+            compressors = ()
+        else:
+            compressors = (compressor,)
+    elif zarr_format == 2 and compressor == compressors == "auto":
+        compressors = ({"id": "blosc"},)
     return compressors

src/zarr/core/group.py

@@ -1011,7 +1011,7 @@ async def create_array(
         shards: ShardsLike | None = None,
         filters: FiltersLike = "auto",
         compressors: CompressorsLike = "auto",
-        compressor: CompressorLike = None,
+        compressor: CompressorLike = "auto",
         serializer: SerializerLike = "auto",
         fill_value: Any | None = 0,
         order: MemoryOrder | None = None,
@@ -1114,8 +1114,9 @@ async def create_array(
         AsyncArray
 
         """
-
-        compressors = _parse_deprecated_compressor(compressor, compressors)
+        compressors = _parse_deprecated_compressor(
+            compressor, compressors, zarr_format=self.metadata.zarr_format
+        )
         return await create_array(
             store=self.store_path,
             name=name,
@@ -2244,7 +2245,7 @@ def create_array(
         shards: ShardsLike | None = None,
         filters: FiltersLike = "auto",
         compressors: CompressorsLike = "auto",
-        compressor: CompressorLike = None,
+        compressor: CompressorLike = "auto",
         serializer: SerializerLike = "auto",
         fill_value: Any | None = 0,
         order: MemoryOrder | None = "C",
@@ -2346,7 +2347,9 @@ def create_array(
         -------
         AsyncArray
         """
-        compressors = _parse_deprecated_compressor(compressor, compressors)
+        compressors = _parse_deprecated_compressor(
+            compressor, compressors, zarr_format=self.metadata.zarr_format
+        )
         return Array(
             self._sync(
                 self._async_group.create_array(

tests/test_group.py

@@ -9,7 +9,7 @@
 
 import numpy as np
 import pytest
-from numcodecs import Zstd
+from numcodecs import Blosc
 
 import zarr
 import zarr.api.asynchronous
@@ -499,7 +499,7 @@ def test_group_child_iterators(store: Store, zarr_format: ZarrFormat, consolidat
                     "chunks": (1,),
                     "order": "C",
                     "filters": None,
-                    "compressor": Zstd(level=0),
+                    "compressor": Blosc(),
                     "zarr_format": zarr_format,
                 },
                 "subgroup": {
@@ -1505,13 +1505,3 @@ def test_group_members_concurrency_limit(store: MemoryStore) -> None:
         elapsed = time.time() - start
 
         assert elapsed > num_groups * get_latency
-
-
-@pytest.mark.parametrize("store", ["local", "memory"], indirect=["store"])
-def test_deprecated_compressor(store: Store) -> None:
-    g = zarr.group(store=store, zarr_format=2)
-    with pytest.warns(UserWarning, match="The `compressor` argument is deprecated.*"):
-        a = g.create_array(
-            "foo", shape=(100,), chunks=(10,), dtype="i4", compressor={"id": "blosc"}
-        )
-        assert a.metadata.compressor.codec_id == "blosc"

tests/test_metadata/test_consolidated.py

@@ -5,7 +5,7 @@
 
 import numpy as np
 import pytest
-from numcodecs import Zstd
+from numcodecs import Blosc
 
 import zarr.api.asynchronous
 import zarr.api.synchronous
@@ -522,7 +522,7 @@ async def test_consolidated_metadata_v2(self):
                         attributes={"key": "a"},
                         chunks=(1,),
                         fill_value=0,
-                        compressor=Zstd(level=0),
+                        compressor=Blosc(),
                         order="C",
                     ),
                     "g1": GroupMetadata(

tests/test_v2.py

@@ -7,11 +7,13 @@
 import pytest
 from numcodecs import Delta
 from numcodecs.blosc import Blosc
+from numcodecs.zstd import Zstd
 
 import zarr
 import zarr.core.buffer
 import zarr.storage
 from zarr import config
+from zarr.abc.store import Store
 from zarr.core.buffer.core import default_buffer_prototype
 from zarr.core.sync import sync
 from zarr.storage import MemoryStore, StorePath
@@ -93,11 +95,7 @@ async def test_v2_encode_decode(dtype):
         store = zarr.storage.MemoryStore()
         g = zarr.group(store=store, zarr_format=2)
         g.create_array(
-            name="foo",
-            shape=(3,),
-            chunks=(3,),
-            dtype=dtype,
-            fill_value=b"X",
+            name="foo", shape=(3,), chunks=(3,), dtype=dtype, fill_value=b"X", compressor=None
         )
 
         result = await store.get("foo/.zarray", zarr.core.buffer.default_buffer_prototype())
@@ -166,6 +164,29 @@ def test_v2_filters_codecs(filters: Any, order: Literal["C", "F"]) -> None:
     np.testing.assert_array_equal(result, array_fixture)
 
 
+@pytest.mark.filterwarnings("ignore")
+@pytest.mark.parametrize("store", ["memory"], indirect=True)
+def test_create_array_defaults(store: Store):
+    """
+    Test that passing compressor=None results in no compressor. Also test that the default value of the compressor
+    parameter does produce a compressor.
+    """
+    g = zarr.open(store, mode="w", zarr_format=2)
+    arr = g.create_array("one", dtype="i8", shape=(1,), chunks=(1,), compressor=None)
+    assert arr._async_array.compressor is None
+    assert not (arr.filters)
+    arr = g.create_array("two", dtype="i8", shape=(1,), chunks=(1,))
+    assert arr._async_array.compressor is not None
+    assert not (arr.filters)
+    arr = g.create_array("three", dtype="i8", shape=(1,), chunks=(1,), compressor=Zstd())
+    assert arr._async_array.compressor is not None
+    assert not (arr.filters)
+    with pytest.raises(ValueError):
+        g.create_array(
+            "four", dtype="i8", shape=(1,), chunks=(1,), compressor=None, compressors=None
+        )
+
+
 @pytest.mark.parametrize("array_order", ["C", "F"])
 @pytest.mark.parametrize("data_order", ["C", "F"])
 @pytest.mark.parametrize("memory_order", ["C", "F"])