Merge branch 'main' of https://github.com/zarr-developers/zarr-python into feat/fixed-length-strings

Author: d-v-b

changes/2972.misc.rst

@@ -0,0 +1 @@
+Avoid an unnecessary memory copy when writing Zarr with obstore

changes/3039.bugfix.rst

@@ -0,0 +1,5 @@
+It is now possible to specify no compressor when creating a zarr format 2 array.
+This can be done by passing ``compressor=None`` to the various array creation routines.
+
+The default behaviour of automatically choosing a suitable default compressor remains if the compressor argument is not given.
+To reproduce the behaviour in previous zarr-python versions when ``compressor=None`` was passed, pass ``compressor='auto'`` instead.

src/zarr/api/asynchronous.py

@@ -12,6 +12,7 @@
 from zarr.core.array import (
     Array,
     AsyncArray,
+    CompressorLike,
     _get_default_chunk_encoding_v2,
     create_array,
     from_array,
@@ -844,7 +845,7 @@ async def create(
     *,  # Note: this is a change from v2
     chunks: ChunkCoords | int | None = None,  # TODO: v2 allowed chunks=True
     dtype: npt.DTypeLike | None = None,
-    compressor: dict[str, JSON] | None = None,  # TODO: default and type change
+    compressor: CompressorLike = "auto",
     fill_value: Any | None = 0,  # TODO: need type
     order: MemoryOrder | None = None,
     store: str | StoreLike | None = None,
@@ -995,9 +996,9 @@ async def create(
         if chunks is None:
             chunks = shape
         default_filters, default_compressor = _get_default_chunk_encoding_v2(dtype_wrapped)
-        if filters is None:
+        if not filters:
             filters = default_filters  # type: ignore[assignment]
-        if compressor is None:
+        if compressor == "auto":
             compressor = default_compressor
     elif zarr_format == 3 and chunk_shape is None:  # type: ignore[redundant-expr]
         if chunks is not None:

src/zarr/api/synchronous.py

@@ -7,7 +7,7 @@
 import zarr.api.asynchronous as async_api
 import zarr.core.array
 from zarr._compat import _deprecate_positional_args
-from zarr.core.array import Array, AsyncArray
+from zarr.core.array import Array, AsyncArray, CompressorLike
 from zarr.core.group import Group
 from zarr.core.sync import sync
 from zarr.core.sync_group import create_hierarchy
@@ -600,7 +600,7 @@ def create(
     *,  # Note: this is a change from v2
     chunks: ChunkCoords | int | bool | None = None,
     dtype: npt.DTypeLike | None = None,
-    compressor: dict[str, JSON] | None = None,  # TODO: default and type change
+    compressor: CompressorLike = "auto",
     fill_value: Any | None = 0,  # TODO: need type
     order: MemoryOrder | None = None,
     store: str | StoreLike | None = None,

src/zarr/core/array.py

@@ -108,6 +108,7 @@
     T_ArrayMetadata,
 )
 from zarr.core.metadata.v2 import (
+    CompressorLikev2,
     parse_compressor,
     parse_filters,
 )
@@ -310,7 +311,7 @@ async def create(
         dimension_separator: Literal[".", "/"] | None = None,
         order: MemoryOrder | None = None,
         filters: list[dict[str, JSON]] | None = None,
-        compressor: dict[str, JSON] | None = None,
+        compressor: CompressorLikev2 | Literal["auto"] = "auto",
         # runtime
         overwrite: bool = False,
         data: npt.ArrayLike | None = None,
@@ -401,7 +402,7 @@ async def create(
         dimension_separator: Literal[".", "/"] | None = None,
         order: MemoryOrder | None = None,
         filters: list[dict[str, JSON]] | None = None,
-        compressor: dict[str, JSON] | None = None,
+        compressor: CompressorLike = "auto",
         # runtime
         overwrite: bool = False,
         data: npt.ArrayLike | None = None,
@@ -436,7 +437,7 @@ async def create(
         dimension_separator: Literal[".", "/"] | None = None,
         order: MemoryOrder | None = None,
         filters: list[dict[str, JSON]] | None = None,
-        compressor: dict[str, JSON] | None = None,
+        compressor: CompressorLike = "auto",
         # runtime
         overwrite: bool = False,
         data: npt.ArrayLike | None = None,
@@ -577,7 +578,7 @@ async def _create(
         dimension_separator: Literal[".", "/"] | None = None,
         order: MemoryOrder | None = None,
         filters: list[dict[str, JSON]] | None = None,
-        compressor: dict[str, JSON] | None = None,
+        compressor: CompressorLike = "auto",
         # runtime
         overwrite: bool = False,
         data: npt.ArrayLike | None = None,
@@ -612,7 +613,7 @@ async def _create(
                 raise ValueError(
                     "filters cannot be used for arrays with zarr_format 3. Use array-to-array codecs instead."
                 )
-            if compressor is not None:
+            if compressor != "auto":
                 raise ValueError(
                     "compressor cannot be used for arrays with zarr_format 3. Use bytes-to-bytes codecs instead."
                 )
@@ -782,7 +783,7 @@ def _create_metadata_v2(
         dimension_separator: Literal[".", "/"] | None = None,
         fill_value: float | None = None,
         filters: Iterable[dict[str, JSON] | numcodecs.abc.Codec] | None = None,
-        compressor: dict[str, JSON] | numcodecs.abc.Codec | None = None,
+        compressor: CompressorLikev2 = None,
         attributes: dict[str, JSON] | None = None,
     ) -> ArrayV2Metadata:
         if dimension_separator is None:
@@ -813,7 +814,7 @@ async def _create_v2(
         dimension_separator: Literal[".", "/"] | None = None,
         fill_value: float | None = None,
         filters: Iterable[dict[str, JSON] | numcodecs.abc.Codec] | None = None,
-        compressor: dict[str, JSON] | numcodecs.abc.Codec | None = None,
+        compressor: CompressorLike = "auto",
         attributes: dict[str, JSON] | None = None,
         overwrite: bool = False,
     ) -> AsyncArray[ArrayV2Metadata]:
@@ -825,6 +826,17 @@ async def _create_v2(
         else:
             await ensure_no_existing_node(store_path, zarr_format=2)
 
+        compressor_parsed: CompressorLikev2
+        if compressor == "auto":
+            _, compressor_parsed = _get_default_chunk_encoding_v2(dtype)
+        elif isinstance(compressor, BytesBytesCodec):
+            raise ValueError(
+                "Cannot use a BytesBytesCodec as a compressor for zarr v2 arrays. "
+                "Use a numcodecs codec directly instead."
+            )
+        else:
+            compressor_parsed = compressor
+
         metadata = cls._create_metadata_v2(
             shape=shape,
             dtype=dtype,
@@ -833,7 +845,7 @@ async def _create_v2(
             dimension_separator=dimension_separator,
             fill_value=fill_value,
             filters=filters,
-            compressor=compressor,
+            compressor=compressor_parsed,
             attributes=attributes,
         )
 
@@ -1780,7 +1792,7 @@ def create(
         dimension_separator: Literal[".", "/"] | None = None,
         order: MemoryOrder | None = None,
         filters: list[dict[str, JSON]] | None = None,
-        compressor: dict[str, JSON] | None = None,
+        compressor: CompressorLike = "auto",
         # runtime
         overwrite: bool = False,
         config: ArrayConfigLike | None = None,
@@ -1909,7 +1921,7 @@ def _create(
         dimension_separator: Literal[".", "/"] | None = None,
         order: MemoryOrder | None = None,
         filters: list[dict[str, JSON]] | None = None,
-        compressor: dict[str, JSON] | None = None,
+        compressor: CompressorLike = "auto",
         # runtime
         overwrite: bool = False,
         config: ArrayConfigLike | None = None,
@@ -3814,7 +3826,11 @@ def _build_parents(
     | Literal["auto"]
     | None
 )
-CompressorLike: TypeAlias = dict[str, JSON] | BytesBytesCodec | numcodecs.abc.Codec | None
+# Union of acceptable types for users to pass in for both v2 and v3 compressors
+CompressorLike: TypeAlias = (
+    dict[str, JSON] | BytesBytesCodec | numcodecs.abc.Codec | Literal["auto"] | None
+)
+
 CompressorsLike: TypeAlias = (
     Iterable[dict[str, JSON] | BytesBytesCodec | numcodecs.abc.Codec]
     | dict[str, JSON]

src/zarr/core/buffer/core.py

@@ -255,6 +255,19 @@ def as_numpy_array(self) -> npt.NDArray[Any]:
         """
         ...
 
+    def as_buffer_like(self) -> BytesLike:
+        """Returns the buffer as an object that implements the Python buffer protocol.
+
+        Notes
+        -----
+        Might have to copy data, since the implementation uses `.as_numpy_array()`.
+
+        Returns
+        -------
+            An object that implements the Python buffer protocol
+        """
+        return memoryview(self.as_numpy_array())  # type: ignore[arg-type]
+
     def to_bytes(self) -> bytes:
         """Returns the buffer as `bytes` (host memory).
 

src/zarr/core/metadata/v2.py

@@ -3,7 +3,7 @@
 import base64
 import warnings
 from collections.abc import Iterable, Sequence
-from typing import TYPE_CHECKING, Any, TypedDict
+from typing import TYPE_CHECKING, Any, TypeAlias, TypedDict
 
 import numcodecs.abc
 
@@ -41,6 +41,10 @@ class ArrayV2MetadataDict(TypedDict):
     attributes: dict[str, JSON]
 
 
+# Union of acceptable types for v2 compressors
+CompressorLikev2: TypeAlias = dict[str, JSON] | numcodecs.abc.Codec | None
+
+
 @dataclass(frozen=True, kw_only=True)
 class ArrayV2Metadata(Metadata):
     shape: ChunkCoords
@@ -50,7 +54,7 @@ class ArrayV2Metadata(Metadata):
     order: MemoryOrder = "C"
     filters: tuple[numcodecs.abc.Codec, ...] | None = None
     dimension_separator: Literal[".", "/"] = "."
-    compressor: numcodecs.abc.Codec | None = None
+    compressor: CompressorLikev2
     attributes: dict[str, JSON] = field(default_factory=dict)
     zarr_format: Literal[2] = field(init=False, default=2)
 
@@ -63,7 +67,7 @@ def __init__(
         fill_value: Any,
         order: MemoryOrder,
         dimension_separator: Literal[".", "/"] = ".",
-        compressor: numcodecs.abc.Codec | dict[str, JSON] | None = None,
+        compressor: CompressorLikev2 = None,
         filters: Iterable[numcodecs.abc.Codec | dict[str, JSON]] | None = None,
         attributes: dict[str, JSON] | None = None,
     ) -> None:

src/zarr/storage/_local.py

@@ -52,10 +52,10 @@ def _put(
         with path.open("r+b") as f:
             f.seek(start)
             # write takes any object supporting the buffer protocol
-            f.write(value.as_numpy_array())  # type: ignore[arg-type]
+            f.write(value.as_buffer_like())
         return None
     else:
-        view = memoryview(value.as_numpy_array())  # type: ignore[arg-type]
+        view = value.as_buffer_like()
         if exclusive:
             mode = "xb"
         else:

src/zarr/storage/_obstore.py

@@ -161,15 +161,15 @@ async def set(self, key: str, value: Buffer) -> None:
 
         self._check_writable()
 
-        buf = value.to_bytes()
+        buf = value.as_buffer_like()
         await obs.put_async(self.store, key, buf)
 
     async def set_if_not_exists(self, key: str, value: Buffer) -> None:
         # docstring inherited
         import obstore as obs
 
         self._check_writable()
-        buf = value.to_bytes()
+        buf = value.as_buffer_like()
         with contextlib.suppress(obs.exceptions.AlreadyExistsError):
             await obs.put_async(self.store, key, buf, mode="create")
 

tests/test_api.py

@@ -2,6 +2,8 @@
 
 from typing import TYPE_CHECKING
 
+import zarr.codecs
+
 if TYPE_CHECKING:
     import pathlib
 
@@ -1190,3 +1192,20 @@ def test_gpu_basic(store: Store, zarr_format: ZarrFormat | None) -> None:
         # assert_array_equal doesn't check the type
         assert isinstance(result, type(src))
         cp.testing.assert_array_equal(result, src[:10, :10])
+
+
+def test_v2_without_compressor() -> None:
+    # Make sure it's possible to set no compressor for v2 arrays
+    arr = zarr.create(store={}, shape=(1), dtype="uint8", zarr_format=2, compressor=None)
+    assert arr.compressors == ()
+
+
+def test_v2_with_v3_compressor() -> None:
+    # Check trying to create a v2 array with a v3 compressor fails
+    with pytest.raises(
+        ValueError,
+        match="Cannot use a BytesBytesCodec as a compressor for zarr v2 arrays. Use a numcodecs codec directly instead.",
+    ):
+        zarr.create(
+            store={}, shape=(1), dtype="uint8", zarr_format=2, compressor=zarr.codecs.BloscCodec()
+        )