add dtype parsing, and tweak auto_partitioning func

Author: d-v-b

src/zarr/api/asynchronous.py

@@ -18,6 +18,7 @@
     ChunkCoords,
     MemoryOrder,
     ZarrFormat,
+    parse_dtype,
 )
 from zarr.core.config import config
 from zarr.core.group import AsyncGroup, ConsolidatedMetadata, GroupMetadata
@@ -972,7 +973,7 @@ async def create_array(
 
     store_path = await make_store_path(store, path=path, mode=mode, storage_options=storage_options)
     sub_codecs = (*filters, BytesCodec(), *compressors)
-
+    _dtype_parsed = parse_dtype(dtype, zarr_format=zarr_format)
     if zarr_format == 2:
         if shard_shape is not None or shard_shape != "auto":
             msg = (
@@ -988,7 +989,7 @@ async def create_array(
         return await AsyncArray._create_v2(
             store_path=store_path,
             shape=shape,
-            dtype=dtype,
+            dtype=_dtype_parsed,
             chunks=chunk_shape,
             dimension_separator="/",
             fill_value=fill_value,
@@ -1018,7 +1019,7 @@ async def create_array(
         return await AsyncArray._create_v3(
             store_path=store_path,
             shape=shape,
-            dtype=dtype,
+            dtype=_dtype_parsed,
             fill_value=fill_value,
             attributes=attributes,
             chunk_shape=chunks_out,

src/zarr/core/chunk_grids.py

@@ -25,8 +25,10 @@
 if TYPE_CHECKING:
     from collections.abc import Iterator
     from typing import Self
+
     import numpy.typing as npt
 
+
 def _guess_chunks(
     shape: ShapeLike,
     typesize: int,
@@ -144,7 +146,6 @@ def normalize_chunks(chunks: Any, shape: tuple[int, ...], typesize: int) -> tupl
     return tuple(int(c) for c in chunks)
 
 
-
 def _auto_partition(
     shape: tuple[int, ...],
     dtype: npt.DTypeLike,
@@ -170,12 +171,13 @@ def _auto_partition(
     else:
         if chunk_shape == "auto":
             # aim for a 1MiB chunk
-            _chunks_out = _guess_chunks(shape, item_size, max_bytes=1024**2)
+            _chunks_out = _guess_chunks(shape, item_size, max_bytes=1024)
         else:
             _chunks_out = chunk_shape
+
         if shard_shape == "auto":
             # TODO: fix me! this should be capped at some sane shard shape
-            _shards_out = tuple(c * 8 for c in _chunks_out)
+            _shards_out = tuple(c * 2 for c in _chunks_out)
         else:
             _shards_out = shard_shape