fix string codecs for np1.25

Author: brokkoli71

.github/workflows/test.yml

@@ -19,7 +19,6 @@ jobs:
     name: os=${{ matrix.os }}, py=${{ matrix.python-version }}, np=${{ matrix.numpy-version }}, deps=${{ matrix.dependency-set }}
 
     strategy:
-      fail-fast: false
       matrix:
         python-version: ['3.11', '3.12', '3.13']
         numpy-version: ['1.25', '2.1']

src/zarr/core/metadata/v3.py

@@ -37,7 +37,7 @@
 )
 from zarr.core.config import config
 from zarr.core.metadata.common import parse_attributes
-from zarr.core.strings import _STRING_DTYPE as STRING_NP_DTYPE
+from zarr.core.strings import _STRING_DTYPE as STRING_NP_DTYPE, _NUMPY_SUPPORTS_VLEN_STRING
 from zarr.errors import MetadataValidationError, NodeTypeValidationError
 from zarr.registry import get_codec_class
 
@@ -606,6 +606,10 @@ def from_numpy(cls, dtype: np.dtype[Any]) -> DataType:
             return DataType.string
         elif dtype.kind == "S":
             return DataType.bytes
+        elif not _NUMPY_SUPPORTS_VLEN_STRING and dtype.kind == "O":
+            # numpy < 2.0 does not support vlen string dtype
+            # so we fall back on object array of strings
+            return DataType.string
         dtype_to_data_type = {
             "|b1": "bool",
             "bool": "bool",

tests/test_config.py

@@ -8,8 +8,8 @@
 import pytest
 
 import zarr
-from zarr import Array, zeros
-from zarr.abc.codec import CodecInput, CodecOutput, CodecPipeline
+from zarr import Array, zeros, AsyncArray
+from zarr.abc.codec import CodecInput, CodecOutput, CodecPipeline, Codec
 from zarr.abc.store import ByteSetter, Store
 from zarr.codecs import (
     BloscCodec,
@@ -25,6 +25,7 @@
 from zarr.core.codec_pipeline import BatchedCodecPipeline
 from zarr.core.config import BadConfigError, config
 from zarr.core.indexing import SelectorTuple
+from zarr.core.strings import _STRING_DTYPE
 from zarr.registry import (
     fully_qualified_name,
     get_buffer_class,
@@ -36,6 +37,7 @@
     register_ndbuffer,
     register_pipeline,
 )
+from zarr.storage import MemoryStore
 from zarr.testing.buffer import (
     NDBufferUsingTestNDArrayLike,
     StoreExpectingTestBuffer,
@@ -254,8 +256,14 @@ def test_config_buffer_implementation() -> None:
         assert np.array_equal(arr_Crc32c[:], data2d)
 
 
-@pytest.mark.parametrize("dtype", ["int", "bytes", str])
-def test_default_codecs(dtype: str) -> None:
+@pytest.mark.parametrize(("dtype", "expected_codecs"),
+    [
+        ("int", [BytesCodec(), GzipCodec()]),
+        ("bytes", [VLenBytesCodec()]),
+        ("str", [VLenUTF8Codec()]),
+    ]
+                         )
+async def test_default_codecs(dtype: str, expected_codecs: list[Codec]) -> None:
     with config.set(
         {
             "array.v3_default_codecs": {
@@ -265,10 +273,5 @@ def test_default_codecs(dtype: str) -> None:
             }
         }
     ):
-        arr = zeros(shape=(100), dtype=np.dtype(dtype), zarr_format=3)
-        if dtype == "int":
-            assert arr.metadata.codecs == [BytesCodec(), GzipCodec()]
-        elif dtype == "bytes":
-            assert arr.metadata.codecs == [VLenBytesCodec()]
-        elif dtype == "str":
-            assert arr.metadata.codecs == [VLenUTF8Codec()]
+        arr = await AsyncArray.create(shape=(100,), chunk_shape=(100,),dtype=np.dtype(dtype), zarr_format=3, store=MemoryStore())
+        assert arr.metadata.codecs == expected_codecs