start fixing config

Author: d-v-b

src/zarr/core/array.py

@@ -70,7 +70,6 @@
 from zarr.core.dtype import (
     DTypeWrapper,
     FixedLengthAsciiString,
-    FixedLengthUnicodeString,
     VariableLengthString,
     parse_data_type,
 )
@@ -4248,19 +4247,15 @@ def _get_default_chunk_encoding_v2(
     """
     Get the default chunk encoding for Zarr format 2 arrays, given a dtype
     """
-    from numcodecs import VLenBytes as numcodecs_VLenBytes
-    from numcodecs import VLenUTF8 as numcodecs_VLenUTF8
-    from numcodecs import Zstd as numcodecs_zstd
-
-    if isinstance(dtype, VariableLengthString | FixedLengthUnicodeString):
-        filters = (numcodecs_VLenUTF8(),)
-    elif isinstance(dtype, FixedLengthAsciiString):
-        filters = (numcodecs_VLenBytes(),)
+    if dtype._zarr_v3_name in zarr_config.get("array.v2_default_filters"):
+        filters = zarr_config.get(f"array.v2_default_filters.{dtype._zarr_v3_name}")
     else:
-        filters = None
-
-    compressor = numcodecs_zstd(level=0, checksum=False)
+        filters = zarr_config.get("array.v2_default_filters.default")
 
+    if dtype._zarr_v3_name in zarr_config.get("array.v2_default_compressor"):
+        compressor = zarr_config.get(f"array.v2_default_compressor.{dtype._zarr_v3_name}")
+    else:
+        compressor = zarr_config.get("array.v2_default_compressor.default")
     return filters, compressor
 
 

src/zarr/core/config.py

@@ -36,6 +36,8 @@
 if TYPE_CHECKING:
     from donfig.config_obj import ConfigSet
 
+from collections import defaultdict
+
 
 class BadConfigError(ValueError):
     _msg = "bad Config: %r"
@@ -77,37 +79,24 @@ def enable_gpu(self) -> ConfigSet:
             "array": {
                 "order": "C",
                 "write_empty_chunks": False,
-                "v2_default_compressor": {
-                    "numeric": {"id": "zstd", "level": 0, "checksum": False},
-                    "string": {"id": "zstd", "level": 0, "checksum": False},
-                    "bytes": {"id": "zstd", "level": 0, "checksum": False},
-                },
+                "v2_default_compressor": {"default": {"id": "zstd", "level": 0, "checksum": False}},
                 "v2_default_filters": {
-                    "numeric": None,
-                    "string": [{"id": "vlen-utf8"}],
-                    "bytes": [{"id": "vlen-bytes"}],
-                    "raw": None,
+                    "default": None,
+                    "numpy.variable_length_unicode_string": [{"id": "vlen-utf8"}],
+                    "numpy.fixed_length_unicode_string": [{"id": "vlen-utf8"}],
+                    "r*": [{"id": "vlen-bytes"}],
                 },
-                "v3_default_filters": {"boolean": [], "numeric": [], "string": [], "bytes": []},
+                "v3_default_filters": defaultdict(list),
                 "v3_default_serializer": {
-                    "boolean": {"name": "bytes", "configuration": {"endian": "little"}},
-                    "numeric": {"name": "bytes", "configuration": {"endian": "little"}},
-                    "string": {"name": "vlen-utf8"},
-                    "bytes": {"name": "vlen-bytes"},
+                    "default": {"name": "bytes", "configuration": {"endian": "little"}},
+                    "numpy.variable_length_unicode_string": [{"name": "vlen-utf8"}],
+                    "numpy.fixed_length_unicode_string": [{"name": "vlen-utf8"}],
+                    "r*": {"name": "vlen-bytes"},
                 },
                 "v3_default_compressors": {
-                    "boolean": [
-                        {"name": "zstd", "configuration": {"level": 0, "checksum": False}},
-                    ],
-                    "numeric": [
-                        {"name": "zstd", "configuration": {"level": 0, "checksum": False}},
-                    ],
-                    "string": [
-                        {"name": "zstd", "configuration": {"level": 0, "checksum": False}},
-                    ],
-                    "bytes": [
+                    "default": [
                         {"name": "zstd", "configuration": {"level": 0, "checksum": False}},
-                    ],
+                    ]
                 },
             },
             "async": {"concurrency": 10, "timeout": None},

src/zarr/core/dtype/_numpy.py

@@ -627,7 +627,7 @@ def from_json_value(self, data: JSON, *, zarr_format: ZarrFormat) -> np.str_:
     @dataclass(frozen=True, kw_only=True)
     class VariableLengthString(DTypeWrapper[np.dtypes.StringDType, str]):
         dtype_cls = np.dtypes.StringDType
-        _zarr_v3_name = "numpy.vlen_string"
+        _zarr_v3_name = "numpy.variable_length_string"
 
         @classmethod
         def _from_dtype_unsafe(cls, dtype: np.dtypes.StringDType) -> Self:
@@ -658,7 +658,7 @@ def from_json_value(self, data: JSON, *, zarr_format: ZarrFormat) -> str:
     @dataclass(frozen=True, kw_only=True)
     class VariableLengthString(DTypeWrapper[np.dtypes.ObjectDType, str]):
         dtype_cls = np.dtypes.ObjectDType
-        _zarr_v3_name = "numpy.vlen_string"
+        _zarr_v3_name = "numpy.variable_length_string"
 
         @classmethod
         def _from_dtype_unsafe(cls, dtype: np.dtypes.ObjectDType) -> Self: