Bootstrap v3 branch with zarrita

zarr/v3/__init__.py

@@ -0,0 +1,40 @@
+from __future__ import annotations
+
+from typing import Union
+
+import zarr.v3.codecs  # noqa: F401
+from zarr.v3.array import Array  # noqa: F401
+from zarr.v3.array_v2 import ArrayV2  # noqa: F401
+from zarr.v3.group import Group  # noqa: F401
+from zarr.v3.group_v2 import GroupV2  # noqa: F401
+from zarr.v3.metadata import RuntimeConfiguration, runtime_configuration  # noqa: F401
+from zarr.v3.store import (  # noqa: F401
+    LocalStore,
+    RemoteStore,
+    Store,
+    StoreLike,
+    StorePath,
+    make_store_path,
+)
+from zarr.v3.sync import sync as _sync
+
+
+async def open_auto_async(
+    store: StoreLike,
+    runtime_configuration_: RuntimeConfiguration = RuntimeConfiguration(),
+) -> Union[Array, ArrayV2, Group, GroupV2]:
+    store_path = make_store_path(store)
+    try:
+        return await Group.open_or_array(store_path, runtime_configuration=runtime_configuration_)
+    except KeyError:
+        return await GroupV2.open_or_array(store_path, runtime_configuration_)
+
+
+def open_auto(
+    store: StoreLike,
+    runtime_configuration_: RuntimeConfiguration = RuntimeConfiguration(),
+) -> Union[Array, ArrayV2, Group, GroupV2]:
+    return _sync(
+        open_auto_async(store, runtime_configuration_),
+        runtime_configuration_.asyncio_loop,
+    )

zarr/v3/abc/array.py

@@ -0,0 +1,140 @@
+from __future__ import annotations
+from abc import abstractproperty, abstractmethod, ABC
+from typing import Tuple, Any, Dict
+
+import numpy as np
+
+from zarr.v3.abc.store import ReadStore, WriteStore
+from zarr.v3.common import Selection
+
+
+class BaseArray(ABC):
+    @abstractproperty
+    def store_path(self) -> str:  # TODO: rename to `path`?
+        """Path to this array in the underlying store."""
+        ...
+
+    @abstractproperty
+    def dtype(self) -> np.dtype:
+        """Data type of the array elements.
+
+        Returns
+        -------
+        dtype
+            array data type
+        """
+        ...
+
+    @abstractproperty
+    def ndim(self) -> int:
+        """Number of array dimensions (axes).
+
+        Returns
+        -------
+        int
+            number of array dimensions (axes)
+        """
+        ...
+
+    @abstractproperty
+    def shape(self) -> Tuple[int, ...]:
+        """Array dimensions.
+
+        Returns
+        -------
+        tuple of int
+            array dimensions
+        """
+        ...
+
+    @abstractproperty
+    def size(self) -> int:
+        """Number of elements in the array.
+
+        Returns
+        -------
+        int
+            number of elements in an array.
+        """
+
+    @abstractproperty
+    def attrs(self) -> Dict[str, Any]:
+        """Array attributes.
+
+        Returns
+        -------
+        dict
+            user defined attributes
+        """
+        ...
+
+    @abstractproperty
+    def info(self) -> Any:
+        """Report some diagnostic information about the array.
+
+        Returns
+        -------
+        out
+        """
+        ...
+
+
+class AsynchronousArray(BaseArray):
+    """This class can be implemented as a v2 or v3 array"""
+
+    @classmethod
+    @abstractmethod
+    async def from_json(cls, zarr_json: Any, store: ReadStore) -> AsynchronousArray:
+        ...
+
+    @classmethod
+    @abstractmethod
+    async def open(cls, store: ReadStore) -> AsynchronousArray:
+        ...
+
+    @classmethod
+    @abstractmethod
+    async def create(cls, store: WriteStore, *, shape, **kwargs) -> AsynchronousArray:
+        ...
+
+    @abstractmethod
+    async def getitem(self, selection: Selection):
+        ...
+
+    @abstractmethod
+    async def setitem(self, selection: Selection, value: np.ndarray) -> None:
+        ...
+
+
+class SynchronousArray(BaseArray):
+    """
+    This class can be implemented as a v2 or v3 array
+    """
+
+    @classmethod
+    @abstractmethod
+    def from_json(cls, zarr_json: Any, store: ReadStore) -> SynchronousArray:
+        ...
+
+    @classmethod
+    @abstractmethod
+    def open(cls, store: ReadStore) -> SynchronousArray:
+        ...
+
+    @classmethod
+    @abstractmethod
+    def create(cls, store: WriteStore, *, shape, **kwargs) -> SynchronousArray:
+        ...
+
+    @abstractmethod
+    def __getitem__(self, selection: Selection):  # TODO: type as np.ndarray | scalar
+        ...
+
+    @abstractmethod
+    def __setitem__(self, selection: Selection, value: np.ndarray) -> None:
+        ...
+
+    # some day ;)
+    # @property
+    # def __array_api_version__(self) -> str:
+    #     return "2022.12"

zarr/v3/abc/codec.py

@@ -0,0 +1,84 @@
+# Notes:
+# 1. These are missing methods described in the spec. I expected to see these method definitions:
+# def compute_encoded_representation_type(self, decoded_representation_type):
+# def encode(self, decoded_value):
+# def decode(self, encoded_value, decoded_representation_type):
+# def partial_decode(self, input_handle, decoded_representation_type, decoded_regions):
+# def compute_encoded_size(self, input_size):
+# 2. Understand why array metadata is included on all codecs
+
+
+from __future__ import annotations
+
+from abc import abstractmethod, ABC
+from typing import TYPE_CHECKING, Optional
+
+import numpy as np
+
+from zarr.v3.common import BytesLike
+
+
+if TYPE_CHECKING:
+    from zarr.v3.metadata import CoreArrayMetadata
+
+
+class Codec(ABC):
+    supports_partial_decode: bool
+    supports_partial_encode: bool
+    is_fixed_size: bool
+    array_metadata: CoreArrayMetadata
+
+    @abstractmethod
+    def compute_encoded_size(self, input_byte_length: int) -> int:
+        pass
+
+    def resolve_metadata(self) -> CoreArrayMetadata:
+        return self.array_metadata
+
+
+class ArrayArrayCodec(Codec):
+    @abstractmethod
+    async def decode(
+        self,
+        chunk_array: np.ndarray,
+    ) -> np.ndarray:
+        pass
+
+    @abstractmethod
+    async def encode(
+        self,
+        chunk_array: np.ndarray,
+    ) -> Optional[np.ndarray]:
+        pass
+
+
+class ArrayBytesCodec(Codec):
+    @abstractmethod
+    async def decode(
+        self,
+        chunk_array: BytesLike,
+    ) -> np.ndarray:
+        pass
+
+    @abstractmethod
+    async def encode(
+        self,
+        chunk_array: np.ndarray,
+    ) -> Optional[BytesLike]:
+        pass
+
+
+class BytesBytesCodec(Codec):
+    @abstractmethod
+    async def decode(
+        self,
+        chunk_array: BytesLike,
+    ) -> BytesLike:
+        pass
+
+    @abstractmethod
+    async def encode(
+        self,
+        chunk_array: BytesLike,
+    ) -> Optional[BytesLike]:
+        pass

zarr/v3/abc/group.py

@@ -0,0 +1,86 @@
+from __future__ import annotations
+
+from abc import abstractproperty, ABC
+from collections.abc import MutableMapping
+from typing import Dict, Any
+
+
+class BaseGroup(ABC):
+    @abstractproperty
+    def attrs(self) -> Dict[str, Any]:
+        """User-defined attributes."""
+        ...
+
+    @abstractproperty
+    def info(self) -> Any:  # TODO: type this later
+        """Return diagnostic information about the group."""
+        ...
+
+
+class AsynchronousGroup(BaseGroup):
+    pass
+    # TODO: (considering the following api)
+    # store_path (rename to path?)
+    # nchildren - number of child groups + arrays
+    # children (async iterator)
+    # contains - check if child exists
+    # getitem - get child
+    # group_keys (async iterator)
+    # groups (async iterator)
+    # array_keys (async iterator)
+    # arrays (async iterator)
+    # visit
+    # visitkeys
+    # visitvalues
+    # tree
+    # create_group
+    # require_group
+    # create_groups
+    # require_groups
+    # create_dataset
+    # require_dataset
+    # create
+    # empty
+    # zeros
+    # ones
+    # full
+    # array
+    # empty_like
+    # zeros_like
+    # ones_like
+    # full_like
+    # move
+
+
+class SynchronousGroup(BaseGroup, MutableMapping):
+    # TODO - think about if we want to keep the MutableMapping abstraction or
+    pass
+    # store_path (rename to path?)
+    # __enter__
+    # __exit__
+    # group_keys
+    # groups
+    # array_keys
+    # arrays
+    # visit
+    # visitkeys
+    # visitvalues
+    # visititems
+    # tree
+    # create_group
+    # require_group
+    # create_groups
+    # require_groups
+    # create_dataset
+    # require_dataset
+    # create
+    # empty
+    # zeros
+    # ones
+    # full
+    # array
+    # empty_like
+    # zeros_like
+    # ones_like
+    # full_like
+    # move