Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly
Appearance settings

Commit a2a562b

Browse filesBrowse files
hameerabbasihameerabbasi
committed
Add __binsparse_descriptor__ and __binsparse_dlpack__.
1 parent fb0affe commit a2a562b
Copy full SHA for a2a562b

File tree

4 files changed

+106
-1
lines changed
Filter options

4 files changed

+106
-1
lines changed

‎sparse/numba_backend/_compressed/compressed.py

Copy file name to clipboardExpand all lines: sparse/numba_backend/_compressed/compressed.py
+37-1Lines changed: 37 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -844,6 +844,15 @@ def isinf(self):
844844
def isnan(self):
845845
return self.tocoo().isnan().asformat("gcxs", compressed_axes=self.compressed_axes)
846846

847+
# `GCXS` is a reshaped/transposed `CSR`, but it can't (usually)
848+
# be expressed in the `binsparse` 0.1 language.
849+
# We are missing index maps.
850+
def __binsparse_descriptor__(self) -> dict:
851+
return super().__binsparse_descriptor__()
852+
853+
def __binsparse_dlpack__(self) -> dict[str, np.ndarray]:
854+
return super().__binsparse_dlpack__()
855+
847856

848857
class _Compressed2d(GCXS):
849858
class_compressed_axes: tuple[int]
@@ -883,6 +892,34 @@ def from_numpy(cls, x, fill_value=0, idx_dtype=None):
883892
coo = COO.from_numpy(x, fill_value=fill_value, idx_dtype=idx_dtype)
884893
return cls.from_coo(coo, cls.class_compressed_axes, idx_dtype)
885894

895+
def __binsparse_descriptor__(self) -> dict:
896+
from sparse._version import __version__
897+
898+
data_dt = str(self.data.dtype)
899+
if np.issubdtype(data_dt, np.complexfloating):
900+
data_dt = f"complex[float{self.data.dtype.itemsize // 2}]"
901+
return {
902+
"binsparse": {
903+
"version": "0.1",
904+
"format": self.format.upper(),
905+
"shape": list(self.shape),
906+
"number_of_stored_values": self.nnz,
907+
"data_types": {
908+
"pointers_to_1": str(self.indices.dtype),
909+
"indices_1": str(self.indptr.dtype),
910+
"values": data_dt,
911+
},
912+
},
913+
"original_source": f"`sparse`, version {__version__}",
914+
}
915+
916+
def __binsparse_dlpack__(self) -> dict[str, np.ndarray]:
917+
return {
918+
"pointers_to_1": self.indices,
919+
"indices_1": self.indptr,
920+
"values": self.data,
921+
}
922+
886923

887924
class CSR(_Compressed2d):
888925
"""
@@ -915,7 +952,6 @@ def transpose(self, axes: None = None, copy: bool = False) -> Union["CSC", "CSR"
915952
return self
916953
return CSC((self.data, self.indices, self.indptr), self.shape[::-1])
917954

918-
919955
class CSC(_Compressed2d):
920956
"""
921957
The CSC or CCS scheme stores a n-dimensional array using n+1 one-dimensional arrays.

‎sparse/numba_backend/_coo/core.py

Copy file name to clipboardExpand all lines: sparse/numba_backend/_coo/core.py
+38Lines changed: 38 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1537,6 +1537,44 @@ def isnan(self):
15371537
prune=True,
15381538
)
15391539

1540+
def __binsparse_descriptor__(self) -> dict:
1541+
from sparse._version import __version__
1542+
1543+
data_dt = str(self.data.dtype)
1544+
if np.issubdtype(data_dt, np.complexfloating):
1545+
data_dt = f"complex[float{self.data.dtype.itemsize // 2}]"
1546+
return {
1547+
"binsparse": {
1548+
"version": "0.1",
1549+
"format": {
1550+
"custom": {
1551+
"level": {
1552+
"level_desc": "sparse",
1553+
"rank": self.ndim,
1554+
"level": {
1555+
"level_desc": "element",
1556+
},
1557+
}
1558+
}
1559+
},
1560+
"shape": list(self.shape),
1561+
"number_of_stored_values": self.nnz,
1562+
"data_types": {
1563+
"pointers_to_1": "uint8",
1564+
"indices_1": str(self.coords.dtype),
1565+
"values": data_dt,
1566+
},
1567+
},
1568+
"original_source": f"`sparse`, version {__version__}",
1569+
}
1570+
1571+
def __binsparse_dlpack__(self) -> dict[str, np.ndarray]:
1572+
return {
1573+
"pointers_to_1": np.array([0, self.nnz], dtype=np.uint8),
1574+
"indices_1": self.coords,
1575+
"values": self.data,
1576+
}
1577+
15401578

15411579
def as_coo(x, shape=None, fill_value=None, idx_dtype=None):
15421580
"""

‎sparse/numba_backend/_dok.py

Copy file name to clipboardExpand all lines: sparse/numba_backend/_dok.py
+6Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -548,6 +548,12 @@ def reshape(self, shape, order="C"):
548548

549549
return DOK.from_coo(self.to_coo().reshape(shape))
550550

551+
def __binsparse_descriptor__(self) -> dict:
552+
raise RuntimeError("`DOK` doesn't support the `__binsparse_descriptor__` protocol.")
553+
554+
def __binsparse_dlpack__(self) -> dict[str, np.ndarray]:
555+
raise RuntimeError("`DOK` doesn't support the `__binsparse_dlpack__` protocol.")
556+
551557

552558
def to_slice(k):
553559
"""Convert integer indices to one-element slices for consistency"""

‎sparse/numba_backend/_sparse_array.py

Copy file name to clipboardExpand all lines: sparse/numba_backend/_sparse_array.py
+25Lines changed: 25 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -218,6 +218,31 @@ def _str_impl(self, summary):
218218
except (ImportError, ValueError):
219219
return summary
220220

221+
@abstractmethod
222+
def __binsparse_descriptor__(self) -> dict:
223+
"""Return a `dict` equivalent to a parsed JSON [`binsparse` descriptor](https://graphblas.org/binsparse-specification/#descriptor)
224+
of this array.
225+
226+
Returns
227+
-------
228+
dict
229+
Parsed `binsparse` descriptor.
230+
"""
231+
raise NotImplementedError
232+
233+
@abstractmethod
234+
def __binsparse_dlpack__(self) -> dict[str, np.ndarray]:
235+
"""A `dict` containing the constituent arrays of this sparse array. The keys are compatible with the
236+
[`binsparse`](https://graphblas.org/binsparse-specification/) scheme, and the values are [`__dlpack__`](https://data-apis.org/array-api/latest/API_specification/generated/array_api.array.__dlpack__.html)
237+
compatible objects.
238+
239+
Returns
240+
-------
241+
dict[str, np.ndarray]
242+
The constituent arrays.
243+
"""
244+
raise NotImplementedError
245+
221246
@abstractmethod
222247
def asformat(self, format):
223248
"""

0 commit comments

Comments
0 (0)
Morty Proxy This is a proxified and sanitized view of the page, visit original site.