Fix fetching boolean data as binary

h5grove/utils.py

@@ -208,25 +208,21 @@ def get_type_metadata(type_id: h5py.h5t.TypeID) -> TypeMetadata:
 
 
 def _sanitize_dtype(dtype: np.dtype) -> np.dtype:
-    """Sanitize numpy dtype to one with a matching typed array in modern JavaScript.
-
-    :raises ValueError: If trying to sanitize a non-numeric numpy dtype
-    """
-    if dtype.kind not in ("f", "i", "u"):
-        raise ValueError(f"Unsupported numpy dtype `{dtype}`. Expected numeric dtype.")
-
-    # Convert to little endian
-    result = dtype.newbyteorder("<")
+    """Sanitize numpy dtype to one with a matching typed array in modern JavaScript."""
 
     # Convert float16 to float32
-    if result.kind == "f" and result.itemsize < 4:
+    if dtype.kind == "f" and dtype.itemsize < 4:
         return np.dtype("<f4")
 
     # Convert float128 to float64 (unavoidable loss of precision)
-    if result.kind == "f" and result.itemsize > 8:
+    if dtype.kind == "f" and dtype.itemsize > 8:
         return np.dtype("<f8")
 
-    return result
+    # Convert native and big-endian byte-orders to little-endian
+    if dtype.byteorder == "=" or dtype.byteorder == ">":
+        return dtype.newbyteorder("<")
+
+    return dtype
 
 
 T = TypeVar("T", np.ndarray, np.number, np.bool_)

test/base_test.py

@@ -140,6 +140,27 @@ def test_data_on_slice_with_format_and_flatten(self, server, format_arg):
 
         assert retrieved_data - data[100, 0] < 1e-8
 
+    def test_data_on_bool(self, server):
+        """Test /data/ endpoint on boolean dataset with format=bin"""
+        tested_h5entity_path = "/bool"
+        data = np.array([True, False, True, True])
+
+        filename = "test.h5"
+        with h5py.File(server.served_directory / filename, mode="w") as h5file:
+            dset = h5file.create_dataset(tested_h5entity_path, data=data)
+            dtype = dset.dtype
+            shape = dset.shape
+
+        response = server.get(
+            f"/data/?{urlencode({'file': filename, 'path': tested_h5entity_path, 'format': 'bin'})}"
+        )
+
+        content_type = response.find_header_value("content-type")
+        assert content_type == "application/octet-stream"
+
+        retrieved_data = decode_array_response(response, "bin", dtype.str, shape)
+        assert np.array_equal(retrieved_data, data)
+
     def test_data_on_opaque(self, server):
         """Test /data/ endpoint on opaque dataset with format=bin"""
         tested_h5entity_path = "/opaque"