Ensure view holds a reference to original array

Author: lshaw8317

CMakeLists.txt

@@ -50,7 +50,7 @@ else()
     include(FetchContent)
     FetchContent_Declare(blosc2
         GIT_REPOSITORY https://github.com/Blosc/c-blosc2
-        GIT_TAG ab7a0cc3b5ca3cf09c5ddaeb66ba1100c1986f23  # fix view
+        GIT_TAG ae52f60b0c7caeac68d461f1ae5b0b2d57060e6a  # v2.21.2
     )
     FetchContent_MakeAvailable(blosc2)
     include_directories("${blosc2_SOURCE_DIR}/include")

src/blosc2/blosc2_ext.pyx

@@ -2357,9 +2357,10 @@ cdef class slice_flatter:
 cdef class NDArray:
     cdef b2nd_array_t* array
 
-    def __init__(self, array):
+    def __init__(self, array, base=None):
         self._dtype = None
         self.array = <b2nd_array_t *> PyCapsule_GetPointer(array, <char *> "b2nd_array_t*")
+        self.base = base # add reference to base if NDArray is a view
 
     @property
     def shape(self) -> tuple[int]:
@@ -2997,8 +2998,7 @@ def expand_dims(arr1: NDArray, axis_mask: list[bool], final_dims: int) -> blosc2
         mask_[i] = axis_mask[i]
     _check_rc(b2nd_expand_dims(arr1.array, &view, mask_, final_dims),"Error while expanding the arrays")
 
-    # Increase the reference count of the original array
-    Py_INCREF(arr1)
-
+    # create view with reference to arr1 to hold onto
+    new_base = arr1 if arr1.base is None else arr1.base
     return blosc2.NDArray(_schunk=PyCapsule_New(view.sc, <char *> "blosc2_schunk*", NULL),
-                          _array=PyCapsule_New(view, <char *> "b2nd_array_t*", NULL))
+                          _array=PyCapsule_New(view, <char *> "b2nd_array_t*", NULL), _base=new_base)

src/blosc2/ndarray.py

@@ -1170,10 +1170,12 @@ def __next__(self):
 class NDArray(blosc2_ext.NDArray, Operand):
     def __init__(self, **kwargs):
         self._schunk = SChunk(_schunk=kwargs["_schunk"], _is_view=True)  # SChunk Python instance
+        self.base = SChunk(_schunk=kwargs["_schunk"], _is_view=True)  # SChunk Python instance
         self._keep_last_read = False
         # Where to store the last read data
         self._last_read = {}
-        super().__init__(kwargs["_array"])
+        base = kwargs.pop("_base", None)
+        super().__init__(kwargs["_array"], base=base)
         # Accessor to fields
         self._fields = {}
         if self.dtype.fields:

tests/ndarray/test_resize.py

@@ -42,16 +42,21 @@ def test_resize(shape, new_shape, chunks, blocks, fill_value):
 )
 def test_expand_dims(shape, axis, chunks, blocks, fill_value):
     a = blosc2.full(shape, fill_value=fill_value, chunks=chunks, blocks=blocks)
-
+    npa = a[:]
     b = blosc2.expand_dims(a, axis=axis)
-    npa = np.expand_dims(a[:], axis)
-    assert npa.shape == b.shape
-    np.testing.assert_array_equal(npa, b[:])
+    npb = np.expand_dims(npa, axis)
+    assert npb.shape == b.shape
+    np.testing.assert_array_equal(npb, b[:])
 
     # Repeated expansion
     axis = (axis,) if isinstance(axis, int) else axis
     axis = axis[0] if (len(axis) + b.ndim) > blosc2.MAX_DIM else axis
     b = blosc2.expand_dims(b, axis=axis)
+    npb = np.expand_dims(npb, axis)
+    assert npb.shape == b.shape
+    np.testing.assert_array_equal(npb, b[:])
+
+    # Check that handling of views is correct
+    a = blosc2.expand_dims(a, axis=axis)  # could lose ref to original array and thus dealloc data
     npa = np.expand_dims(npa, axis)
-    assert npa.shape == b.shape
-    np.testing.assert_array_equal(npa, b[:])
+    assert a[()].shape == npa[()].shape  # getitem fails if deallocate has happened