fix: don't use np.asarray on Index or Content objects (#2740)

* test: update tests

* test: cleanup more test lines!

* fix: asarray usage

* fix: re-evaluate!

src/awkward/_connect/numexpr.py

@@ -91,11 +91,14 @@ def action(inputs, **ignore):
             or not isinstance(x, ak.contents.Content)
             for x in inputs
         ):
+            input_primitives = [
+                x.data if isinstance(x, ak.contents.NumpyArray) else x for x in inputs
+            ]
             return (
                 ak.contents.NumpyArray(
                     numexpr.evaluate(
                         expression,
-                        dict(zip(names, inputs)),
+                        dict(zip(names, input_primitives)),
                         {},
                         order=order,
                         casting=casting,
@@ -138,8 +141,13 @@ def action(inputs, **ignore):
             or not isinstance(x, ak.contents.Content)
             for x in inputs
         ):
+            input_primitives = [
+                x.data if isinstance(x, ak.contents.NumpyArray) else x for x in inputs
+            ]
             return (
-                ak.contents.NumpyArray(numexpr.re_evaluate(dict(zip(names, inputs)))),
+                ak.contents.NumpyArray(
+                    numexpr.re_evaluate(dict(zip(names, input_primitives)))
+                ),
             )
         else:
             return None

src/awkward/_connect/numpy.py

@@ -220,8 +220,8 @@ def _array_ufunc_categorical(
         assert method == "__call__"
         one, two = inputs
 
-        one_index = numpy.asarray(one.index)
-        two_index = numpy.asarray(two.index)
+        one_index = numpy.asarray(one.index.data)
+        two_index = numpy.asarray(two.index.data)
         one_content = wrap_layout(one.content, behavior)
         two_content = wrap_layout(two.content, behavior)
 
@@ -280,12 +280,17 @@ def _array_ufunc_string_likes(
         right = ak.without_parameters(right, highlevel=False)
 
         # first condition: string lengths must be the same
-        counts1 = nplike.asarray(
-            ak._do.reduce(left, ak._reducers.Count(), axis=-1, mask=False)
+        left_counts_layout = ak._do.reduce(
+            left, ak._reducers.Count(), axis=-1, mask=False
         )
-        counts2 = nplike.asarray(
-            ak._do.reduce(right, ak._reducers.Count(), axis=-1, mask=False)
+        assert left_counts_layout.is_numpy
+        right_counts_layout = ak._do.reduce(
+            right, ak._reducers.Count(), axis=-1, mask=False
         )
+        assert right_counts_layout.is_numpy
+
+        counts1 = nplike.asarray(left_counts_layout.data)
+        counts2 = nplike.asarray(right_counts_layout.data)
 
         out = counts1 == counts2
 

src/awkward/contents/emptyarray.py

@@ -151,9 +151,6 @@ def to_NumpyArray(self, dtype, backend=None):
             backend=backend,
         )
 
-    def __array__(self, **kwargs):
-        return numpy.empty(0, dtype=np.float64)
-
     def __iter__(self):
         return iter([])
 

src/awkward/contents/indexedoptionarray.py

@@ -1518,7 +1518,7 @@ def _pad_none(self, target, axis, depth, clip):
             )
 
     def _to_arrow(self, pyarrow, mask_node, validbytes, length, options):
-        index = numpy.asarray(self._index, copy=True)
+        index = numpy.asarray(self._index.data, copy=True)
         this_validbytes = self.mask_as_bool(valid_when=True)
         index[~this_validbytes] = 0
 

src/awkward/contents/numpyarray.py

@@ -273,9 +273,6 @@ def to_RegularArray(self):
     def maybe_to_NumpyArray(self) -> Self:
         return self
 
-    def __array__(self, dtype=None):
-        return self._backend.nplike.asarray(self._data, dtype=dtype)
-
     def __iter__(self):
         return iter(self._data)
 

src/awkward/operations/ak_concatenate.py

@@ -264,8 +264,7 @@ def action(inputs, depth, **kwargs):
 
                 for x in nextinputs:
                     o, f = x._offsets_and_flattened(1, 1)
-                    o = backend.index_nplike.asarray(o)
-                    c = o[1:] - o[:-1]
+                    c = o.data[1:] - o.data[:-1]
                     backend.index_nplike.add(counts, c, maybe_out=counts)
                     all_counts.append(c)
                     all_flatten.append(f)

src/awkward/operations/ak_from_json.py

@@ -430,8 +430,8 @@ def action(node, **kwargs):
                     ):
                         with numpy._module.errstate(invalid="ignore"):
                             return ak.contents.NumpyArray(
-                                node.backend.nplike.asarray(real)
-                                + node.backend.nplike.asarray(imag) * 1j
+                                node.backend.nplike.asarray(real.data)
+                                + node.backend.nplike.asarray(imag.data) * 1j
                             )
                     else:
                         raise ValueError(

src/awkward/operations/ak_mask.py

@@ -100,8 +100,8 @@ def mask(array, mask, *, valid_when=True, highlevel=True, behavior=None):
 def _impl(array, mask, valid_when, highlevel, behavior):
     def action(inputs, backend, **kwargs):
         layoutarray, layoutmask = inputs
-        if isinstance(layoutmask, ak.contents.NumpyArray):
-            m = backend.nplike.asarray(layoutmask)
+        if layoutmask.is_numpy:
+            m = backend.nplike.asarray(layoutmask.data)
             if not issubclass(m.dtype.type, (bool, np.bool_)):
                 raise ValueError(f"mask must have boolean type, not {m.dtype!r}")
             bytemask = ak.index.Index8(m.view(np.int8))

src/awkward/operations/ak_run_lengths.py

@@ -156,10 +156,12 @@ def action(layout, **kwargs):
                 nextcontent, _ = lengths_of(ak.highlevel.Array(layout), None)
                 return ak.contents.NumpyArray(nextcontent)
 
-            if not isinstance(layout, (ak.contents.NumpyArray, ak.contents.EmptyArray)):
+            if layout.is_unknown:
+                layout = layout.to_NumpyArray(np.float64)
+            elif not layout.is_numpy:
                 raise NotImplementedError("run_lengths on " + type(layout).__name__)
 
-            nextcontent, _ = lengths_of(backend.nplike.asarray(layout), None)
+            nextcontent, _ = lengths_of(backend.nplike.asarray(layout.data), None)
             return ak.contents.NumpyArray(nextcontent)
 
         elif layout.branch_depth == (False, 2):
@@ -197,9 +199,9 @@ def action(layout, **kwargs):
             if content.is_indexed:
                 content = content.project()
 
-            if not isinstance(
-                content, (ak.contents.NumpyArray, ak.contents.EmptyArray)
-            ):
+            if content.is_unknown:
+                content = content.to_NumpyArray(np.float64)
+            elif not content.is_numpy:
                 raise NotImplementedError(
                     "run_lengths on "
                     + type(layout).__name__
@@ -208,7 +210,7 @@ def action(layout, **kwargs):
                 )
 
             nextcontent, nextoffsets = lengths_of(
-                backend.nplike.asarray(content), offsets - offsets[0]
+                backend.nplike.asarray(content.data), offsets - offsets[0]
             )
             return ak.contents.ListOffsetArray(
                 ak.index.Index64(nextoffsets),

src/awkward/operations/ak_where.py

@@ -105,7 +105,7 @@ def _impl3(condition, x, y, mergebool, highlevel, behavior):
     def action(inputs, **kwargs):
         condition, x, y = inputs
         if isinstance(condition, ak.contents.NumpyArray):
-            npcondition = backend.index_nplike.asarray(condition)
+            npcondition = backend.index_nplike.asarray(condition.data)
             tags = ak.index.Index8((npcondition == 0).view(np.int8))
             index = ak.index.Index64(
                 backend.index_nplike.arange(tags.length, dtype=np.int64),

tests/test_0002_minimal_listarray.py

@@ -10,7 +10,7 @@ def test():
     data = np.array([0, 2, 2, 3], dtype="i8")
     offsets = ak.index.Index64(data)
 
-    assert np.asarray(offsets).tolist() == [0, 2, 2, 3]
+    assert np.asarray(offsets.data).tolist() == [0, 2, 2, 3]
     assert offsets[0] == 0
     assert offsets[1] == 2
     assert offsets[2] == 2
@@ -21,7 +21,7 @@ def test():
     data = np.array([0, 2, 2, 3], dtype="i4")
     offsets = ak.index.Index32(data)
 
-    assert np.asarray(offsets).tolist() == [0, 2, 2, 3]
+    assert np.asarray(offsets.data).tolist() == [0, 2, 2, 3]
     assert offsets[0] == 0
     assert offsets[1] == 2
     assert offsets[2] == 2
@@ -34,12 +34,12 @@ def test():
     offsets = ak.index.Index32(data)
     array = ak.contents.ListOffsetArray(offsets, content)
 
-    assert np.asarray(content).tolist() == [[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]
-    assert np.asarray(content[0]).tolist() == [0, 1, 2, 3]
+    assert ak.to_numpy(content).tolist() == [[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]
+    assert ak.to_numpy(content[0]).tolist() == [0, 1, 2, 3]
     assert content.to_typetracer()[0].form == content[0].form
-    assert np.asarray(content[1]).tolist() == [4, 5, 6, 7]
+    assert ak.to_numpy(content[1]).tolist() == [4, 5, 6, 7]
     assert content.to_typetracer()[1].form == content[1].form
-    assert np.asarray(content[2]).tolist() == [8, 9, 10, 11]
+    assert ak.to_numpy(content[2]).tolist() == [8, 9, 10, 11]
     assert content.to_typetracer()[2].form == content[2].form
     assert [content[i][j] for i in range(3) for j in range(4)] == [
         0,
@@ -56,17 +56,17 @@ def test():
         11,
     ]
 
-    assert np.asarray(array[0]).tolist() == [[0, 1, 2, 3], [4, 5, 6, 7]]
+    assert ak.to_numpy(array[0]).tolist() == [[0, 1, 2, 3], [4, 5, 6, 7]]
     assert array.to_typetracer()[0].form == array[0].form
-    assert np.asarray(array[1]).tolist() == []
+    assert ak.to_numpy(array[1]).tolist() == []
     assert array.to_typetracer()[1].form == array[1].form
-    assert np.asarray(array[2]).tolist() == [[8, 9, 10, 11]]
+    assert ak.to_numpy(array[2]).tolist() == [[8, 9, 10, 11]]
     assert array.to_typetracer()[2].form == array[2].form
-    assert np.asarray(array[1:3][0]).tolist() == []
+    assert ak.to_numpy(array[1:3][0]).tolist() == []
     assert array.to_typetracer()[1:3][0].form == array[1:3][0].form
-    assert np.asarray(array[1:3][1]).tolist() == [[8, 9, 10, 11]]
+    assert ak.to_numpy(array[1:3][1]).tolist() == [[8, 9, 10, 11]]
     assert array.to_typetracer()[1:3][1].form == array[1:3][1].form
-    assert np.asarray(array[2:3][0]).tolist() == [[8, 9, 10, 11]]
+    assert ak.to_numpy(array[2:3][0]).tolist() == [[8, 9, 10, 11]]
     assert array.to_typetracer()[2:3][0].form == array[2:3][0].form
 
 
@@ -85,16 +85,16 @@ def test_members():
     array = ak.contents.ListOffsetArray(offsets, content)
     new = ak.contents.ListOffsetArray(offsets, array)
 
-    assert np.asarray(array.offsets).tolist() == [0, 2, 2, 3]
-    assert np.asarray(array.content).tolist() == [
+    assert np.asarray(array.offsets.data).tolist() == [0, 2, 2, 3]
+    assert ak.to_numpy(array.content).tolist() == [
         [0, 1, 2, 3],
         [4, 5, 6, 7],
         [8, 9, 10, 11],
     ]
 
-    assert np.asarray(new.offsets).tolist() == [0, 2, 2, 3]
-    assert np.asarray(new.content.offsets).tolist() == [0, 2, 2, 3]
-    assert np.asarray(new.content.content).tolist() == [
+    assert np.asarray(new.offsets.data).tolist() == [0, 2, 2, 3]
+    assert np.asarray(new.content.offsets.data).tolist() == [0, 2, 2, 3]
+    assert ak.to_numpy(new.content.content).tolist() == [
         [0, 1, 2, 3],
         [4, 5, 6, 7],
         [8, 9, 10, 11],

tests/test_0006_deep_iteration.py

@@ -12,4 +12,4 @@ def test_iterator():
     array = ak.contents.ListOffsetArray(offsets, content)
 
     assert list(content) == [1.1, 2.2, 3.3]
-    assert [np.asarray(x).tolist() for x in array] == [[1.1, 2.2], [], [3.3]]
+    assert [ak.to_numpy(x).tolist() for x in array] == [[1.1, 2.2], [], [3.3]]

tests/test_0074_argsort_and_sort.py

@@ -117,18 +117,18 @@ def test_NumpyArray():
 
     assert to_list(
         ak.operations.sort(array2, axis=1, ascending=True, stable=False)
-    ) == to_list(np.sort(np.asarray(array2), axis=1))
+    ) == to_list(np.sort(ak.to_numpy(array2), axis=1))
     assert to_list(
         ak.operations.sort(array2, axis=0, ascending=True, stable=False)
-    ) == to_list(np.sort(np.asarray(array2), axis=0))
+    ) == to_list(np.sort(ak.to_numpy(array2), axis=0))
 
     assert to_list(
         ak.operations.argsort(array2, axis=1, ascending=True, stable=False)
-    ) == to_list(np.argsort(np.asarray(array2), 1))
+    ) == to_list(np.argsort(ak.to_numpy(array2), 1))
 
     assert to_list(
         ak.operations.argsort(array2, axis=0, ascending=True, stable=False)
-    ) == to_list(np.argsort(np.asarray(array2), 0))
+    ) == to_list(np.argsort(ak.to_numpy(array2), 0))
 
     with pytest.raises(ValueError) as err:
         ak.operations.argsort(array2, axis=2, ascending=True, stable=False)
@@ -344,20 +344,20 @@ def test_3d():
     )  # 5
 
     sorted = ak.operations.argsort(array, axis=1, ascending=True, stable=False)
-    assert to_list(sorted) == to_list(np.argsort(np.asarray(array), 1))
+    assert to_list(sorted) == to_list(np.argsort(np.asarray(array.data), 1))
 
     sorted = ak.operations.argsort(array, axis=2, ascending=True, stable=False)
-    assert to_list(sorted) == to_list(np.argsort(np.asarray(array), 2))
+    assert to_list(sorted) == to_list(np.argsort(np.asarray(array.data), 2))
 
     sorted = ak.operations.sort(array, axis=2, ascending=True, stable=False)
-    assert to_list(sorted) == to_list(np.sort(np.asarray(array), 2))
+    assert to_list(sorted) == to_list(np.sort(np.asarray(array.data), 2))
 
     sorted = ak.operations.argsort(array, axis=1, ascending=True, stable=False)
 
-    assert to_list(sorted) == to_list(np.argsort(np.asarray(array), 1))
+    assert to_list(sorted) == to_list(np.argsort(np.asarray(array.data), 1))
 
     sorted = ak.operations.sort(array, axis=1, ascending=True, stable=False)
-    assert to_list(sorted) == to_list(np.sort(np.asarray(array), 1))
+    assert to_list(sorted) == to_list(np.sort(np.asarray(array.data), 1))
 
     sorted = ak.operations.sort(array, axis=1, ascending=False, stable=False)
     assert to_list(sorted) == [
@@ -374,11 +374,11 @@ def test_3d():
     ]
 
     sorted = ak.operations.sort(array, axis=0, ascending=True, stable=False)
-    assert to_list(sorted) == to_list(np.sort(np.asarray(array), 0))
+    assert to_list(sorted) == to_list(np.sort(np.asarray(array.data), 0))
 
     assert to_list(
         ak.operations.argsort(array, axis=0, ascending=True, stable=False)
-    ) == to_list(np.argsort(np.asarray(array), 0))
+    ) == to_list(np.argsort(np.asarray(array.data), 0))
 
 
 def test_ByteMaskedArray():

tests/test_0093_simplify_uniontypes_and_optiontypes.py

@@ -60,11 +60,11 @@ def test_numpyarray_merge():
             one = ak.contents.NumpyArray(np.array([1, 2, 3], dtype=x))
             two = ak.contents.NumpyArray(np.array([4, 5], dtype=y))
             three = one._mergemany([two])
-            assert np.asarray(three).dtype == np.dtype(z), "{} {} {} {}".format(
-                x, y, z, np.asarray(three).dtype.type
+            assert ak.to_numpy(three).dtype == np.dtype(z), "{} {} {} {}".format(
+                x, y, z, ak.to_numpy(three).dtype.type
             )
             assert to_list(three) == to_list(
-                np.concatenate([np.asarray(one), np.asarray(two)])
+                np.concatenate([ak.to_numpy(one), ak.to_numpy(two)])
             )
             assert to_list(one._mergemany([emptyarray])) == to_list(one)
             assert to_list(emptyarray._mergemany([one])) == to_list(one)
@@ -927,7 +927,7 @@ def test_indexedarray_simplify():
     index2 = ak.index.Index64(np.array([2, 2, 1, 6, 5], dtype=np.int64))
 
     array2 = ak.contents.IndexedArray.simplified(index2, array)
-    assert np.asarray(array.index).tolist() == [0, 1, -1, 2, -1, -1, 3, 4]
+    assert np.asarray(array.index.data).tolist() == [0, 1, -1, 2, -1, -1, 3, 4]
     assert to_list(array2) == to_list(array2) == [None, None, "two", "four", None]
 
     assert array2.to_typetracer().form == array2.form