Skip to content

Fix invalid codegen for #7451. #7456

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
sklam merged 2 commits into from
Oct 5, 2021
Merged

Fix invalid codegen for #7451. #7456

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6562250
Add unit test to demonstrate #7451 failing case.
stuartarchibald Oct 5, 2021
f3755f2
Fix issue with negative axis
stuartarchibald Oct 5, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 6 additions & 1 deletion numba/np/arrayobj.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5545,6 +5545,11 @@ def _take_along_axis_impl(
# With addition of pre-broadcasting:
# https://github.com/numpy/numpy/issues/19704

# Wrap axis, it's used in tuple_setitem so must be (axis >= 0) to ensure
# the GEP is in bounds.
if axis < 0:
axis = arr.ndim + axis

# Broadcast the two arrays to matching shapes:
arr_shape = list(arr.shape)
arr_shape[axis] = 1
Expand Down Expand Up @@ -5620,7 +5625,7 @@ def take_along_axis_impl(arr, indices, axis):
check_is_integer(axis, "axis")
if not isinstance(axis, types.IntegerLiteral):
raise ValueError(
"axis must be a literal value (i.e. not an argument) for now"
"axis must be a literal value (i.e. not an argument)"
)
axis = axis.literal_value
if axis < 0:
Expand Down
19 changes: 12 additions & 7 deletions numba/tests/test_np_functions.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4461,14 +4461,19 @@ def test_take_along_axis_broadcasting(self):
arr = np.ones((3, 4, 1))
ai = np.ones((1, 2, 5), dtype=np.intp)

@njit
def check(a, i):
return np.take_along_axis(a, i, axis=1)
def gen(axis):
@njit
def impl(a, i):
return np.take_along_axis(a, i, axis)
return impl

expected = check.py_func(arr, ai)
actual = check(arr, ai)
self.assertPreciseEqual(expected, actual)
self.assertEqual(actual.shape, (3, 2, 5))
# Check same axis but expressed as positive/negative value
for i in (1, -2):
check = gen(i)
expected = check.py_func(arr, ai)
actual = check(arr, ai)
self.assertPreciseEqual(expected, actual)
self.assertEqual(actual.shape, (3, 2, 5))

def test_take_along_axis_exceptions(self):
arr2d = np.arange(8).reshape(2, 4)
Expand Down