Skip to content

Fix chop_threshold crashes in the pandas-free complex repr #620

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

Open
eriknw wants to merge 2 commits into from
Open

Fix chop_threshold crashes in the pandas-free complex repr #620

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
eb5878e
Fix chop_threshold crashes in the pandas-free complex repr
eriknw Aug 5, 2026
d09186f
Skip the complex chop_threshold test when FC64 is unavailable
eriknw Aug 6, 2026
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
13 changes: 12 additions & 1 deletion graphblas/core/formatting.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -443,7 +443,18 @@ def base(v):
return base

def formatter(v):
return base(v) if abs(v) > threshold else base(0.0)
try:
mag = abs(v)
except OverflowError:
# abs() of a python complex overflows for finite components near
# the float max, where numpy's abs returns inf. pandas computes the
# magnitude with numpy, so such a value is never chopped.
return base(v)
# Chop to the value's own type of zero. pandas chops at the array
# level, so a chopped complex renders " 0.000000+0.000000j"; a bare
# float zero here would put a j-less string into the complex column,
# which _trim_zeros_complex cannot parse.
return base(v) if mag > threshold else base(type(v)(0))

return formatter

Expand Down
30 changes: 30 additions & 0 deletions graphblas/tests/test_formatting.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4931,3 +4931,33 @@ def test_vector_as_matrix():
"</table>\n"
"</div></details></div>"
)


@pytest.mark.skipif("not pd")
@pytest.mark.skipif("not dtypes._supports_complex")
def test_chop_threshold_complex():
# The COO (long) form formats complex columns through the float column
# formatter, where display.chop_threshold compares each value's magnitude.
big = float(np.finfo(np.float64).max)
v = Vector.from_coo(
[10_000, 500_000], [complex(big, big), complex(1e-7, 0)], size=10**6, dtype="FC64"
)
with pd.option_context("display.chop_threshold", 1e-8):
# abs() of a python complex overflows for components near the float
# max; numpy's abs is inf, so nothing chops and this must render.
# Byte-identical to the pandas renderer, verified against 6f1eb02.
r = repr(v)
assert "1.797693e+308+1.797693e+308j" in r
assert "1.000000e-07+ 0.000000e+00j" in r
v._repr_html_()
# A value below the threshold chops to complex zero. pandas 3.0.3
# itself crashes on this input (its own _trim_zeros_complex cannot
# parse the float zero its chop substitutes into a complex column),
# so rendering at all is deliberately better than byte-identity.
w = Vector.from_coo(
[10_000, 500_000], [complex(1.0, 1.0), complex(5e-9, 0)], size=10**6, dtype="FC64"
)
r = repr(w)
assert "1.000000e+00+1.000000e+00j" in r
assert "0.000000e+00+0.000000e+00j" in r
w._repr_html_()
Loading