Skip to content
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

Jump to bottom

Fix groupby head/tail for empty dataframe #13398

Merged
merged 3 commits into from
May 22, 2023
Merged

Fix groupby head/tail for empty dataframe #13398

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
8126ed3
Fix groupby head/tail for empty dataframe
shwina May 20, 2023
eeb672b
Merge branch 'branch-23.06' of github.com:rapidsai/cudf into fix-grou…
shwina May 22, 2023
bce865d
Apply suggested patch
shwina May 22, 2023
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
2 changes: 1 addition & 1 deletion python/cudf/cudf/core/groupby/groupby.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -678,7 +678,7 @@ def _head_tail(self, n, *, take_head: bool, preserve_order: bool):
# subsample the gather map from the full input ordering,
# rather than permuting the gather map of the output.
_, (ordering,), _ = self._groupby.groups(
[arange(0, self.obj._data.nrows)]
[arange(0, len(self.obj))]
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The same error occurs (previously and after the fix for scans in #13389, which was me again sorry!). Although it's impossible to reach that code path right now because mimic pandas order for scans only happens if the dataframe is not empty.

But can you apply this patch too please?

diff --git a/python/cudf/cudf/core/groupby/groupby.py b/python/cudf/cudf/core/groupby/groupby.py
index fb242a49ad..b3be6d9de0 100644
--- a/python/cudf/cudf/core/groupby/groupby.py
+++ b/python/cudf/cudf/core/groupby/groupby.py
@@ -2277,9 +2277,7 @@ class GroupBy(Serializable, Reducible, Scannable):
         # result coming back from libcudf has null_count few rows than
         # the input, so we must produce an ordering from the full
         # input range.
-        _, (ordering,), _ = self._groupby.groups(
-            [arange(0, self.obj._data.nrows)]
-        )
+        _, (ordering,), _ = self._groupby.groups([arange(0, len(self.obj))])
         if self._dropna and any(
             c.has_nulls(include_nan=True) > 0
             for c in self.grouping._key_columns
@@ -2287,7 +2285,7 @@ class GroupBy(Serializable, Reducible, Scannable):
             # Scan aggregations with null/nan keys put nulls in the
             # corresponding output rows in pandas, to do that here
             # expand the result by reindexing.
-            ri = cudf.RangeIndex(0, self.obj._data.nrows)
+            ri = cudf.RangeIndex(0, len(self.obj))
             result.index = cudf.Index(ordering)
             # This reorders and expands
             result = result.reindex(ri)

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done

)
# Invert permutation from original order to groups on the
# subset of entries we want.
Expand Down
16 changes: 16 additions & 0 deletions python/cudf/cudf/tests/test_groupby.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3231,3 +3231,19 @@ def test_head_tail(self, df, n, take_head, expected, preserve_order):
else:
actual = df.groupby("a").tail(n=n, preserve_order=preserve_order)
assert_eq(actual, expected)


def test_head_tail_empty():
# GH #13397

values = [1, 2, 3]
pdf = pd.DataFrame({}, index=values)
df = cudf.DataFrame({}, index=values)

expected = pdf.groupby(pd.Series(values)).head()
got = df.groupby(cudf.Series(values)).head()
assert_eq(expected, got)

expected = pdf.groupby(pd.Series(values)).tail()
got = df.groupby(cudf.Series(values)).tail()
assert_eq(expected, got)