Merge pull request #20037 from rapidsai/branch-25.10

Forward-merge branch-25.10 into branch-25.12

Author: GPUtester

docs/cudf/source/user_guide/api_docs/dataframe.rst

@@ -246,6 +246,14 @@ Time Series-related
    DataFrame.shift
    DataFrame.resample
 
+
+Metadata
+~~~~~~~~
+.. autosummary::
+   :toctree: api/
+
+   DataFrame.attrs
+
 Serialization / IO / conversion
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 .. autosummary::

docs/cudf/source/user_guide/api_docs/series.rst

@@ -236,6 +236,13 @@ Time Series-related
    Series.shift
    Series.resample
 
+Metadata
+~~~~~~~~
+.. autosummary::
+   :toctree: api/
+
+   Series.attrs
+
 Accessors
 ---------
 

python/cudf/cudf/core/accessors/base_accessor.py

@@ -81,13 +81,16 @@ def _return_or_inplace(
                     return idx
                 else:
                     return self._parent._constructor_expanddim._from_data(
-                        data=table, index=self._parent.index
+                        data=table,
+                        index=self._parent.index,
+                        attrs=self._parent.attrs,
                     )
             elif isinstance(self._parent, cudf.Series):
                 return cudf.Series._from_column(
                     new_col,
                     name=self._parent.name,
                     index=self._parent.index if retain_index else None,
+                    attrs=self._parent.attrs,
                 )
             elif isinstance(self._parent, cudf.Index):
                 return cudf.Index._from_column(new_col, name=self._parent.name)

python/cudf/cudf/core/dataframe.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import copy
 import functools
 import inspect
 import itertools
@@ -1002,7 +1003,7 @@ def __init__(
 
         second_index = None
         second_columns = None
-
+        attrs = None
         if isinstance(data, (DataFrame, pd.DataFrame)):
             if isinstance(data, pd.DataFrame):
                 cols = {
@@ -1017,6 +1018,7 @@ def __init__(
             col_accessor = data._data
             index, second_index = data.index, index
             second_columns = columns
+            attrs = data.attrs
         elif isinstance(data, (Series, pd.Series)):
             if isinstance(data, pd.Series):
                 data = Series(data, nan_as_null=nan_as_null)
@@ -1197,7 +1199,7 @@ def __init__(
                 label_dtype=second_columns.dtype,
             )
 
-        super().__init__(col_accessor, index=index)
+        super().__init__(col_accessor, index=index, attrs=attrs)
         if second_index is not None:
             reindexed = self.reindex(index=second_index, copy=False)
             self._data = reindexed._data
@@ -1207,13 +1209,14 @@ def __init__(
             self._data = self.astype(dtype)._data
 
     @classmethod
-    def _from_data(
+    def _from_data(  # type: ignore[override]
         cls,
         data: MutableMapping,
         index: Index | None = None,
         columns: Any = None,
+        attrs: dict | None = None,
     ) -> Self:
-        out = super()._from_data(data=data, index=index)
+        out = super()._from_data(data=data, index=index, attrs=attrs)
         if columns is not None:
             out.columns = columns
         return out
@@ -1370,10 +1373,10 @@ def _getitem_preprocessed(
         inputs.
         """
         if col_is_scalar:
-            series = Series._from_data(ca, index=self.index)
+            series = Series._from_data(ca, index=self.index, attrs=self.attrs)
             return series._getitem_preprocessed(spec)
         if ca.names != self._column_names:
-            frame = self._from_data(ca, index=self.index)
+            frame = self._from_data(ca, index=self.index, attrs=self.attrs)
         else:
             frame = self
         if isinstance(spec, indexing_utils.MapIndexer):
@@ -1405,6 +1408,7 @@ def _getitem_preprocessed(
                 )
                 result.index = result_index
                 result.name = new_name
+                result._attrs = frame.attrs
                 return result
             except TypeError:
                 if get_option("mode.pandas_compatible"):
@@ -1501,7 +1505,9 @@ def __getitem__(self, arg):
                     and all(n == "" for n in out._column_names[0])
                 )
             ):
-                out = self._constructor_sliced._from_data(out._data)
+                out = self._constructor_sliced._from_data(
+                    out._data, attrs=self.attrs
+                )
                 out._data.multiindex = False
                 out.index = self.index
                 out.name = arg
@@ -3436,17 +3442,16 @@ def reset_index(
         allow_duplicates: bool = False,
         names: Hashable | Sequence[Hashable] | None = None,
     ):
+        data, index = self._reset_index(
+            level=level,
+            drop=drop,
+            col_level=col_level,
+            col_fill=col_fill,
+            allow_duplicates=allow_duplicates,
+            names=names,
+        )
         return self._mimic_inplace(
-            DataFrame._from_data(
-                *self._reset_index(
-                    level=level,
-                    drop=drop,
-                    col_level=col_level,
-                    col_fill=col_fill,
-                    allow_duplicates=allow_duplicates,
-                    names=names,
-                )
-            ),
+            DataFrame._from_data(data=data, index=index, attrs=self.attrs),
             inplace=inplace,
         )
 
@@ -4328,6 +4333,7 @@ def transpose(self) -> Self:
         result = type(self)._from_data(
             ColumnAccessor(dict(enumerate(result_columns)), verify=False),
             index=Index(index),
+            attrs=self.attrs,
         )
         # Set the old index as the new column names
         result.columns = self.index
@@ -5071,7 +5077,7 @@ def _func(x):  # pragma: no cover
             apply_sr = Series._from_column(col)
             result[name] = apply_sr.apply(_func)._column
 
-        return DataFrame._from_data(result, index=self.index)
+        return DataFrame._from_data(result, index=self.index, attrs=self.attrs)
 
     @_performance_tracking
     @applyutils.doc_applychunks()
@@ -5667,6 +5673,7 @@ def to_pandas(
 
         out_df = pd.DataFrame(out_data, index=out_index)
         out_df.columns = self._data.to_pandas_index
+        out_df.attrs = self.attrs
 
         return out_df
 
@@ -5720,6 +5727,7 @@ def from_pandas(cls, dataframe, nan_as_null=no_default):
             df = cls._from_data(data, index)
             # Checks duplicate columns and sets column metadata
             df.columns = dataframe.columns
+            df._attrs = copy.deepcopy(dataframe.attrs)
             return df
         else:
             raise TypeError(
@@ -6310,7 +6318,10 @@ def quantile(
             if q_is_number:
                 result = result.transpose()
                 return Series._from_column(
-                    result._columns[0], name=q, index=result.index
+                    result._columns[0],
+                    name=q,
+                    index=result.index,
+                    attrs=self.attrs,
                 )
         elif method == "single":
             # Ensure that qs is non-scalar so that we always get a column back.
@@ -6328,7 +6339,7 @@ def quantile(
                     if len(res) == 0:
                         res = column_empty(row_count=len(qs), dtype=ser.dtype)
                     result[k] = res
-            result = DataFrame._from_data(result)
+            result = DataFrame._from_data(result, attrs=self.attrs)
 
             if q_is_number and numeric_only:
                 result = result.fillna(np.nan).iloc[0]
@@ -6478,7 +6489,7 @@ def make_false_column_like_self():
             )
 
         # TODO: Update this logic to properly preserve MultiIndex columns.
-        return DataFrame._from_data(result, self.index)
+        return DataFrame._from_data(result, self.index, attrs=self.attrs)
 
     #
     # Stats
@@ -6590,6 +6601,7 @@ def count(self, axis=0, numeric_only=False):
                 ]
             ),
             index=Index(self._column_names),
+            attrs=self.attrs,
         )
 
     _SUPPORT_AXIS_LOOKUP = {
@@ -6639,12 +6651,14 @@ def _reduce(
             )
             source = self._get_columns_by_label(numeric_cols)
             if source.empty:
-                return Series(
+                res = Series(
                     index=self._data.to_pandas_index[:0]
                     if axis == 0
                     else source.index,
                     dtype="float64",
                 )
+                res._attrs = self._attrs
+                return res
         if (
             axis == 2
             and op in {"kurtosis", "skew"}
@@ -6746,7 +6760,7 @@ def _reduce(
                     new_dtype = get_dtype_of_same_kind(common_dtype, res_dtype)
                     res = res.astype(new_dtype)
 
-                return Series._from_column(res, index=idx)
+                return Series._from_column(res, index=idx, attrs=self.attrs)
 
     @_performance_tracking
     def _scan(
@@ -7028,10 +7042,13 @@ def _apply_cupy_method_axis_1(self, method, *args, **kwargs):
             result = as_column(result, dtype=result_dtype)
             if mask is not None:
                 result = result.set_mask(mask._column.as_mask())
-            return Series._from_column(result, index=self.index)
+            return Series._from_column(
+                result, index=self.index, attrs=self.attrs
+            )
         else:
             result_df = DataFrame(result, index=self.index)
             result_df._set_columns_like(prepared._data)
+            result_df._attrs = self.attrs
             return result_df
 
     @_performance_tracking
@@ -7652,7 +7669,9 @@ def unnamed_group_generator():
 
         # Construct the resulting dataframe / series
         if not has_unnamed_levels:
-            result = Series._from_column(stacked[0], index=new_index)
+            result = Series._from_column(
+                stacked[0], index=new_index, attrs=self.attrs
+            )
         else:
             if unnamed_level_values.nlevels == 1:
                 unnamed_level_values = unnamed_level_values.get_level_values(0)
@@ -7677,7 +7696,9 @@ def unnamed_group_generator():
                 unnamed_level_values.names,
             )
 
-            result = DataFrame._from_data(data, index=new_index)
+            result = DataFrame._from_data(
+                data, index=new_index, attrs=self.attrs
+            )
 
         if not future_stack and dropna:
             return result.dropna(how="all")
@@ -7724,6 +7745,7 @@ def cov(self, min_periods=None, ddof: int = 1, numeric_only: bool = False):
         cols = self._data.to_pandas_index
         df = DataFrame(cupy.asfortranarray(cov), index=cols)
         df._set_columns_like(self._data)
+        df._attrs = self.attrs
         return df
 
     def corr(
@@ -7770,6 +7792,7 @@ def corr(
         cols = self._data.to_pandas_index
         df = DataFrame(cupy.asfortranarray(corr), index=cols)
         df._set_columns_like(self._data)
+        df._attrs = self.attrs
         return df
 
     @_performance_tracking
@@ -8080,9 +8103,13 @@ def nunique(self, axis=0, dropna: bool = True) -> Series:
         if axis != 0:
             raise NotImplementedError("axis parameter is not supported yet.")
         counts = [col.distinct_count(dropna=dropna) for col in self._columns]
-        return self._constructor_sliced(
-            counts, index=self._data.to_pandas_index
+        res = self._constructor_sliced(
+            counts,
+            index=self._data.to_pandas_index,
+            dtype="float64" if len(counts) == 0 else None,
         )
+        res._attrs = self.attrs
+        return res
 
     def _sample_axis_1(
         self,