Disallow cudf.Index accepting column in favor of ._from_column (#16549)

Similar to #16454, this PR disallows the public `cudf.Index` accepting a private `ColumnBase` object in favor of `_from_column` (which was added in the linked PR)

Authors:
  - Matthew Roeschke (https://github.com/mroeschke)
  - GALI PREM SAGAR (https://github.com/galipremsagar)

Approvers:
  - GALI PREM SAGAR (https://github.com/galipremsagar)

URL: #16549

python/cudf/cudf/_lib/parquet.pyx

@@ -222,7 +222,7 @@ cdef object _process_metadata(object df,
                 if len(filtered_idx) > 0:
                     idx = cudf.concat(filtered_idx)
                 else:
-                    idx = cudf.Index(cudf.core.column.column_empty(0))
+                    idx = cudf.Index._from_column(cudf.core.column.column_empty(0))
             else:
                 start = range_index_meta["start"] + skip_rows
                 stop = range_index_meta["stop"]
@@ -240,7 +240,7 @@ cdef object _process_metadata(object df,
             index_data = df[index_col]
             actual_index_names = list(index_col_names.values())
             if len(index_data._data) == 1:
-                idx = cudf.Index(
+                idx = cudf.Index._from_column(
                     index_data._data.columns[0],
                     name=actual_index_names[0]
                 )

python/cudf/cudf/_lib/utils.pyx

@@ -93,12 +93,12 @@ cpdef generate_pandas_metadata(table, index):
     materialize_index = False
     if index is not False:
         for level, name in enumerate(table._index.names):
-            if isinstance(table._index, cudf.core.multiindex.MultiIndex):
+            if isinstance(table._index, cudf.MultiIndex):
                 idx = table.index.get_level_values(level)
             else:
                 idx = table.index
 
-            if isinstance(idx, cudf.core.index.RangeIndex):
+            if isinstance(idx, cudf.RangeIndex):
                 if index is None:
                     descr = {
                         "kind": "range",
@@ -110,7 +110,7 @@ cpdef generate_pandas_metadata(table, index):
                 else:
                     materialize_index = True
                     # When `index=True`, RangeIndex needs to be materialized.
-                    materialized_idx = cudf.Index(idx._values, name=idx.name)
+                    materialized_idx = idx._as_int_index()
                     descr = _index_level_name(
                         index_name=materialized_idx.name,
                         level=level,

python/cudf/cudf/api/types.py

@@ -249,7 +249,7 @@ def _union_categoricals(
             new_categories=sorted_categories
         )
 
-    return cudf.Index(result_col)
+    return cudf.CategoricalIndex._from_column(result_col)
 
 
 def is_bool_dtype(arr_or_dtype):

python/cudf/cudf/core/_base_index.py

@@ -1979,7 +1979,7 @@ def from_pandas(cls, index: pd.Index, nan_as_null=no_default):
                 name=index.name,
             )
         else:
-            return cudf.Index(
+            return cudf.Index._from_column(
                 column.as_column(index, nan_as_null=nan_as_null),
                 name=index.name,
             )

python/cudf/cudf/core/algorithms.py

@@ -8,7 +8,7 @@
 import numpy as np
 
 from cudf.core.column import as_column
-from cudf.core.index import RangeIndex, ensure_index
+from cudf.core.index import Index, RangeIndex
 from cudf.core.scalar import Scalar
 from cudf.options import get_option
 from cudf.utils.dtypes import can_convert_to_column
@@ -112,7 +112,9 @@ def factorize(values, sort=False, use_na_sentinel=True, size_hint=None):
         dtype="int64" if get_option("mode.pandas_compatible") else None,
     ).values
 
-    return labels, cats.values if return_cupy_array else ensure_index(cats)
+    return labels, cats.values if return_cupy_array else Index._from_column(
+        cats
+    )
 
 
 def _interpolation(column: ColumnBase, index: BaseIndex) -> ColumnBase:

python/cudf/cudf/core/column/categorical.py

@@ -601,11 +601,13 @@ def __setitem__(self, key, value):
             to_add_categories = 0
         else:
             if cudf.api.types.is_scalar(value):
-                arr = [value]
+                arr = column.as_column(value, length=1, nan_as_null=False)
             else:
-                arr = value
+                arr = column.as_column(value, nan_as_null=False)
             to_add_categories = len(
-                cudf.Index(arr, nan_as_null=False).difference(self.categories)
+                cudf.Index._from_column(arr).difference(
+                    cudf.Index._from_column(self.categories)
+                )
             )
 
         if to_add_categories > 0:

python/cudf/cudf/core/column/datetime.py

@@ -250,6 +250,10 @@ def __contains__(self, item: ScalarLike) -> bool:
     def time_unit(self) -> str:
         return np.datetime_data(self.dtype)[0]
 
+    @property
+    def quarter(self) -> ColumnBase:
+        return libcudf.datetime.extract_quarter(self)
+
     @property
     def year(self) -> ColumnBase:
         return self.get_dt_field("year")
@@ -308,14 +312,18 @@ def is_quarter_start(self) -> ColumnBase:
     @property
     def is_year_end(self) -> ColumnBase:
         day_of_year = self.day_of_year
-        leap_dates = libcudf.datetime.is_leap_year(self)
+        leap_dates = self.is_leap_year
 
         leap = day_of_year == cudf.Scalar(366)
         non_leap = day_of_year == cudf.Scalar(365)
         return libcudf.copying.copy_if_else(leap, non_leap, leap_dates).fillna(
             False
         )
 
+    @property
+    def is_leap_year(self) -> ColumnBase:
+        return libcudf.datetime.is_leap_year(self)
+
     @property
     def is_year_start(self) -> ColumnBase:
         return (self.day_of_year == 1).fillna(False)

python/cudf/cudf/core/column/methods.py

@@ -65,8 +65,8 @@ def _return_or_inplace(
         """
         if inplace:
             self._parent._mimic_inplace(
-                self._parent.__class__._from_data(
-                    {self._parent.name: new_col}
+                type(self._parent)._from_column(
+                    new_col, name=self._parent.name
                 ),
                 inplace=True,
             )
@@ -92,6 +92,6 @@ def _return_or_inplace(
                     index=self._parent.index if retain_index else None,
                 )
             elif isinstance(self._parent, cudf.BaseIndex):
-                return cudf.Index(new_col, name=self._parent.name)
+                return cudf.Index._from_column(new_col, name=self._parent.name)
             else:
                 return self._parent._mimic_inplace(new_col, inplace=False)

python/cudf/cudf/core/column/string.py

@@ -4693,7 +4693,7 @@ def character_tokenize(self) -> SeriesOrIndex:
                 result_col, name=self._parent.name, index=index
             )
         elif isinstance(self._parent, cudf.BaseIndex):
-            return cudf.Index(result_col, name=self._parent.name)
+            return cudf.Index._from_column(result_col, name=self._parent.name)
         else:
             return result_col
 

python/cudf/cudf/core/cut.py

@@ -292,7 +292,7 @@ def cut(
     )
 
     # we return a categorical index, as we don't have a Categorical method
-    categorical_index = cudf.CategoricalIndex._from_data({None: col})
+    categorical_index = cudf.CategoricalIndex._from_column(col)
 
     if isinstance(orig_x, (pd.Series, cudf.Series)):
         # if we have a series input we return a series output

python/cudf/cudf/core/dataframe.py

@@ -326,7 +326,7 @@ def _getitem_tuple_arg(self, arg):
                                 range(len(tmp_arg[0]))
                             )
                         },
-                        index=cudf.Index(tmp_arg[0]),
+                        index=cudf.Index._from_column(tmp_arg[0]),
                     )
                     columns_df[cantor_name] = column.as_column(
                         range(len(columns_df))
@@ -1758,7 +1758,7 @@ def _concat(
         for cols in columns:
             table_index = None
             if 1 == first_data_column_position:
-                table_index = cudf.Index(cols[0])
+                table_index = cudf.Index._from_column(cols[0])
             elif first_data_column_position > 1:
                 table_index = cudf.MultiIndex._from_data(
                     data=dict(
@@ -1810,7 +1810,7 @@ def _concat(
             if not isinstance(out.index, MultiIndex) and isinstance(
                 out.index.dtype, cudf.CategoricalDtype
             ):
-                out = out.set_index(cudf.Index(out.index._values))
+                out = out.set_index(out.index)
         for name, col in out._data.items():
             out._data[name] = col._with_type_metadata(
                 tables[0]._data[name].dtype
@@ -3007,7 +3007,7 @@ def set_index(
             and not isinstance(keys[0], (cudf.MultiIndex, pd.MultiIndex))
         ):
             # Don't turn single level MultiIndex into an Index
-            idx = cudf.Index(data_to_add[0], name=names[0])
+            idx = cudf.Index._from_column(data_to_add[0], name=names[0])
         else:
             idx = MultiIndex._from_data(dict(enumerate(data_to_add)))
             idx.names = names

python/cudf/cudf/core/dtypes.py

@@ -182,7 +182,7 @@ def __init__(self, categories=None, ordered: bool = False) -> None:
         self._ordered = ordered
 
     @property
-    def categories(self) -> "cudf.core.index.Index":
+    def categories(self) -> cudf.Index:
         """
         An ``Index`` containing the unique categories allowed.
 
@@ -194,10 +194,12 @@ def categories(self) -> "cudf.core.index.Index":
         Index(['b', 'a'], dtype='object')
         """
         if self._categories is None:
-            return cudf.Index(
-                cudf.core.column.column_empty(0, dtype="object", masked=False)
+            col = cudf.core.column.column_empty(
+                0, dtype="object", masked=False
             )
-        return cudf.Index(self._categories, copy=False)
+        else:
+            col = self._categories
+        return cudf.Index._from_column(col)
 
     @property
     def type(self):
@@ -259,7 +261,9 @@ def to_pandas(self) -> pd.CategoricalDtype:
             categories = self._categories.to_pandas()
         return pd.CategoricalDtype(categories=categories, ordered=self.ordered)
 
-    def _init_categories(self, categories: Any):
+    def _init_categories(
+        self, categories: Any
+    ) -> cudf.core.column.ColumnBase | None:
         if categories is None:
             return categories
         if len(categories) == 0 and not isinstance(

python/cudf/cudf/core/groupby/groupby.py

@@ -403,8 +403,7 @@ def indices(self) -> dict[ScalarLike, cp.ndarray]:
         if len(group_keys) > 1:
             index = cudf.MultiIndex.from_arrays(group_keys)
         else:
-            (group_keys,) = group_keys
-            index = cudf.Index(group_keys)
+            index = cudf.Index._from_column(group_keys[0])
         return dict(
             zip(index.to_pandas(), cp.split(indices.values, offsets[1:-1]))
         )
@@ -2583,7 +2582,7 @@ def _mimic_pandas_order(
             # corresponding output rows in pandas, to do that here
             # expand the result by reindexing.
             ri = cudf.RangeIndex(0, len(self.obj))
-            result.index = cudf.Index(ordering)
+            result.index = cudf.Index._from_column(ordering)
             # This reorders and expands
             result = result.reindex(ri)
         else:
@@ -3154,7 +3153,9 @@ def keys(self):
                 dict(zip(range(nkeys), self._key_columns))
             )._set_names(self.names)
         else:
-            return cudf.Index(self._key_columns[0], name=self.names[0])
+            return cudf.Index._from_column(
+                self._key_columns[0], name=self.names[0]
+            )
 
     @property
     def values(self) -> cudf.core.frame.Frame: