Use as_column instead of arange for range like inputs (#14689)

1. Allows range-like inputs in `as_column` to short circuit and not materialize when creating columns
2. Avoids diverging column construction logic between `column.arange` and `column.as_column`

Authors:
  - Matthew Roeschke (https://github.com/mroeschke)

Approvers:
  - Vyas Ramasubramani (https://github.com/vyasr)

URL: #14689

python/cudf/cudf/core/column/__init__.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2023, NVIDIA CORPORATION.
+# Copyright (c) 2020-2024, NVIDIA CORPORATION.
 
 """
 isort: skip_file
@@ -8,7 +8,6 @@
 from cudf.core.column.categorical import CategoricalColumn
 from cudf.core.column.column import (
     ColumnBase,
-    arange,
     as_column,
     build_categorical_column,
     build_column,

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

@@ -1159,7 +1159,7 @@ def find_and_replace(
         new_cats_col = new_cats_col.apply_boolean_mask(bmask)
         new_cats = cudf.DataFrame._from_data(
             {
-                "index": cudf.core.column.arange(len(new_cats_col)),
+                "index": column.as_column(range(len(new_cats_col))),
                 "cats": new_cats_col,
             }
         )
@@ -1531,9 +1531,13 @@ def _set_categories(
         )
         out_code_dtype = min_unsigned_type(max_cat_size)
 
-        cur_order = column.arange(len(cur_codes))
-        old_codes = column.arange(len(cur_cats), dtype=out_code_dtype)
-        new_codes = column.arange(len(new_cats), dtype=out_code_dtype)
+        cur_order = column.as_column(range(len(cur_codes)))
+        old_codes = column.as_column(
+            range(len(cur_cats)), dtype=out_code_dtype
+        )
+        new_codes = column.as_column(
+            range(len(new_cats)), dtype=out_code_dtype
+        )
 
         new_df = cudf.DataFrame._from_data(
             data={"new_codes": new_codes, "cats": new_cats}

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

@@ -554,10 +554,8 @@ def slice(
             ]._with_type_metadata(self.dtype)
         else:
             # Need to create a gather map for given slice with stride
-            gather_map = arange(
-                start=start,
-                stop=stop,
-                step=stride,
+            gather_map = as_column(
+                range(start, stop, stride),
                 dtype=cudf.dtype(np.int32),
             )
             return self.take(gather_map)
@@ -626,10 +624,8 @@ def _scatter_by_slice(
                 )
 
         # step != 1, create a scatter map with arange
-        scatter_map = arange(
-            start=start,
-            stop=stop,
-            step=step,
+        scatter_map = as_column(
+            range(start, stop, step),
             dtype=cudf.dtype(np.int32),
         )
 
@@ -745,7 +741,7 @@ def indices_of(
             assert len(value) == 1
         mask = libcudf.search.contains(value, self)
         return apply_boolean_mask(
-            [arange(0, len(self), dtype=size_type_dtype)], mask
+            [as_column(range(0, len(self)), dtype=size_type_dtype)], mask
         )[0]
 
     def _find_first_and_last(self, value: ScalarLike) -> Tuple[int, int]:
@@ -1379,7 +1375,9 @@ def _return_sentinel_column():
             [self], [cats], how="left"
         )
         codes = libcudf.copying.gather(
-            [arange(len(cats), dtype=dtype)], right_gather_map, nullify=True
+            [as_column(range(len(cats)), dtype=dtype)],
+            right_gather_map,
+            nullify=True,
         )
         del right_gather_map
         # reorder `codes` so that its values correspond to the
@@ -1905,13 +1903,26 @@ def as_column(
     * Objects exposing ``__array_interface__``(e.g., numpy arrays)
     * pyarrow array
     * pandas.Categorical objects
+    * range objects
     """
-    if isinstance(arbitrary, ColumnBase):
+    if isinstance(arbitrary, (range, pd.RangeIndex, cudf.RangeIndex)):
+        column = libcudf.filling.sequence(
+            len(arbitrary),
+            as_device_scalar(arbitrary.start, dtype=cudf.dtype("int64")),
+            as_device_scalar(arbitrary.step, dtype=cudf.dtype("int64")),
+        )
+        if cudf.get_option("default_integer_bitwidth") and dtype is None:
+            dtype = cudf.dtype(
+                f'i{cudf.get_option("default_integer_bitwidth")//8}'
+            )
+        if dtype is not None:
+            column = column.astype(dtype)
+        return column
+    elif isinstance(arbitrary, ColumnBase):
         if dtype is not None:
             return arbitrary.astype(dtype)
         else:
             return arbitrary
-
     elif isinstance(arbitrary, cudf.Series):
         data = arbitrary._column
         if dtype is not None:
@@ -2614,70 +2625,6 @@ def deserialize_columns(headers: List[dict], frames: List) -> List[ColumnBase]:
     return columns
 
 
-def arange(
-    start: Union[int, float],
-    stop: Optional[Union[int, float]] = None,
-    step: Union[int, float] = 1,
-    dtype=None,
-) -> cudf.core.column.NumericalColumn:
-    """
-    Returns a column with evenly spaced values within a given interval.
-
-    Values are generated within the half-open interval [start, stop).
-    The first three arguments are mapped like the range built-in function,
-    i.e. start and step are optional.
-
-    Parameters
-    ----------
-    start : int/float
-        Start of the interval.
-    stop : int/float, default is None
-        Stop of the interval.
-    step : int/float, default 1
-        Step width between each pair of consecutive values.
-    dtype : default None
-        Data type specifier. It is inferred from other arguments by default.
-
-    Returns
-    -------
-    cudf.core.column.NumericalColumn
-
-    Examples
-    --------
-    >>> import cudf
-    >>> col = cudf.core.column.arange(2, 7, 1, dtype='int16')
-    >>> col
-    <cudf.core.column.numerical.NumericalColumn object at 0x7ff7998f8b90>
-    >>> cudf.Series(col)
-    0    2
-    1    3
-    2    4
-    3    5
-    4    6
-    dtype: int16
-    """
-    if stop is None:
-        stop = start
-        start = 0
-
-    if step is None:
-        step = 1
-
-    size = len(range(int(start), int(stop), int(step)))
-    if size == 0:
-        if dtype is None:
-            dtype = cudf.dtype("int64")
-        return cast(
-            cudf.core.column.NumericalColumn, column_empty(0, dtype=dtype)
-        )
-
-    return libcudf.filling.sequence(
-        size,
-        as_device_scalar(start, dtype=dtype),
-        as_device_scalar(step, dtype=dtype),
-    )
-
-
 def full(
     size: int, fill_value: ScalarLike, dtype: Optional[Dtype] = None
 ) -> ColumnBase:

python/cudf/cudf/core/dataframe.py

@@ -342,10 +342,16 @@ def _getitem_tuple_arg(self, arg):
                         tmp_col_name = (tmp_col_name, *extra)
                         cantor_name = (cantor_name, *extra)
                     other_df = DataFrame(
-                        {tmp_col_name: column.arange(len(tmp_arg[0]))},
+                        {
+                            tmp_col_name: column.as_column(
+                                range(len(tmp_arg[0]))
+                            )
+                        },
                         index=as_index(tmp_arg[0]),
                     )
-                    columns_df[cantor_name] = column.arange(len(columns_df))
+                    columns_df[cantor_name] = column.as_column(
+                        range(len(columns_df))
+                    )
                     df = other_df.join(columns_df, how="inner")
                     # as join is not assigning any names to index,
                     # update it over here

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

@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2023, NVIDIA CORPORATION.
+# Copyright (c) 2020-2024, NVIDIA CORPORATION.
 
 import copy
 import itertools
@@ -23,7 +23,7 @@
 from cudf._typing import AggType, DataFrameOrSeries, MultiColumnAggType
 from cudf.api.types import is_bool_dtype, is_float_dtype, is_list_like
 from cudf.core.abc import Serializable
-from cudf.core.column.column import ColumnBase, arange, as_column
+from cudf.core.column.column import ColumnBase, as_column
 from cudf.core.column_accessor import ColumnAccessor
 from cudf.core.join._join_helpers import _match_join_keys
 from cudf.core.mixins import Reducible, Scannable
@@ -761,7 +761,7 @@ 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, len(self.obj))]
+                [as_column(range(0, len(self.obj)))]
             )
             # Invert permutation from original order to groups on the
             # subset of entries we want.
@@ -2543,7 +2543,9 @@ def _mimic_pandas_order(
         # 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, len(self.obj))])
+        _, (ordering,), _ = self._groupby.groups(
+            [as_column(range(0, len(self.obj)))]
+        )
         if self._dropna and any(
             c.has_nulls(include_nan=True) > 0
             for c in self.grouping._key_columns

python/cudf/cudf/core/index.py

@@ -286,9 +286,7 @@ def _num_rows(self):
     @_cudf_nvtx_annotate
     def _values(self):
         if len(self) > 0:
-            return column.arange(
-                self._start, self._stop, self._step, dtype=self.dtype
-            )
+            return column.as_column(self._range, dtype=self.dtype)
         else:
             return column.column_empty(0, masked=False, dtype=self.dtype)
 

python/cudf/cudf/core/indexed_frame.py

@@ -182,12 +182,8 @@ def _indices_from_labels(obj, labels):
     # join is not guaranteed to maintain the index ordering
     # so we will sort it with its initial ordering which is stored
     # in column "__"
-    lhs = cudf.DataFrame(
-        {"__": cudf.core.column.arange(len(labels))}, index=labels
-    )
-    rhs = cudf.DataFrame(
-        {"_": cudf.core.column.arange(len(obj))}, index=obj.index
-    )
+    lhs = cudf.DataFrame({"__": as_column(range(len(labels)))}, index=labels)
+    rhs = cudf.DataFrame({"_": as_column(range(len(obj)))}, index=obj.index)
     return lhs.join(rhs).sort_values(by=["__", "_"])["_"]
 
 
@@ -1897,10 +1893,8 @@ def _slice(self, arg: slice, keep_index: bool = True) -> Self:
         if stride != 1:
             return self._gather(
                 GatherMap.from_column_unchecked(
-                    cudf.core.column.arange(
-                        start,
-                        stop=stop,
-                        step=stride,
+                    as_column(
+                        range(start, stop, stride),
                         dtype=libcudf.types.size_type_dtype,
                     ),
                     len(self),
@@ -2541,9 +2535,9 @@ def _align_to_index(
         # to recover ordering after index alignment.
         sort_col_id = str(uuid4())
         if how == "left":
-            lhs[sort_col_id] = cudf.core.column.arange(len(lhs))
+            lhs[sort_col_id] = as_column(range(len(lhs)))
         elif how == "right":
-            rhs[sort_col_id] = cudf.core.column.arange(len(rhs))
+            rhs[sort_col_id] = as_column(range(len(rhs)))
 
         result = lhs.join(rhs, how=how, sort=sort)
         if how in ("left", "right"):

python/cudf/cudf/core/join/join.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2023, NVIDIA CORPORATION.
+# Copyright (c) 2020-2024, NVIDIA CORPORATION.
 from __future__ import annotations
 
 import itertools
@@ -232,7 +232,11 @@ def _gather_maps(self, left_cols, right_cols):
         key_order = list(
             itertools.chain.from_iterable(
                 libcudf.copying.gather(
-                    [cudf.core.column.arange(n, dtype=size_type_dtype)],
+                    [
+                        cudf.core.column.as_column(
+                            range(n), dtype=size_type_dtype
+                        )
+                    ],
                     map_,
                     nullify=null,
                 )

python/cudf/cudf/core/multiindex.py

@@ -501,9 +501,9 @@ def __repr__(self):
             # TODO: Update the following two arange calls to
             # a single arange call once arange has support for
             # a vector start/end points.
-            indices = column.arange(start=0, stop=n, step=1)
+            indices = column.as_column(range(n))
             indices = indices.append(
-                column.arange(start=len(self) - n, stop=len(self), step=1)
+                column.as_column(range(len(self) - n, len(self), 1))
             )
             preprocess = self.take(indices)
         else:
@@ -795,7 +795,7 @@ def _compute_validity_mask(self, index, row_tuple, max_length):
             [
                 frame,
                 cudf.DataFrame(
-                    {"idx": cudf.Series(column.arange(len(frame)))}
+                    {"idx": cudf.Series(column.as_column(range(len(frame))))}
                 ),
             ],
             axis=1,
@@ -807,7 +807,7 @@ def _compute_validity_mask(self, index, row_tuple, max_length):
         # obtain deterministic ordering.
         if cudf.get_option("mode.pandas_compatible"):
             lookup_order = "_" + "_".join(map(str, lookup._data.names))
-            lookup[lookup_order] = column.arange(len(lookup))
+            lookup[lookup_order] = column.as_column(range(len(lookup)))
             postprocess = operator.methodcaller(
                 "sort_values", by=[lookup_order, "idx"]
             )
@@ -840,14 +840,16 @@ def _get_valid_indices_by_tuple(self, index, row_tuple, max_length):
             ):
                 stop = row_tuple.stop or max_length
                 start, stop, step = row_tuple.indices(stop)
-                return column.arange(start, stop, step)
+                return column.as_column(range(start, stop, step))
             start_values = self._compute_validity_mask(
                 index, row_tuple.start, max_length
             )
             stop_values = self._compute_validity_mask(
                 index, row_tuple.stop, max_length
             )
-            return column.arange(start_values.min(), stop_values.max() + 1)
+            return column.as_column(
+                range(start_values.min(), stop_values.max() + 1)
+            )
         elif isinstance(row_tuple, numbers.Number):
             return row_tuple
         return self._compute_validity_mask(index, row_tuple, max_length)
@@ -1024,7 +1026,7 @@ def __getitem__(self, index):
             index = np.array(index)
         elif isinstance(index, slice):
             start, stop, step = index.indices(len(self))
-            index = column.arange(start, stop, step)
+            index = column.as_column(range(start, stop, step))
         result = MultiIndex.from_frame(
             self.to_frame(index=False, name=range(0, self.nlevels)).take(
                 index

python/cudf/cudf/core/series.py

@@ -55,7 +55,6 @@
     DatetimeColumn,
     IntervalColumn,
     TimeDeltaColumn,
-    arange,
     as_column,
     full,
 )
@@ -1366,7 +1365,9 @@ def map(self, arg, na_action=None) -> "Series":
                 raise NotImplementedError(
                     "default values in dicts are currently not supported."
                 )
-            lhs = cudf.DataFrame({"x": self, "orig_order": arange(len(self))})
+            lhs = cudf.DataFrame(
+                {"x": self, "orig_order": as_column(range(len(self)))}
+            )
             rhs = cudf.DataFrame(
                 {
                     "x": arg.keys(),
@@ -1386,7 +1387,9 @@ def map(self, arg, na_action=None) -> "Series":
                     "Reindexing only valid with"
                     " uniquely valued Index objects"
                 )
-            lhs = cudf.DataFrame({"x": self, "orig_order": arange(len(self))})
+            lhs = cudf.DataFrame(
+                {"x": self, "orig_order": as_column(range(len(self)))}
+            )
             rhs = cudf.DataFrame(
                 {
                     "x": arg.keys(),