Remove cudf._lib.utils in favor of python APIs (#17625)

Contributes to #17317

Dependent on #17582

Did a search across RAPIDS and Morpheus and didn't find usage of these methods.

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

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

URL: #17625

python/cudf/cudf/_lib/CMakeLists.txt

@@ -12,7 +12,7 @@
 # the License.
 # =============================================================================
 
-set(cython_sources column.pyx scalar.pyx strings_udf.pyx types.pyx utils.pyx)
+set(cython_sources column.pyx scalar.pyx strings_udf.pyx types.pyx)
 set(linked_libraries cudf::cudf)
 
 rapids_cython_create_modules(

python/cudf/cudf/core/frame.py

@@ -861,7 +861,9 @@ def _quantile_table(
                 column_order,
                 null_precedence,
             )
-            columns = libcudf.utils.columns_from_pylibcudf_table(plc_table)
+            columns = [
+                ColumnBase.from_pylibcudf(col) for col in plc_table.columns()
+            ]
         return self._from_columns_like_self(
             columns,
             column_names=self._column_names,

python/cudf/cudf/io/avro.py

@@ -3,7 +3,7 @@
 import pylibcudf as plc
 
 import cudf
-from cudf._lib.utils import data_from_pylibcudf_io
+from cudf._lib.column import Column
 from cudf.utils import ioutils
 
 
@@ -46,5 +46,12 @@ def read_avro(
         options.set_columns(columns)
 
     plc_result = plc.io.avro.read_avro(options)
-
-    return cudf.DataFrame._from_data(*data_from_pylibcudf_io(plc_result))
+    data = {
+        name: Column.from_pylibcudf(col)
+        for name, col in zip(
+            plc_result.column_names(include_children=False),
+            plc_result.columns,
+            strict=True,
+        )
+    }
+    return cudf.DataFrame._from_data(data)

python/cudf/cudf/io/csv.py

@@ -15,8 +15,8 @@
 import pylibcudf as plc
 
 import cudf
+from cudf._lib.column import Column
 from cudf._lib.types import dtype_to_pylibcudf_type
-from cudf._lib.utils import data_from_pylibcudf_io
 from cudf.api.types import is_hashable, is_scalar
 from cudf.core.buffer import acquire_spill_lock
 from cudf.utils import ioutils
@@ -251,9 +251,17 @@ def read_csv(
     if na_values is not None:
         options.set_na_values([str(val) for val in na_values])
 
-    df = cudf.DataFrame._from_data(
-        *data_from_pylibcudf_io(plc.io.csv.read_csv(options))
-    )
+    table_w_meta = plc.io.csv.read_csv(options)
+    data = {
+        name: Column.from_pylibcudf(col)
+        for name, col in zip(
+            table_w_meta.column_names(include_children=False),
+            table_w_meta.columns,
+            strict=True,
+        )
+    }
+
+    df = cudf.DataFrame._from_data(data)
 
     if isinstance(dtype, abc.Mapping):
         for k, v in dtype.items():

python/cudf/cudf/io/json.py

@@ -15,7 +15,6 @@
 import cudf
 from cudf._lib.column import Column
 from cudf._lib.types import dtype_to_pylibcudf_type
-from cudf._lib.utils import _data_from_columns, data_from_pylibcudf_io
 from cudf.core.buffer import acquire_spill_lock
 from cudf.utils import ioutils
 from cudf.utils.dtypes import _maybe_convert_to_default_type
@@ -178,13 +177,11 @@ def read_json(
                     )
                 )
             )
-            df = cudf.DataFrame._from_data(
-                *_data_from_columns(
-                    columns=[Column.from_pylibcudf(col) for col in res_cols],
-                    column_names=res_col_names,
-                    index_names=None,
-                )
-            )
+            data = {
+                name: Column.from_pylibcudf(col)
+                for name, col in zip(res_col_names, res_cols, strict=True)
+            }
+            df = cudf.DataFrame._from_data(data)
             ioutils._add_df_col_struct_names(df, res_child_names)
             return df
         else:
@@ -207,10 +204,15 @@ def read_json(
                     extra_parameters=kwargs,
                 )
             )
-
-            df = cudf.DataFrame._from_data(
-                *data_from_pylibcudf_io(table_w_meta)
-            )
+            data = {
+                name: Column.from_pylibcudf(col)
+                for name, col in zip(
+                    table_w_meta.column_names(include_children=False),
+                    table_w_meta.columns,
+                    strict=True,
+                )
+            }
+            df = cudf.DataFrame._from_data(data)
 
             # Post-processing to add in struct column names
             ioutils._add_df_col_struct_names(df, table_w_meta.child_names)

python/cudf/cudf/io/orc.py

@@ -10,10 +10,11 @@
 import pylibcudf as plc
 
 import cudf
+from cudf._lib.column import Column
 from cudf._lib.types import dtype_to_pylibcudf_type
-from cudf._lib.utils import data_from_pylibcudf_io
 from cudf.api.types import is_list_like
 from cudf.core.buffer import acquire_spill_lock
+from cudf.core.index import _index_from_data
 from cudf.utils import ioutils
 
 try:
@@ -323,11 +324,35 @@ def read_orc(
                     actual_index_names = list(index_col_names.values())
                     col_names = names[len(actual_index_names) :]
 
-            data, index = data_from_pylibcudf_io(
-                tbl_w_meta,
-                col_names if columns is None else names,
-                actual_index_names,
-            )
+            result_col_names = col_names if columns is None else names
+            if actual_index_names is None:
+                index = None
+                data = {
+                    name: Column.from_pylibcudf(col)
+                    for name, col in zip(
+                        result_col_names, tbl_w_meta.columns, strict=True
+                    )
+                }
+            else:
+                result_columns = [
+                    Column.from_pylibcudf(col) for col in tbl_w_meta.columns
+                ]
+                index = _index_from_data(
+                    dict(
+                        zip(
+                            actual_index_names,
+                            result_columns[: len(actual_index_names)],
+                            strict=True,
+                        )
+                    )
+                )
+                data = dict(
+                    zip(
+                        result_col_names,
+                        result_columns[len(actual_index_names) :],
+                        strict=True,
+                    )
+                )
 
             if is_range_index:
                 index = range_idx

python/cudf/cudf/io/parquet.py

@@ -23,10 +23,6 @@
 
 import cudf
 from cudf._lib.column import Column
-from cudf._lib.utils import (
-    _data_from_columns,
-    data_from_pylibcudf_io,
-)
 from cudf.api.types import is_list_like
 from cudf.core.buffer import acquire_spill_lock
 from cudf.core.column import as_column, column_empty
@@ -1238,16 +1234,11 @@ def _read_parquet(
                     # Drop residual columns to save memory
                     tbl._columns[i] = None
 
-            df = cudf.DataFrame._from_data(
-                *_data_from_columns(
-                    columns=[
-                        Column.from_pylibcudf(plc)
-                        for plc in concatenated_columns
-                    ],
-                    column_names=column_names,
-                    index_names=None,
-                )
-            )
+            data = {
+                name: Column.from_pylibcudf(col)
+                for name, col in zip(column_names, concatenated_columns)
+            }
+            df = cudf.DataFrame._from_data(data)
             df = _process_metadata(
                 df,
                 column_names,
@@ -1287,8 +1278,16 @@ def _read_parquet(
                 options.set_filter(filters)
 
             tbl_w_meta = plc.io.parquet.read_parquet(options)
+            data = {
+                name: Column.from_pylibcudf(col)
+                for name, col in zip(
+                    tbl_w_meta.column_names(include_children=False),
+                    tbl_w_meta.columns,
+                    strict=True,
+                )
+            }
 
-            df = cudf.DataFrame._from_data(*data_from_pylibcudf_io(tbl_w_meta))
+            df = cudf.DataFrame._from_data(data)
 
             df = _process_metadata(
                 df,