handle runtime conversion failure

rapidsai · Oct 16, 2024 · 6d699ac · 6d699ac
1 parent 7feb1f3
commit 6d699ac
Show file tree

Hide file tree

Showing 2 changed files with 40 additions and 8 deletions.
diff --git a/python/cudf_polars/cudf_polars/dsl/expressions/unary.py b/python/cudf_polars/cudf_polars/dsl/expressions/unary.py
@@ -9,8 +9,16 @@
 
 import pyarrow as pa
 import pylibcudf as plc
+from pylibcudf.strings.convert.convert_floats import from_floats, is_float, to_floats
+from pylibcudf.strings.convert.convert_integers import (
+    from_integers,
+    is_integer,
+    to_integers,
+)
 from pylibcudf.traits import is_floating_point
 
+from polars.exceptions import InvalidOperationError
+
 from cudf_polars.containers import Column
 from cudf_polars.dsl.expressions.base import AggInfo, ExecutionContext, Expr
 from cudf_polars.dsl.expressions.literal import Literal
@@ -61,18 +69,32 @@ def do_evaluate(
     def _handle_string_cast(self, column: Column) -> plc.Column:
         if self.dtype.id() == plc.TypeId.STRING:
             if is_floating_point(column.obj.type()):
-                result = plc.strings.convert.convert_floats.from_floats(column.obj)
+                result = from_floats(column.obj)
             else:
-                result = plc.strings.convert.convert_integers.from_integers(column.obj)
+                result = from_integers(column.obj)
         else:
             if is_floating_point(self.dtype):
-                result = plc.strings.convert.convert_floats.to_floats(
-                    column.obj, self.dtype
-                )
+                floats = is_float(column.obj)
+                if not plc.interop.to_arrow(
+                    plc.reduce.reduce(
+                        floats,
+                        plc.aggregation.all(),
+                        plc.DataType(plc.TypeId.BOOL8),
+                    )
+                ).as_py():
+                    raise InvalidOperationError("Conversion from `str` failed.")
+                result = to_floats(column.obj, self.dtype)
             else:
-                result = plc.strings.convert.convert_integers.to_integers(
-                    column.obj, self.dtype
-                )
+                integers = is_integer(column.obj)
+                if not plc.interop.to_arrow(
+                    plc.reduce.reduce(
+                        integers,
+                        plc.aggregation.all(),
+                        plc.DataType(plc.TypeId.BOOL8),
+                    )
+                ).as_py():
+                    raise InvalidOperationError("Conversion from `str` failed.")
+                result = to_integers(column.obj, self.dtype)
         return result
 
     def collect_agg(self, *, depth: int) -> AggInfo:

diff --git a/python/cudf_polars/tests/expressions/test_stringfunction.py b/python/cudf_polars/tests/expressions/test_stringfunction.py
@@ -375,3 +375,13 @@ def test_string_to_numeric(str_to_numeric_data, numeric_type):
 def test_string_from_numeric(str_from_numeric_data):
     query = str_from_numeric_data.select(pl.col("a").cast(pl.String))
     assert_gpu_result_equal(query)
+
+
+def test_string_to_numeric_invalid(numeric_type):
+    df = pl.LazyFrame({"a": ["a", "b", "c"]})
+    q = df.select(pl.col("a").cast(numeric_type))
+    assert_collect_raises(
+        q,
+        polars_except=pl.exceptions.InvalidOperationError,
+        cudf_except=pl.exceptions.ComputeError,
+    )