From daa95e4c7a05a0067336f2967c66e11379042cb5 Mon Sep 17 00:00:00 2001
From: Matthew Roeschke <10647082+mroeschke@users.noreply.github.com>
Date: Fri, 29 Dec 2023 11:40:40 -0800
Subject: [PATCH 1/3] Fix constructing Series/Index from arrow array and dtype

---
 python/cudf/cudf/core/column/column.py | 19 ++++++++++++-------
 python/cudf/cudf/tests/test_series.py  |  9 +++++++++
 2 files changed, 21 insertions(+), 7 deletions(-)

diff --git a/python/cudf/cudf/core/column/column.py b/python/cudf/cudf/core/column/column.py
index 296fd6a41b0..81bc0b08d01 100644
--- a/python/cudf/cudf/core/column/column.py
+++ b/python/cudf/cudf/core/column/column.py
@@ -2036,6 +2036,8 @@ def as_column(
                 new_dtype = "str"
 
             col = col.astype(new_dtype)
+        elif dtype is not None:
+            col = col.astype(dtype)
 
         return col
 
@@ -2112,6 +2114,15 @@ def as_column(
                 arbitrary, nan_as_null=nan_as_null, dtype=dtype, length=length
             )
         elif arbitrary.dtype.kind == "O":
+            if len(arbitrary) == 0:
+                # TODO: Can remove once empty constructor default becomes
+                # object instead of float.
+                return as_column(
+                    pa.array([], type=pa.string()),
+                    nan_as_null=nan_as_null,
+                    dtype=dtype,
+                    length=length,
+                )
             if isinstance(arbitrary, pd.arrays.PandasArray):
                 # infer_dtype does not handle PandasArray
                 arbitrary = np.array(arbitrary, dtype=object)
@@ -2140,15 +2151,9 @@ def as_column(
                 arbitrary,
                 from_pandas=True,
             )
-            if isinstance(pyarrow_array.type, pa.Decimal128Type):
-                pyarrow_type = cudf.Decimal128Dtype.from_arrow(
-                    pyarrow_array.type
-                )
-            else:
-                pyarrow_type = arbitrary.dtype
             data = as_column(
                 pyarrow_array,
-                dtype=pyarrow_type,
+                dtype=dtype,
                 nan_as_null=nan_as_null,
                 length=length,
             )
diff --git a/python/cudf/cudf/tests/test_series.py b/python/cudf/cudf/tests/test_series.py
index 39da34fa89c..ffe4a32bdbf 100644
--- a/python/cudf/cudf/tests/test_series.py
+++ b/python/cudf/cudf/tests/test_series.py
@@ -2640,6 +2640,15 @@ def test_astype_pandas_nullable_pandas_compat(dtype, klass, kind):
             ser.astype(kind(dtype))
 
 
+@pytest.mark.parametrize("klass", [cudf.Series, cudf.Index])
+@pytest.mark.parametrize(
+    "data", [pa.array([1, None]), pa.chunked_array([[1, None]])]
+)
+def test_from_arrow_array_dtype(klass, data):
+    obj = klass(data, dtype="int8")
+    assert obj.dtype == np.dtype("int8")
+
+
 def test_series_where_mixed_bool_dtype():
     s = cudf.Series([True, False, True])
     with pytest.raises(TypeError):

From 0ddaf77bd7f4fc0e22778c5fb3faa5bfda748023 Mon Sep 17 00:00:00 2001
From: Matthew Roeschke <10647082+mroeschke@users.noreply.github.com>
Date: Fri, 29 Dec 2023 11:48:12 -0800
Subject: [PATCH 2/3] Add another test casting from object

---
 python/cudf/cudf/tests/test_series.py | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/python/cudf/cudf/tests/test_series.py b/python/cudf/cudf/tests/test_series.py
index ffe4a32bdbf..b8233e38f1b 100644
--- a/python/cudf/cudf/tests/test_series.py
+++ b/python/cudf/cudf/tests/test_series.py
@@ -2642,13 +2642,24 @@ def test_astype_pandas_nullable_pandas_compat(dtype, klass, kind):
 
 @pytest.mark.parametrize("klass", [cudf.Series, cudf.Index])
 @pytest.mark.parametrize(
-    "data", [pa.array([1, None]), pa.chunked_array([[1, None]])]
+    "data",
+    [
+        pa.array([1, None], type=pa.int64()),
+        pa.chunked_array([[1, None]], type=pa.int64()),
+    ],
 )
 def test_from_arrow_array_dtype(klass, data):
     obj = klass(data, dtype="int8")
     assert obj.dtype == np.dtype("int8")
 
 
+@pytest.mark.parametrize("klass", [cudf.Series, cudf.Index])
+def test_from_pandas_object_dtype_passed_dtype(klass):
+    result = klass(pd.Series([True, False], dtype=object), dtype="int8")
+    expected = klass(pa.array([1, 0], type=pa.int8()))
+    assert_eq(result, expected)
+
+
 def test_series_where_mixed_bool_dtype():
     s = cudf.Series([True, False, True])
     with pytest.raises(TypeError):

From a9fc10411258665b41065c84f39809eaf091aec9 Mon Sep 17 00:00:00 2001
From: Matthew Roeschke <10647082+mroeschke@users.noreply.github.com>
Date: Fri, 29 Dec 2023 13:08:45 -0800
Subject: [PATCH 3/3] Remove buggy dtype passing

---
 python/cudf/cudf/core/column/column.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/cudf/cudf/core/column/column.py b/python/cudf/cudf/core/column/column.py
index 81bc0b08d01..8d2c49e74c1 100644
--- a/python/cudf/cudf/core/column/column.py
+++ b/python/cudf/cudf/core/column/column.py
@@ -2285,7 +2285,7 @@ def as_column(
             if dtype is not None:
                 data = data.astype(dtype)
         elif arb_dtype.kind in ("O", "U"):
-            data = as_column(pa.array(arbitrary), dtype=arbitrary.dtype)
+            data = as_column(pa.array(arbitrary), dtype=dtype)
             # There is no cast operation available for pa.Array from int to
             # str, Hence instead of handling in pa.Array block, we
             # will have to type-cast here.