GH-48457: [Python] Overflow check in datetime conversion in pa.array

python/pyarrow/src/arrow/python/numpy_to_arrow.cc

@@ -364,15 +364,33 @@ Status CastBuffer(const std::shared_ptr<DataType>& in_type,
   return Status::OK();
 }
 
+// Downcast buffer from FromType to ToType with optional overflow checking.
+// This function only supports narrowing casts (FromType wider than ToType).
+// Do not use this function for widening casts (ToType wider than FromType).
 template <typename FromType, typename ToType>
-Status StaticCastBuffer(const Buffer& input, const int64_t length, MemoryPool* pool,
-                        std::shared_ptr<Buffer>* out) {
+Status StaticDowncastBuffer(const Buffer& input, int64_t length, MemoryPool* pool,
+                            const uint8_t* null_bitmap,
+                            const compute::CastOptions& cast_options,
+                            std::shared_ptr<Buffer>* out) {
   ARROW_ASSIGN_OR_RAISE(auto result, AllocateBuffer(sizeof(ToType) * length, pool));
 
   auto in_values = reinterpret_cast<const FromType*>(input.data());
   auto out_values = reinterpret_cast<ToType*>(result->mutable_data());
+
+  constexpr FromType kMin = std::numeric_limits<ToType>::min();
+  constexpr FromType kMax = std::numeric_limits<ToType>::max();
+
   for (int64_t i = 0; i < length; ++i) {
-    *out_values++ = static_cast<ToType>(*in_values++);
+    FromType value = *in_values++;
+    // Check overflow only when cast_options.allow_int_overflow is false and value is not
+    // null
+    bool check_overflow = !cast_options.allow_int_overflow &&
+                          ((null_bitmap == nullptr) || bit_util::GetBit(null_bitmap, i));
+    if (check_overflow && (value < kMin || value > kMax)) {
+      return Status::Invalid("Integer value ", value, " out of bounds for int",
+                             sizeof(ToType) * 8, " conversion at index ", i);
+    }
+    *out_values++ = static_cast<ToType>(value);
   }
   *out = std::move(result);
   return Status::OK();
@@ -496,10 +514,10 @@ inline Status NumPyConverter::ConvertData<Date32Type>(std::shared_ptr<Buffer>* d
     // separately here from int64_t to int32_t, because this data is not
     // supported in compute::Cast
     if (date_dtype->meta.base == NPY_FR_D) {
-      // TODO(wesm): How pedantic do we really want to be about checking for int32
-      // overflow here?
-      Status s = StaticCastBuffer<int64_t, int32_t>(**data, length_, pool_, data);
-      RETURN_NOT_OK(s);
+      // Downcast from int64 to int32 with overflow checking
+      const uint8_t* null_bitmap_ptr = null_bitmap_ ? null_bitmap_->data() : nullptr;
+      RETURN_NOT_OK((StaticDowncastBuffer<int64_t, int32_t>(
+          **data, length_, pool_, null_bitmap_ptr, cast_options_, data)));
     } else {
       ARROW_ASSIGN_OR_RAISE(input_type, NumPyDtypeToArrow(dtype_));
       if (!input_type->Equals(*type_)) {

python/pyarrow/tests/test_array.py

@@ -2445,6 +2445,32 @@ def test_array_roundtrip_from_numpy_datetimeD():
     assert result.dtype == arr.dtype
 
 
+@pytest.mark.numpy
+def test_array_from_numpy_datetime_overflow():
+    # datetime64[D] to date32 conversion should check for int32 overflow
+    # when safe=True (default)
+    overflow_value = np.int64(3000000000)
+    arr = np.array([overflow_value], dtype='datetime64[D]')
+    with pytest.raises(pa.ArrowInvalid, match='value .* out of bounds'):
+        pa.array(arr, type=pa.date32())
+
+    underflow_value = np.int64(-3000000000)
+    arr = np.array([underflow_value], dtype='datetime64[D]')
+    with pytest.raises(pa.ArrowInvalid, match='value .* out of bounds'):
+        pa.array(arr, type=pa.date32())
+
+    # safe=False should allow overflow
+    result = pa.array(np.array([overflow_value], dtype='datetime64[D]'),
+                      type=pa.date32(), safe=False)
+    assert len(result) == 1
+
+    # Values within int32 range should work
+    valid_arr = np.array([0, 100, -100, 2147483647, -2147483648],
+                         dtype='datetime64[D]')
+    result = pa.array(valid_arr, type=pa.date32())
+    assert len(result) == 5
+
+
 def test_array_from_naive_datetimes():
     arr = pa.array([
         None,