From 24eaba083a484881f5d42aeb4d2b5528c3931a04 Mon Sep 17 00:00:00 2001
From: Mithun RK <mythrocks@gmail.com>
Date: Tue, 4 May 2021 11:55:42 -0700
Subject: [PATCH] Fix scatter output size for structs.

1. Output struct size must match target column, not source.
2. Test case for #1.
3. Also, move scatter_struct_tests and scatter_list_tests
   from .cu to .cpp, for faster compile.
---
 cpp/include/cudf/detail/scatter.cuh           |  2 +-
 cpp/tests/CMakeLists.txt                      |  4 ++--
 ...r_list_tests.cu => scatter_list_tests.cpp} |  2 --
 ...ruct_tests.cu => scatter_struct_tests.cpp} | 21 +++++++++++++++++++
 4 files changed, 24 insertions(+), 5 deletions(-)
 rename cpp/tests/copying/{scatter_list_tests.cu => scatter_list_tests.cpp} (99%)
 rename cpp/tests/copying/{scatter_struct_tests.cu => scatter_struct_tests.cpp} (93%)

diff --git a/cpp/include/cudf/detail/scatter.cuh b/cpp/include/cudf/detail/scatter.cuh
index d069ed06cae..68f4adfadc9 100644
--- a/cpp/include/cudf/detail/scatter.cuh
+++ b/cpp/include/cudf/detail/scatter.cuh
@@ -317,7 +317,7 @@ struct column_scatterer_impl<struct_view> {
 
     // Need to put the result column in a vector to call `gather_bitmask`
     std::vector<std::unique_ptr<column>> result;
-    result.emplace_back(cudf::make_structs_column(source.size(),
+    result.emplace_back(cudf::make_structs_column(target.size(),
                                                   std::move(output_struct_members),
                                                   0,
                                                   rmm::device_buffer{0, stream, mr},
diff --git a/cpp/tests/CMakeLists.txt b/cpp/tests/CMakeLists.txt
index 6dd50592274..29dd4319bfc 100644
--- a/cpp/tests/CMakeLists.txt
+++ b/cpp/tests/CMakeLists.txt
@@ -206,8 +206,8 @@ ConfigureTest(COPYING_TEST
     copying/pack_tests.cu
     copying/sample_tests.cpp
     copying/scatter_tests.cpp
-    copying/scatter_list_tests.cu
-    copying/scatter_struct_tests.cu
+    copying/scatter_list_tests.cpp
+    copying/scatter_struct_tests.cpp
     copying/segmented_gather_list_tests.cpp
     copying/shift_tests.cpp
     copying/slice_tests.cpp
diff --git a/cpp/tests/copying/scatter_list_tests.cu b/cpp/tests/copying/scatter_list_tests.cpp
similarity index 99%
rename from cpp/tests/copying/scatter_list_tests.cu
rename to cpp/tests/copying/scatter_list_tests.cpp
index 92f1d44c46f..64fdf6d00d5 100644
--- a/cpp/tests/copying/scatter_list_tests.cu
+++ b/cpp/tests/copying/scatter_list_tests.cpp
@@ -18,8 +18,6 @@
 
 #include <cudf/column/column_view.hpp>
 #include <cudf/copying.hpp>
-#include <cudf/detail/iterator.cuh>
-#include <cudf/detail/scatter.cuh>
 #include <cudf/lists/lists_column_view.hpp>
 #include <cudf/table/table.hpp>
 #include <cudf/table/table_view.hpp>
diff --git a/cpp/tests/copying/scatter_struct_tests.cu b/cpp/tests/copying/scatter_struct_tests.cpp
similarity index 93%
rename from cpp/tests/copying/scatter_struct_tests.cu
rename to cpp/tests/copying/scatter_struct_tests.cpp
index a9bb1980d53..3993664168b 100644
--- a/cpp/tests/copying/scatter_struct_tests.cu
+++ b/cpp/tests/copying/scatter_struct_tests.cpp
@@ -291,3 +291,24 @@ TYPED_TEST(TypedStructScatterTest, ScatterStructOfListsTest)
   auto const scatter_map = int32s_col{-3, -2, -1, 5, 4, 3, 2}.release();
   test_scatter(structs_src, structs_tgt, structs_expected, scatter_map);
 }
+
+TYPED_TEST(TypedStructScatterTest, ScatterSourceSmallerThanTarget)
+{
+  using namespace cudf;
+  using namespace cudf::test;
+
+  using fixed_width     = fixed_width_column_wrapper<TypeParam, int32_t>;
+  using structs_col     = structs_column_wrapper;
+  using scatter_map_col = fixed_width_column_wrapper<offset_type, int32_t>;
+
+  auto source_child   = fixed_width{22, 55};
+  auto target_child   = fixed_width{0, 1, 2, 3, 4, 5, 6};
+  auto expected_child = fixed_width{0, 1, 22, 3, 4, 55, 6};
+
+  auto const source      = structs_col{{source_child}}.release();
+  auto const target      = structs_col{{target_child}}.release();
+  auto const scatter_map = scatter_map_col{2, 5}.release();
+  auto const expected    = structs_col{{expected_child}}.release();
+
+  test_scatter(source, target, expected, scatter_map);
+}