Fixes sliced list and struct column bug in JSON chunked writer (#13108)

Fixes the OOM access error while using chunked JSON writer on list columns.
The issue is present in struct columns also, which is fixed in this change.
Fixes #13030

Authors:
  - Karthikeyan (https://github.com/karthikeyann)
  - GALI PREM SAGAR (https://github.com/galipremsagar)

Approvers:
  - Nghia Truong (https://github.com/ttnghia)
  - David Wendt (https://github.com/davidwendt)

URL: #13108

cpp/src/io/json/write_json.cu

@@ -20,6 +20,7 @@
  */
 
 #include <io/csv/durations.hpp>
+#include <lists/utilities.hpp>
 
 #include <cudf/column/column_device_view.cuh>
 #include <cudf/column/column_factories.hpp>
@@ -437,15 +438,16 @@ struct column_to_strings_fn {
       };
       auto child_string = cudf::strings::detail::replace_nulls(
         child_string_with_null()->view(), narep, stream_, rmm::mr::get_current_device_resource());
-      auto const list_child_string =
-        column_view(column.type(),
-                    column.size(),
-                    column.head(),
-                    column.null_mask(),
-                    column.null_count(),
-                    column.offset(),
-                    {lists_column_view(column).offsets(), child_string->view()});
-      return strings::detail::join_list_elements(lists_column_view(list_child_string),
+      auto new_offsets = cudf::lists::detail::get_normalized_offsets(
+        lists_column_view(column), stream_, rmm::mr::get_current_device_resource());
+      auto const list_child_string = make_lists_column(
+        column.size(),
+        std::move(new_offsets),
+        std::move(child_string),
+        column.null_count(),
+        cudf::detail::copy_bitmask(column, stream_, rmm::mr::get_current_device_resource()),
+        stream_);
+      return strings::detail::join_list_elements(lists_column_view(*list_child_string),
                                                  string_scalar{","},
                                                  narep,
                                                  strings::separator_on_nulls::YES,
@@ -470,7 +472,11 @@ struct column_to_strings_fn {
   std::enable_if_t<std::is_same_v<column_type, cudf::struct_view>, std::unique_ptr<column>>
   operator()(column_view const& column, host_span<column_name_info const> children_names) const
   {
-    auto col_string = operator()(column.child_begin(), column.child_end(), children_names);
+    auto const child_it = cudf::detail::make_counting_transform_iterator(
+      0, [structs_view = structs_column_view{column}](auto const child_idx) {
+        return structs_view.get_sliced_child(child_idx);
+      });
+    auto col_string = operator()(child_it, child_it + column.num_children(), children_names);
     col_string->set_null_mask(cudf::detail::copy_bitmask(column, stream_, mr_),
                               column.null_count());
     return col_string;

cpp/tests/io/json_writer.cpp

@@ -394,4 +394,49 @@ TEST_F(JsonWriterTest, NullList)
   EXPECT_EQ(expected, std::string(out_buffer.data(), out_buffer.size()));
 }
 
+TEST_F(JsonWriterTest, ChunkedNested)
+{
+  std::string const data = R"(
+{"a": 1, "b": -2, "c": {      }, "e": [{"f": 1}]}
+{"a": 2, "b": -2, "c": {      }, "e": null}
+{"a": 3, "b": -2, "c": {"d": 9}, "e": [{"f": 2}, null]}
+{"a": 4, "b": -2, "c": {"d": 16}, "e": [{"f": 3}, {"f": 4}, {"f": 5}]}
+{"a": 5, "b": -2, "c": {      }, "e": []}
+{"a": 6, "b": -2, "c": {"d": 36}, "e": [{"f": 6}]}
+{"a": 7, "b": -2, "c": {"d": 49}, "e": [{"f": 7}]}
+{"a": 8, "b": -2, "c": {"d": 64}, "e": [{"f": 8}]}
+{"a": 9, "b": -2, "c": {"d": 81}, "e": [{"f": 9}]}
+)";
+  cudf::io::json_reader_options in_options =
+    cudf::io::json_reader_options::builder(cudf::io::source_info{data.data(), data.size()})
+      .lines(true);
+
+  cudf::io::table_with_metadata result = cudf::io::read_json(in_options);
+  cudf::table_view tbl_view            = result.tbl->view();
+  cudf::io::table_metadata mt{result.metadata};
+
+  std::vector<char> out_buffer;
+  auto destination     = cudf::io::sink_info(&out_buffer);
+  auto options_builder = cudf::io::json_writer_options_builder(destination, tbl_view)
+                           .include_nulls(false)
+                           .metadata(mt)
+                           .lines(true)
+                           .na_rep("null")
+                           .rows_per_chunk(8);
+
+  cudf::io::write_json(options_builder.build(), rmm::mr::get_current_device_resource());
+  std::string const expected =
+    R"({"a":1,"b":-2,"c":{},"e":[{"f":1}]}
+{"a":2,"b":-2,"c":{}}
+{"a":3,"b":-2,"c":{"d":9},"e":[{"f":2},null]}
+{"a":4,"b":-2,"c":{"d":16},"e":[{"f":3},{"f":4},{"f":5}]}
+{"a":5,"b":-2,"c":{},"e":[]}
+{"a":6,"b":-2,"c":{"d":36},"e":[{"f":6}]}
+{"a":7,"b":-2,"c":{"d":49},"e":[{"f":7}]}
+{"a":8,"b":-2,"c":{"d":64},"e":[{"f":8}]}
+{"a":9,"b":-2,"c":{"d":81},"e":[{"f":9}]}
+)";
+  EXPECT_EQ(expected, std::string(out_buffer.data(), out_buffer.size()));
+}
+
 CUDF_TEST_PROGRAM_MAIN()