Method Finalize

cpp/src/arrow/compute/kernels/hash_aggregate.cc:1483–1538

struct GroupedDistinctImpl : public GroupedCountDistinctImpl {
  Result<Datum> Finalize() override {
    ARROW_ASSIGN_OR_RAISE(auto uniques, grouper_->GetUniques());
    ARROW_ASSIGN_OR_RAISE(
        auto groupings, Grouper::MakeGroupings(*uniques[1].array_as<UInt32Array>(),
                                               static_cast<uint32_t>(num_groups_), ctx_));
    ARROW_ASSIGN_OR_RAISE(
        auto list, Grouper::ApplyGroupings(*groupings, *uniques[0].make_array(), ctx_));
    const auto& values = list->values();
    DCHECK_EQ(values->offset(), 0);
    auto* offsets = list->value_offsets()->mutable_data_as<int32_t>();
    if (options_.mode == CountOptions::ALL ||
        (options_.mode == CountOptions::ONLY_VALID && values->null_count() == 0)) {
      return list;
    } else if (options_.mode == CountOptions::ONLY_VALID) {
      int32_t prev_offset = offsets[0];
      for (int64_t i = 0; i < list->length(); i++) {
        const int32_t slot_length = offsets[i + 1] - prev_offset;
        const int64_t null_count =
            slot_length - arrow::internal::CountSetBits(values->null_bitmap()->data(),
                                                        prev_offset, slot_length);
        DCHECK_LE(null_count, 1);
        const int32_t offset = null_count > 0 ? slot_length - 1 : slot_length;
        prev_offset = offsets[i + 1];
        offsets[i + 1] = offsets[i] + offset;
      }
      auto filter =
          std::make_shared<BooleanArray>(values->length(), values->null_bitmap());
      ARROW_ASSIGN_OR_RAISE(
          auto new_values,
          Filter(std::move(values), filter, FilterOptions(FilterOptions::DROP), ctx_));
      return std::make_shared<ListArray>(list->type(), list->length(),
                                         list->value_offsets(), new_values.make_array());
    }
    // ONLY_NULL
    if (values->null_count() == 0) {
      std::fill(offsets + 1, offsets + list->length() + 1, offsets[0]);
    } else {
      int32_t prev_offset = offsets[0];
      for (int64_t i = 0; i < list->length(); i++) {
        const int32_t slot_length = offsets[i + 1] - prev_offset;
        const int64_t null_count =
            slot_length - arrow::internal::CountSetBits(values->null_bitmap()->data(),
                                                        prev_offset, slot_length);
        const int32_t offset = null_count > 0 ? 1 : 0;
        prev_offset = offsets[i + 1];
        offsets[i + 1] = offsets[i] + offset;
      }
    }
    ARROW_ASSIGN_OR_RAISE(
        auto new_values,
        MakeArrayOfNull(out_type_,
                        list->length() > 0 ? offsets[list->length()] - offsets[0] : 0,
                        pool_));
    return std::make_shared<ListArray>(list->type(), list->length(),
                                       list->value_offsets(), std::move(new_values));
  }

  std::shared_ptr<DataType> out_type() const override { return list(out_type_); }