Function AppendUnionBatchInternal

cpp/src/arrow/adapters/orc/util.cc:276–311

template <typename UnionBuilderType>
Status AppendUnionBatchInternal(const liborc::Type* type,
                                liborc::ColumnVectorBatch* column_vector_batch,
                                int64_t offset, int64_t length, ArrayBuilder* abuilder) {
  auto builder = checked_cast<UnionBuilderType*>(abuilder);
  auto batch = checked_cast<liborc::UnionVectorBatch*>(column_vector_batch);

  auto union_type = checked_pointer_cast<UnionType>(abuilder->type());
  const std::vector<int8_t>& type_codes = union_type->type_codes();
  const unsigned char* tags = batch->tags.data();

  for (int64_t i = offset; i < length + offset; i++) {
    if (!batch->hasNulls || batch->notNull[i]) {
      auto child_id = tags[i];
      auto type_code = type_codes[child_id];
      RETURN_NOT_OK(builder->Append(type_code));

      auto child_type = type->getSubtype(child_id);
      auto child_batch = batch->children[child_id];
      auto start = static_cast<int64_t>(batch->offsets[i]);
      RETURN_NOT_OK(AppendBatch(child_type, child_batch, start, /*length=*/1,
                                builder->child_builder(child_id).get()));

      if constexpr (std::is_same_v<UnionBuilderType, SparseUnionBuilder>) {
        // Append null value to other child builders for sparse union type.
        for (int8_t field_id = 0; field_id < union_type->num_fields(); field_id++) {
          if (field_id != child_id) {
            RETURN_NOT_OK(builder->child_builder(field_id)->AppendNull());
          }
        }
      }
    } else {
      RETURN_NOT_OK(builder->AppendNull());
    }
  }
  return Status::OK();
}

Status AppendUnionBatch(const liborc::Type* type,
                        liborc::ColumnVectorBatch* column_vector_batch, int64_t offset,