Function WriteUnionBatch

cpp/src/arrow/adapters/orc/util.cc:815–861

template <typename UnionArrayType>
Status WriteUnionBatch(const Array& array, int64_t orc_offset,
                       liborc::ColumnVectorBatch* column_vector_batch) {
  auto union_array = checked_cast<const UnionArrayType*>(&array);
  auto batch = checked_cast<liborc::UnionVectorBatch*>(column_vector_batch);

  // Union array itself does not have nulls.
  batch->hasNulls = false;

  int64_t arrow_length = array.length();
  int64_t running_arrow_offset = 0, running_orc_offset = orc_offset;

  for (; running_arrow_offset < arrow_length;
       running_orc_offset++, running_arrow_offset++) {
    auto child_id = union_array->child_id(running_arrow_offset);
    batch->tags[running_orc_offset] = static_cast<unsigned char>(child_id);
    batch->notNull[running_orc_offset] = true;

    // Orc union batch is dense, so we need to find the previous row that has the same
    // type code (or child_id) to calculate the offset.
    batch->offsets[running_orc_offset] = 0;
    for (int64_t row = running_orc_offset - 1; row >= 0; row--) {
      if (batch->tags[row] == child_id) {
        batch->offsets[running_orc_offset] = batch->offsets[row] + 1;
        break;
      }
    }

    // Fill child batch.
    liborc::ColumnVectorBatch* child_batch = batch->children[child_id];
    int64_t child_array_orc_offset = batch->offsets[running_orc_offset];
    int64_t child_array_arrow_offset;

    if constexpr (std::is_same_v<UnionArrayType, DenseUnionArray>) {
      child_array_arrow_offset = union_array->value_offset(running_arrow_offset);
    } else {
      static_assert(std::is_same_v<UnionArrayType, SparseUnionArray>);
      child_array_arrow_offset = running_arrow_offset;
    }
    child_batch->resize(child_array_orc_offset + 1);

    std::shared_ptr<Array> child_array = union_array->field(child_id);
    RETURN_NOT_OK(WriteBatch(*(child_array->Slice(child_array_arrow_offset, 1)),
                             child_array_orc_offset, child_batch));
  }

  return Status::OK();
}

Status WriteBatch(const Array& array, int64_t orc_offset,
                  liborc::ColumnVectorBatch* column_vector_batch) {