Function DecodeDict

cpp/src/parquet/encoding_benchmark.cc:1083–1124

template <typename Type>
static void DecodeDict(const std::vector<typename Type::c_type>& values,
                       benchmark::State& state) {
  typedef typename Type::c_type T;
  int num_values = static_cast<int>(values.size());

  MemoryPool* allocator = default_memory_pool();
  std::shared_ptr<ColumnDescriptor> descr = Int64Schema(Repetition::REQUIRED);

  auto base_encoder =
      MakeEncoder(Type::type_num, Encoding::PLAIN, true, descr.get(), allocator);
  auto encoder =
      dynamic_cast<typename EncodingTraits<Type>::Encoder*>(base_encoder.get());
  auto dict_traits = dynamic_cast<DictEncoder<Type>*>(base_encoder.get());
  encoder->Put(values.data(), num_values);

  std::shared_ptr<ResizableBuffer> dict_buffer =
      AllocateBuffer(allocator, dict_traits->dict_encoded_size());

  std::shared_ptr<ResizableBuffer> indices =
      AllocateBuffer(allocator, encoder->EstimatedDataEncodedSize());

  dict_traits->WriteDict(dict_buffer->mutable_data());
  int actual_bytes = dict_traits->WriteIndices(indices->mutable_data(),
                                               static_cast<int>(indices->size()));

  PARQUET_THROW_NOT_OK(indices->Resize(actual_bytes));

  std::vector<T> decoded_values(num_values);
  for (auto _ : state) {
    auto dict_decoder = MakeTypedDecoder<Type>(Encoding::PLAIN, descr.get());
    dict_decoder->SetData(dict_traits->num_entries(), dict_buffer->data(),
                          static_cast<int>(dict_buffer->size()));

    auto decoder = MakeDictDecoder<Type>(descr.get());
    decoder->SetDict(dict_decoder.get());
    decoder->SetData(num_values, indices->data(), static_cast<int>(indices->size()));
    decoder->Decode(decoded_values.data(), num_values);
  }

  state.SetBytesProcessed(state.iterations() * num_values * sizeof(T));
  state.SetItemsProcessed(state.iterations() * num_values);
}

static void BM_DictDecodingInt64_repeats(benchmark::State& state) {
  typedef Int64Type Type;