Function ScanOnly

cpp/src/arrow/dataset/scanner_benchmark.cc:143–170

}

void ScanOnly(
    size_t num_batches, size_t batch_size, const std::string& factory_name,
    std::function<Result<std::shared_ptr<acero::ExecNodeOptions>>(size_t, size_t)>
        options_factory) {
  // ensure arrow::dataset node factories are in the registry
  ::arrow::dataset::internal::Initialize();

  RecordBatchVector batches = GetBatches(num_batches, batch_size);

  std::shared_ptr<Dataset> dataset =
      std::make_shared<InMemoryDataset>(GetSchema(), batches);

  ASSERT_OK_AND_ASSIGN(std::shared_ptr<acero::ExecNodeOptions> node_options,
                       options_factory(num_batches, batch_size));

  // construct the plan
  acero::Declaration scan(factory_name, std::move(node_options));

  ASSERT_OK_AND_ASSIGN(std::shared_ptr<Table> collected,
                       acero::DeclarationToTable(std::move(scan)));

  ASSERT_GT(collected->num_rows(), 0);
  if (factory_name == "scan") {
    ASSERT_EQ(collected->num_columns(), 6);
  } else if (factory_name == "scan2") {
    ASSERT_EQ(collected->num_columns(), 2);
  }
}

static constexpr int kScanIdx = 0;
static constexpr int kScanV2Idx = 1;