Function DoHashJoin

cpp/examples/arrow/join_example.cc:84–125

}

arrow::Status DoHashJoin() {
  ARROW_RETURN_NOT_OK(arrow::compute::Initialize());
  arrow::dataset::internal::Initialize();

  ARROW_ASSIGN_OR_RAISE(auto l_dataset, CreateDataSetFromCSVData(true));
  ARROW_ASSIGN_OR_RAISE(auto r_dataset, CreateDataSetFromCSVData(false));

  auto l_options = std::make_shared<arrow::dataset::ScanOptions>();
  // create empty projection: "default" projection where each field is mapped to a
  // field_ref
  l_options->projection = cp::project({}, {});

  auto r_options = std::make_shared<arrow::dataset::ScanOptions>();
  // create empty projection: "default" projection where each field is mapped to a
  // field_ref
  r_options->projection = cp::project({}, {});

  // construct the scan node
  auto l_scan_node_options = arrow::dataset::ScanNodeOptions{l_dataset, l_options};
  auto r_scan_node_options = arrow::dataset::ScanNodeOptions{r_dataset, r_options};

  arrow::acero::Declaration left{"scan", std::move(l_scan_node_options)};
  arrow::acero::Declaration right{"scan", std::move(r_scan_node_options)};

  arrow::acero::HashJoinNodeOptions join_opts{arrow::acero::JoinType::INNER,
                                              /*in_left_keys=*/{"lkey"},
                                              /*in_right_keys=*/{"rkey"},
                                              /*filter*/ arrow::compute::literal(true),
                                              /*output_suffix_for_left*/ "_l",
                                              /*output_suffix_for_right*/ "_r"};

  arrow::acero::Declaration hashjoin{
      "hashjoin", {std::move(left), std::move(right)}, join_opts};

  // expected columns l_a, l_b
  ARROW_ASSIGN_OR_RAISE(std::shared_ptr<arrow::Table> response_table,
                        arrow::acero::DeclarationToTable(std::move(hashjoin)));

  std::cout << "Results : " << response_table->ToString() << std::endl;

  return arrow::Status::OK();
}

int main(int argc, char** argv) {
  auto status = DoHashJoin();