Function ExecNode_Join

r/src/compute-exec.cpp:409–449

[[acero::export]]

// [[acero::export]]
std::shared_ptr<acero::ExecNode> ExecNode_Join(
    const std::shared_ptr<acero::ExecNode>& input, acero::JoinType join_type,
    const std::shared_ptr<acero::ExecNode>& right_data,
    std::vector<std::string> left_keys, std::vector<std::string> right_keys,
    std::vector<std::string> left_output, std::vector<std::string> right_output,
    std::string output_suffix_for_left, std::string output_suffix_for_right,
    bool na_matches) {
  std::vector<arrow::FieldRef> left_refs, right_refs, left_out_refs, right_out_refs;
  std::vector<acero::JoinKeyCmp> key_cmps;
  for (auto&& name : left_keys) {
    left_refs.emplace_back(std::move(name));
    // Populate key_cmps in this loop, one for each key
    // Note that Acero supports having different values for each key, but dplyr
    // only supports one value for all keys, so we're only going to support that
    // for now.
    key_cmps.emplace_back(na_matches ? acero::JoinKeyCmp::IS : acero::JoinKeyCmp::EQ);
  }
  for (auto&& name : right_keys) {
    right_refs.emplace_back(std::move(name));
  }
  for (auto&& name : left_output) {
    left_out_refs.emplace_back(std::move(name));
  }
  // dplyr::semi_join => LEFT_SEMI; dplyr::anti_join => LEFT_ANTI
  // So ignoring RIGHT_SEMI and RIGHT_ANTI here because dplyr doesn't implement them.
  if (join_type != acero::JoinType::LEFT_SEMI &&
      join_type != acero::JoinType::LEFT_ANTI) {
    // Don't include out_refs in semi/anti join
    for (auto&& name : right_output) {
      right_out_refs.emplace_back(std::move(name));
    }
  }

  return MakeExecNodeOrStop(
      "hashjoin", input->plan(), {input.get(), right_data.get()},
      acero::HashJoinNodeOptions{join_type, std::move(left_refs), std::move(right_refs),
                                 std::move(left_out_refs), std::move(right_out_refs),
                                 std::move(key_cmps), compute::literal(true),
                                 std::move(output_suffix_for_left),
                                 std::move(output_suffix_for_right)});
}

// [[acero::export]]
std::shared_ptr<acero::ExecNode> ExecNode_Union(