Function GenRandomUniqueRecords

cpp/src/arrow/acero/hash_join_node_test.cc:506–540

Generate random arrays given list of data types and null probabilities. Make sure that all generated records are unique. The actual number of generated records may be lower than desired because duplicates will be removed without replacement.

// will be removed without replacement.
//
std::vector<std::shared_ptr<Array>> GenRandomUniqueRecords(
    Random64Bit& rng, const RandomDataTypeVector& data_types, int num_desired,
    int* num_actual) {
  std::vector<std::shared_ptr<Array>> result =
      GenRandomRecords(rng, data_types.data_types, num_desired);

  ExecContext* ctx = default_exec_context();
  std::vector<TypeHolder> val_types;
  for (size_t i = 0; i < result.size(); ++i) {
    val_types.push_back(result[i]->type());
  }
  RowEncoder encoder;
  encoder.Init(val_types, ctx);
  ExecBatch batch({}, num_desired);
  batch.values.resize(result.size());
  for (size_t i = 0; i < result.size(); ++i) {
    batch.values[i] = result[i];
  }
  Status status = encoder.EncodeAndAppend(ExecSpan(batch));
  ARROW_DCHECK(status.ok());

  std::unordered_map<std::string, int> uniques;
  std::vector<int32_t> ids;
  for (int i = 0; i < num_desired; ++i) {
    if (uniques.find(encoder.encoded_row(i)) == uniques.end()) {
      uniques.insert(std::make_pair(encoder.encoded_row(i), i));
      ids.push_back(i);
    }
  }
  *num_actual = static_cast<int>(uniques.size());

  std::vector<std::shared_ptr<Array>> output;
  TakeUsingVector(ctx, result, ids, &output);
  return output;
}

std::vector<bool> NullInKey(const std::vector<JoinKeyCmp>& cmp,
                            const std::vector<std::shared_ptr<Array>>& key) {