Method TranslateMinimalBatch

cpp/src/arrow/csv/writer.cc:667–706

  }

  Status TranslateMinimalBatch(const RecordBatch& batch) {
    if (batch.num_rows() == 0) {
      return Status::OK();
    }
    offsets_.resize(batch.num_rows());
    std::fill(offsets_.begin(), offsets_.end(), 0);

    // Calculate relative offsets for each row (excluding delimiters)
    for (int32_t col = 0; col < static_cast<int32_t>(column_populators_.size()); col++) {
      RETURN_NOT_OK(
          column_populators_[col]->UpdateRowLengths(*batch.column(col), offsets_.data()));
    }
    // Calculate cumulative offsets for each row (including delimiters).
    // - before conversion: offsets_[i] = length of i-th row
    // - after conversion:  offsets_[i] = offset to the starting of i-th row buffer
    //   - offsets_[0] = 0
    //   - offsets_[i] = offsets_[i-1] + len(i-1-th row) + len(delimiters)
    // Delimiters: ',' * (num_columns - 1) + eol
    const int32_t delimiters_length =
        static_cast<int32_t>(batch.num_columns() - 1 + options_.eol.size());
    int64_t last_row_length = offsets_[0] + delimiters_length;
    offsets_[0] = 0;
    for (size_t row = 1; row < offsets_.size(); ++row) {
      const int64_t this_row_length = offsets_[row] + delimiters_length;
      offsets_[row] = offsets_[row - 1] + last_row_length;
      last_row_length = this_row_length;
    }
    // Resize the target buffer to required size. We assume batch to batch sizes
    // should be pretty close so don't shrink the buffer to avoid allocation churn.
    RETURN_NOT_OK(
        data_buffer_->Resize(offsets_.back() + last_row_length, /*shrink_to_fit=*/false));

    // Use the offsets to populate contents.
    for (auto& populator : column_populators_) {
      RETURN_NOT_OK(populator->PopulateRows(
          reinterpret_cast<char*>(data_buffer_->mutable_data()), offsets_.data()));
    }
    DCHECK_EQ(data_buffer_->size(), offsets_.back());
    return Status::OK();
  }

  static constexpr int64_t kColumnSizeGuess = 8;
  io::OutputStream* sink_;