Method BuildDataPageV1

cpp/src/parquet/column_writer.cc:1005–1049

}

void ColumnWriterImpl::BuildDataPageV1(int64_t definition_levels_rle_size,
                                       int64_t repetition_levels_rle_size,
                                       int64_t uncompressed_size,
                                       const std::shared_ptr<Buffer>& values) {
  // Use Arrow::Buffer::shrink_to_fit = false
  // underlying buffer only keeps growing. Resize to a smaller size does not reallocate.
  PARQUET_THROW_NOT_OK(uncompressed_data_->Resize(uncompressed_size, false));
  ConcatenateBuffers(definition_levels_rle_size, repetition_levels_rle_size, values,
                     uncompressed_data_->mutable_data());
  auto [page_stats, page_size_stats] = GetPageStatistics();
  page_stats.ApplyStatSizeLimits(properties_->max_statistics_size(descr_->path()));
  page_stats.set_is_signed(SortOrder::SIGNED == descr_->sort_order());
  ResetPageStatistics();

  std::shared_ptr<Buffer> compressed_data;
  if (pager_->has_compressor()) {
    pager_->Compress(*(uncompressed_data_.get()), compressor_temp_buffer_.get());
    compressed_data = compressor_temp_buffer_;
  } else {
    compressed_data = uncompressed_data_;
  }

  int32_t num_values = static_cast<int32_t>(num_buffered_values_);
  int64_t first_row_index = rows_written_ - num_buffered_rows_;

  // Write the page to OutputStream eagerly if there is no dictionary or
  // if dictionary encoding has fallen back to PLAIN
  if (has_dictionary_ && !fallback_) {  // Save pages until end of dictionary encoding
    PARQUET_ASSIGN_OR_THROW(
        auto compressed_data_copy,
        compressed_data->CopySlice(0, compressed_data->size(), allocator_));
    std::unique_ptr<DataPage> page_ptr = std::make_unique<DataPageV1>(
        compressed_data_copy, num_values, encoding_, Encoding::RLE, Encoding::RLE,
        uncompressed_size, std::move(page_stats), first_row_index,
        std::move(page_size_stats));
    total_compressed_bytes_ += page_ptr->size() + sizeof(format::PageHeader);

    data_pages_.push_back(std::move(page_ptr));
  } else {  // Eagerly write pages
    DataPageV1 page(compressed_data, num_values, encoding_, Encoding::RLE, Encoding::RLE,
                    uncompressed_size, std::move(page_stats), first_row_index,
                    std::move(page_size_stats));
    WriteDataPage(page);
  }
}

void ColumnWriterImpl::BuildDataPageV2(int64_t definition_levels_rle_size,
                                       int64_t repetition_levels_rle_size,