Function WriteTableToBuffer

cpp/src/parquet/chunker_internal_test.cc:335–363

}

Result<std::shared_ptr<Buffer>> WriteTableToBuffer(
    const std::shared_ptr<Table>& table, int64_t min_chunk_size, int64_t max_chunk_size,
    int64_t row_group_length = 1024 * 1024, bool enable_dictionary = false,
    ParquetDataPageVersion data_page_version = ParquetDataPageVersion::V1) {
  auto sink = CreateOutputStream();

  auto builder = WriterProperties::Builder();
  builder.enable_content_defined_chunking()->content_defined_chunking_options(
      {min_chunk_size, max_chunk_size, /*norm_level=*/0});
  builder.data_page_version(data_page_version);
  if (enable_dictionary) {
    builder.enable_dictionary();
  } else {
    builder.disable_dictionary();
  }
  auto write_props = builder.build();
  auto arrow_props = ArrowWriterProperties::Builder().store_schema()->build();
  RETURN_NOT_OK(WriteTable(*table, default_memory_pool(), sink, row_group_length,
                           write_props, arrow_props));
  ARROW_ASSIGN_OR_RAISE(auto buffer, sink->Finish());

  // validate that the data correctly roundtrips
  ARROW_ASSIGN_OR_RAISE(auto readback, ReadTableFromBuffer(buffer));
  RETURN_NOT_OK(readback->ValidateFull());
  ARROW_RETURN_IF(!readback->Equals(*table),
                  Status::Invalid("Readback table not equal to original"));

  return buffer;
}

// Type to represent a list of chunks where each element is the size of the chunk.
using ChunkList = std::vector<int64_t>;