Method CreateColumnWriterForColumn

cpp/src/parquet/file_writer.cc:291–325

  }

  std::shared_ptr<ColumnWriter> CreateColumnWriterForColumn(
      ColumnChunkMetaDataBuilder* col_meta, int32_t column_ordinal) const {
    const auto& path = col_meta->descr()->path();
    const ColumnProperties& column_properties = properties_->column_properties(path);
    auto meta_encryptor =
        file_encryptor_ ? file_encryptor_->GetColumnMetaEncryptor(path->ToDotString())
                        : nullptr;
    auto data_encryptor =
        file_encryptor_ ? file_encryptor_->GetColumnDataEncryptor(path->ToDotString())
                        : nullptr;
    auto ci_builder = page_index_builder_ && column_properties.page_index_enabled()
                          ? page_index_builder_->GetColumnIndexBuilder(column_ordinal)
                          : nullptr;
    auto oi_builder = page_index_builder_ && column_properties.page_index_enabled()
                          ? page_index_builder_->GetOffsetIndexBuilder(column_ordinal)
                          : nullptr;
    BloomFilter* bloom_filter = nullptr;
    if (bloom_filter_builder_) {
      bloom_filter = bloom_filter_builder_->CreateBloomFilter(column_ordinal);
    }
    const CodecOptions* codec_options = column_properties.codec_options()
                                            ? column_properties.codec_options().get()
                                            : nullptr;
    CodecOptions default_codec_options;
    if (!codec_options) {
      codec_options = &default_codec_options;
    }
    DCHECK_NE(nullptr, codec_options);
    std::unique_ptr<PageWriter> pager = PageWriter::Open(
        sink_, column_properties.compression(), col_meta, row_group_ordinal_,
        static_cast<int16_t>(column_ordinal), properties_->memory_pool(),
        buffered_row_group_, meta_encryptor, data_encryptor,
        properties_->page_checksum_enabled(), ci_builder, oi_builder, *codec_options);
    return ColumnWriter::Make(col_meta, std::move(pager), properties_, bloom_filter);
  }

  // If buffered_row_group_ is false, only column_writers_[0] is used as current writer.
  // If buffered_row_group_ is true, multiple column writers are used.