Method WriteDictionaryPage

cpp/src/parquet/column_writer.cc:285–353

  }

  int64_t WriteDictionaryPage(const DictionaryPage& page) override {
    int64_t uncompressed_size = page.buffer()->size();
    if (uncompressed_size > std::numeric_limits<int32_t>::max()) {
      throw ParquetException(
          "Uncompressed dictionary page size overflows INT32_MAX. Size:",
          uncompressed_size);
    }
    std::shared_ptr<Buffer> compressed_data;
    if (has_compressor()) {
      auto buffer = std::static_pointer_cast<ResizableBuffer>(
          AllocateBuffer(pool_, uncompressed_size));
      Compress(*(page.buffer().get()), buffer.get());
      compressed_data = std::static_pointer_cast<Buffer>(buffer);
    } else {
      compressed_data = page.buffer();
    }

    format::DictionaryPageHeader dict_page_header;
    dict_page_header.__set_num_values(page.num_values());
    dict_page_header.__set_encoding(ToThrift(page.encoding()));
    dict_page_header.__set_is_sorted(page.is_sorted());

    const uint8_t* output_data_buffer = compressed_data->data();
    if (compressed_data->size() > std::numeric_limits<int32_t>::max()) {
      throw ParquetException(
          "Compressed dictionary page size overflows INT32_MAX. Size: ",
          uncompressed_size);
    }
    int32_t output_data_len = static_cast<int32_t>(compressed_data->size());

    if (data_encryptor_.get()) {
      UpdateEncryption(encryption::kDictionaryPage);
      PARQUET_THROW_NOT_OK(encryption_buffer_->Resize(
          data_encryptor_->CiphertextLength(output_data_len), false));
      output_data_len =
          data_encryptor_->Encrypt(compressed_data->span_as<uint8_t>(),
                                   encryption_buffer_->mutable_span_as<uint8_t>());
      output_data_buffer = encryption_buffer_->data();
    }

    format::PageHeader page_header;
    page_header.__set_type(format::PageType::DICTIONARY_PAGE);
    page_header.__set_uncompressed_page_size(static_cast<int32_t>(uncompressed_size));
    page_header.__set_compressed_page_size(static_cast<int32_t>(output_data_len));
    page_header.__set_dictionary_page_header(dict_page_header);
    if (page_checksum_verification_) {
      uint32_t crc32 =
          ::arrow::internal::crc32(/* prev */ 0, output_data_buffer, output_data_len);
      page_header.__set_crc(static_cast<int32_t>(crc32));
    }

    PARQUET_ASSIGN_OR_THROW(int64_t start_pos, sink_->Tell());
    if (dictionary_page_offset_ == 0) {
      dictionary_page_offset_ = start_pos;
    }

    if (meta_encryptor_) {
      UpdateEncryption(encryption::kDictionaryPageHeader);