Method StartFile

cpp/src/parquet/file_writer.cc:528–583

  std::unique_ptr<BloomFilterBuilder> bloom_filter_builder_;

  void StartFile() {
    auto file_encryption_properties = properties_->file_encryption_properties();
    if (file_encryption_properties == nullptr) {
      // Unencrypted parquet files always start with PAR1
      PARQUET_THROW_NOT_OK(sink_->Write(kParquetMagic, 4));
    } else {
      // Check that all columns in columnEncryptionProperties exist in the schema.
      auto encrypted_columns = file_encryption_properties->encrypted_columns();
      // if columnEncryptionProperties is empty, every column in file schema will be
      // encrypted with footer key.
      if (encrypted_columns.size() != 0) {
        std::vector<std::string> column_path_vec;
        std::vector<std::string> column_root_vec;
        // First, save all column paths and root column names in schema.
        for (int i = 0; i < num_columns(); i++) {
          column_path_vec.push_back(schema_.Column(i)->path()->ToDotString());
          column_root_vec.push_back(schema_.Column(i)->path()->ToDotVector().at(0));
        }

        for (const auto& elem : encrypted_columns) {
          // Check if this column exists in schema.
          auto it = std::find(column_path_vec.begin(), column_path_vec.end(), elem.first);
          if (it == column_path_vec.end()) {
            // this column does not exist in the schema, this might be a root column name
            // like `a` while there are `a.key_value.key` and `a.key_value.value` in the
            // schema.
            it = std::find(column_root_vec.begin(), column_root_vec.end(), elem.first);
            if (it == column_root_vec.end()) {
              // no encrypted columns exist with this root column name
              std::stringstream ss;
              ss << "Encrypted column " + elem.first + " not in file schema: ";
              for (auto& cp : column_path_vec) {
                ss << cp << " ";
              }
              throw ParquetException(ss.str());
            }
          }
        }
      }

      file_encryptor_ = std::make_unique<InternalFileEncryptor>(
          file_encryption_properties, properties_->memory_pool());
      if (file_encryption_properties->encrypted_footer()) {
        PARQUET_THROW_NOT_OK(sink_->Write(kParquetEMagic, 4));
      } else {
        // Encrypted file with plaintext footer mode.
        PARQUET_THROW_NOT_OK(sink_->Write(kParquetMagic, 4));
      }
    }
    if (properties_->bloom_filter_enabled()) {
      bloom_filter_builder_ = BloomFilterBuilder::Make(schema(), properties_.get());
    }
    if (properties_->page_index_enabled()) {
      page_index_builder_ = PageIndexBuilder::Make(&schema_, file_encryptor_.get());
    }
  }
};