| 284 | } |
| 285 | |
| 286 | int64_t WriteDictionaryPage(const DictionaryPage& page) override { |
| 287 | int64_t uncompressed_size = page.buffer()->size(); |
| 288 | if (uncompressed_size > std::numeric_limits<int32_t>::max()) { |
| 289 | throw ParquetException( |
| 290 | "Uncompressed dictionary page size overflows INT32_MAX. Size:", |
| 291 | uncompressed_size); |
| 292 | } |
| 293 | std::shared_ptr<Buffer> compressed_data; |
| 294 | if (has_compressor()) { |
| 295 | auto buffer = std::static_pointer_cast<ResizableBuffer>( |
| 296 | AllocateBuffer(pool_, uncompressed_size)); |
| 297 | Compress(*(page.buffer().get()), buffer.get()); |
| 298 | compressed_data = std::static_pointer_cast<Buffer>(buffer); |
| 299 | } else { |
| 300 | compressed_data = page.buffer(); |
| 301 | } |
| 302 | |
| 303 | format::DictionaryPageHeader dict_page_header; |
| 304 | dict_page_header.__set_num_values(page.num_values()); |
| 305 | dict_page_header.__set_encoding(ToThrift(page.encoding())); |
| 306 | dict_page_header.__set_is_sorted(page.is_sorted()); |
| 307 | |
| 308 | const uint8_t* output_data_buffer = compressed_data->data(); |
| 309 | if (compressed_data->size() > std::numeric_limits<int32_t>::max()) { |
| 310 | throw ParquetException( |
| 311 | "Compressed dictionary page size overflows INT32_MAX. Size: ", |
| 312 | uncompressed_size); |
| 313 | } |
| 314 | int32_t output_data_len = static_cast<int32_t>(compressed_data->size()); |
| 315 | |
| 316 | if (data_encryptor_.get()) { |
| 317 | UpdateEncryption(encryption::kDictionaryPage); |
| 318 | PARQUET_THROW_NOT_OK(encryption_buffer_->Resize( |
| 319 | data_encryptor_->CiphertextLength(output_data_len), false)); |
| 320 | output_data_len = |
| 321 | data_encryptor_->Encrypt(compressed_data->span_as<uint8_t>(), |
| 322 | encryption_buffer_->mutable_span_as<uint8_t>()); |
| 323 | output_data_buffer = encryption_buffer_->data(); |
| 324 | } |
| 325 | |
| 326 | format::PageHeader page_header; |
| 327 | page_header.__set_type(format::PageType::DICTIONARY_PAGE); |
| 328 | page_header.__set_uncompressed_page_size(static_cast<int32_t>(uncompressed_size)); |
| 329 | page_header.__set_compressed_page_size(static_cast<int32_t>(output_data_len)); |
| 330 | page_header.__set_dictionary_page_header(dict_page_header); |
| 331 | if (page_checksum_verification_) { |
| 332 | uint32_t crc32 = |
| 333 | ::arrow::internal::crc32(/* prev */ 0, output_data_buffer, output_data_len); |
| 334 | page_header.__set_crc(static_cast<int32_t>(crc32)); |
| 335 | } |
| 336 | |
| 337 | PARQUET_ASSIGN_OR_THROW(int64_t start_pos, sink_->Tell()); |
| 338 | if (dictionary_page_offset_ == 0) { |
| 339 | dictionary_page_offset_ = start_pos; |
| 340 | } |
| 341 | |
| 342 | if (meta_encryptor_) { |
| 343 | UpdateEncryption(encryption::kDictionaryPageHeader); |
nothing calls this directly
no test coverage detected