| 394 | } |
| 395 | |
| 396 | std::shared_ptr<Page> SerializedPageReader::NextPage() { |
| 397 | ThriftDeserializer deserializer(properties_); |
| 398 | |
| 399 | // Loop here because there may be unhandled page types that we skip until |
| 400 | // finding a page that we do know what to do with |
| 401 | while (seen_num_values_ < total_num_values_) { |
| 402 | uint32_t header_size = 0; |
| 403 | uint32_t allowed_page_size = kDefaultPageHeaderSize; |
| 404 | |
| 405 | // Page headers can be very large because of page statistics |
| 406 | // We try to deserialize a larger buffer progressively |
| 407 | // until a maximum allowed header limit |
| 408 | while (true) { |
| 409 | PARQUET_ASSIGN_OR_THROW(auto view, stream_->Peek(allowed_page_size)); |
| 410 | if (view.size() == 0) return nullptr; |
| 411 | |
| 412 | // This gets used, then set by DeserializeThriftMsg |
| 413 | header_size = static_cast<uint32_t>(view.size()); |
| 414 | try { |
| 415 | if (meta_decryptor_ != nullptr) { |
| 416 | UpdateDecryption(meta_decryptor_.get(), encryption::kDictionaryPageHeader, |
| 417 | &data_page_header_aad_); |
| 418 | } |
| 419 | // Reset current page header to avoid unclearing the __isset flag. |
| 420 | current_page_header_ = format::PageHeader(); |
| 421 | deserializer.DeserializeMessage(reinterpret_cast<const uint8_t*>(view.data()), |
| 422 | &header_size, ¤t_page_header_, |
| 423 | meta_decryptor_.get()); |
| 424 | break; |
| 425 | } catch (std::exception& e) { |
| 426 | // Failed to deserialize. Double the allowed page header size and try again |
| 427 | std::stringstream ss; |
| 428 | ss << e.what(); |
| 429 | allowed_page_size *= 2; |
| 430 | if (allowed_page_size > max_page_header_size_) { |
| 431 | ss << "Deserializing page header failed.\n"; |
| 432 | throw ParquetException(ss.str()); |
| 433 | } |
| 434 | } |
| 435 | } |
| 436 | // Advance the stream offset |
| 437 | PARQUET_THROW_NOT_OK(stream_->Advance(header_size)); |
| 438 | |
| 439 | int32_t compressed_len = current_page_header_.compressed_page_size; |
| 440 | int32_t uncompressed_len = current_page_header_.uncompressed_page_size; |
| 441 | if (compressed_len < 0 || uncompressed_len < 0) { |
| 442 | throw ParquetException("Invalid page header"); |
| 443 | } |
| 444 | |
| 445 | EncodedStatistics data_page_statistics; |
| 446 | if (ShouldSkipPage(&data_page_statistics)) { |
| 447 | PARQUET_THROW_NOT_OK(stream_->Advance(compressed_len)); |
| 448 | continue; |
| 449 | } |
| 450 | |
| 451 | if (data_decryptor_ != nullptr) { |
| 452 | UpdateDecryption(data_decryptor_.get(), encryption::kDictionaryPage, |
| 453 | &data_page_aad_); |