| 337 | |
| 338 | template <typename ParquetType> |
| 339 | DeltaData<ParquetType> DeltaEncode(const std::vector<ParquetType>& input, int bit_width) { |
| 340 | using UnsignedParquetType = typename std::make_unsigned<ParquetType>::type; |
| 341 | const ParquetType base_value = input.empty() ? 0 : input[0]; |
| 342 | std::vector<ParquetType> deltas; |
| 343 | |
| 344 | /// Calculate the deltas. |
| 345 | for (int i = 1; i < input.size(); i++) { |
| 346 | UnsignedParquetType current = input[i]; |
| 347 | UnsignedParquetType previous = input[i - 1]; |
| 348 | ParquetType delta = current - previous; |
| 349 | deltas.push_back(delta); |
| 350 | } |
| 351 | |
| 352 | /// Offset the deltas. |
| 353 | const auto it = std::min_element(deltas.begin(), deltas.end()); |
| 354 | const ParquetType min_delta = (it == deltas.end()) ? 0 : *it; |
| 355 | |
| 356 | std::for_each(deltas.begin(), deltas.end(), [min_delta](ParquetType& element) { |
| 357 | element = ArithmeticUtil::AsUnsigned<std::minus>(element, min_delta); |
| 358 | }); |
| 359 | |
| 360 | // Bit pack the deltas. |
| 361 | const int bytes_required = BitUtil::RoundUpNumBytes(bit_width * deltas.size()); |
| 362 | std::vector<uint8_t> out_data(bytes_required); |
| 363 | |
| 364 | if (bytes_required > 0) { |
| 365 | BitWriter writer(out_data.data(), bytes_required); |
| 366 | if (bit_width > 0) { |
| 367 | for (const ParquetType delta : deltas) { |
| 368 | const UnsignedParquetType delta_unsigned = delta; |
| 369 | EXPECT_TRUE(writer.PutValue(delta_unsigned, bit_width)); |
| 370 | } |
| 371 | } |
| 372 | writer.Flush(); |
| 373 | } |
| 374 | |
| 375 | return DeltaData<ParquetType>(base_value, min_delta, deltas.size(), out_data); |
| 376 | } |
| 377 | |
| 378 | template <typename INT_T, typename ParquetType> |
| 379 | void RandomUnpackAndDeltaDecodeTest() { |