| 1371 | } |
| 1372 | |
| 1373 | void verifyFlatmapStructEncoding( |
| 1374 | memory::MemoryPool* pool, |
| 1375 | const std::string& filename, |
| 1376 | const std::vector<int32_t>& keysAsFields, |
| 1377 | const std::vector<int32_t>& keysToSelect, |
| 1378 | size_t batchSize = 1000) { |
| 1379 | dwio::common::ReaderOptions readerOpts{pool}; |
| 1380 | auto reader = DwrfReader::create( |
| 1381 | createFileBufferedInput(filename, readerOpts.getMemoryPool()), |
| 1382 | readerOpts); |
| 1383 | |
| 1384 | const std::string projectedColumn = "map1"; |
| 1385 | const vector_size_t projectedColumnIndex = 1; |
| 1386 | const std::vector<std::string> columnSelections = keysToSelect.empty() |
| 1387 | ? std::vector<std::string>{projectedColumn} |
| 1388 | : std::vector<std::string>{ |
| 1389 | projectedColumn + "#[" + folly::join(", ", keysToSelect) + "]"}; |
| 1390 | |
| 1391 | auto cs = std::make_shared<ColumnSelector>( |
| 1392 | std::dynamic_pointer_cast<const RowType>(HiveTypeParser().parse("struct<\ |
| 1393 | id:int,\ |
| 1394 | map1:map<int, array<float>>,\ |
| 1395 | map2:map<string, map<smallint,bigint>>,\ |
| 1396 | map3:map<int,int>,\ |
| 1397 | map4:map<int,struct<field1:int,field2:float,field3:string>>,\ |
| 1398 | memo:string>")), |
| 1399 | columnSelections); |
| 1400 | |
| 1401 | RowReaderOptions rowReaderOpts; |
| 1402 | rowReaderOpts.select(cs); |
| 1403 | |
| 1404 | auto mapEncodingReader = reader->createRowReader(rowReaderOpts); |
| 1405 | |
| 1406 | rowReaderOpts.setFlatmapNodeIdsAsStruct( |
| 1407 | makeStructEncodingOption(*cs, "map1", keysAsFields)); |
| 1408 | auto structEncodingReader = reader->createRowReader(rowReaderOpts); |
| 1409 | |
| 1410 | const auto compare = [&]() { |
| 1411 | VectorPtr batchMap; |
| 1412 | VectorPtr batchStruct; |
| 1413 | |
| 1414 | do { |
| 1415 | bool resultMap = mapEncodingReader->next(batchSize, batchMap); |
| 1416 | bool resultStruct = structEncodingReader->next(batchSize, batchStruct); |
| 1417 | |
| 1418 | EXPECT_EQ(resultMap, resultStruct); |
| 1419 | if (!resultMap) { |
| 1420 | break; |
| 1421 | } |
| 1422 | |
| 1423 | // verify current batch |
| 1424 | auto rowMapEncoding = batchMap->as<RowVector>(); |
| 1425 | auto rowStructEncoding = batchStruct->as<RowVector>(); |
| 1426 | |
| 1427 | EXPECT_EQ(rowMapEncoding->size(), rowStructEncoding->size()); |
| 1428 | |
| 1429 | for (size_t i = 0; i < keysAsFields.size(); ++i) { |
| 1430 | verifyMapColumnEqual( |
no test coverage detected