Function TestMultipleSeeksWithPredicates

c++/test/TestPredicatePushdown.cc:261–308

  }

  void TestMultipleSeeksWithPredicates(Reader* reader) {
    // Build search argument (x >= 300000 AND x < 600000) for column 'int1'. Only the 2nd
    // row group will be selected.
    std::unique_ptr<SearchArgument> sarg =
        SearchArgumentFactory::newBuilder()
            ->startAnd()
            .startNot()
            .lessThan("int1", PredicateDataType::LONG, Literal(static_cast<int64_t>(300000L)))
            .end()
            .lessThan("int1", PredicateDataType::LONG, Literal(static_cast<int64_t>(600000L)))
            .end()
            .build();
    RowReaderOptions rowReaderOpts;
    rowReaderOpts.searchArgument(std::move(sarg));
    auto rowReader = reader->createRowReader(rowReaderOpts);

    // Read only one row after each seek
    auto readBatch = rowReader->createRowBatch(1);
    auto& batch0 = dynamic_cast<StructVectorBatch&>(*readBatch);
    auto& batch1 = dynamic_cast<LongVectorBatch&>(*batch0.fields[0]);
    auto& batch2 = dynamic_cast<StringVectorBatch&>(*batch0.fields[1]);

    // Seek within the 1st row group will go to the start of the 2nd row group
    rowReader->seekToRow(10);
    EXPECT_TRUE(rowReader->next(*readBatch));
    EXPECT_EQ(1000, rowReader->getRowNumber()) << "Should start at the 2nd row group";
    EXPECT_EQ(1, readBatch->numElements);
    EXPECT_EQ(300000, batch1.data[0]);
    EXPECT_EQ("10000", std::string(batch2.data[0], static_cast<size_t>(batch2.length[0])));

    // Seek within the 2nd row group (1000 rows) which is selected by the search argument
    uint64_t seekRowNum[] = {1001, 1010, 1100, 1500, 1999};
    for (uint64_t pos : seekRowNum) {
      rowReader->seekToRow(pos);
      EXPECT_TRUE(rowReader->next(*readBatch));
      EXPECT_EQ(pos, rowReader->getRowNumber());
      EXPECT_EQ(1, readBatch->numElements);
      EXPECT_EQ(300 * pos, batch1.data[0]);
      EXPECT_EQ(std::to_string(10 * pos),
                std::string(batch2.data[0], static_cast<size_t>(batch2.length[0])));
    }

    // Seek advance the 2nd row group will go to the end of file
    rowReader->seekToRow(2000);
    EXPECT_FALSE(rowReader->next(*readBatch));
    EXPECT_EQ(3500, rowReader->getRowNumber());
    EXPECT_EQ(0, readBatch->numElements);
  }

  TEST(TestPredicatePushdown, testPredicatePushdown) {
    MemoryOutputStream memStream(DEFAULT_MEM_STREAM_SIZE);