Method TestMergeNullCount

cpp/src/parquet/statistics_test.cc:749–791

Default statistics should have null_count even if no nulls is written. However, if statistics is created from thrift message, it might not have null_count. Merging statistics from such page will result in an invalid null_count as well.

  // have null_count. Merging statistics from such page will result in an
  // invalid null_count as well.
  void TestMergeNullCount() {
    this->GenerateData(/*num_values=*/1000);

    // Page should have null-count even if no nulls
    std::shared_ptr<TypedStatistics<TestType>> statistics1;
    {
      statistics1 = MakeStatistics<TestType>(this->schema_.Column(0));
      statistics1->Update(this->values_ptr_, /*num_values=*/this->values_.size(),
                          /*null_count=*/0);
      auto encoded_stats1 = statistics1->Encode();
      EXPECT_TRUE(statistics1->HasNullCount());
      EXPECT_EQ(0, statistics1->null_count());
      EXPECT_TRUE(statistics1->Encode().has_null_count);
    }
    // Merge with null-count should also have null count
    VerifyMergedStatistics(*statistics1, *statistics1,
                           [](TypedStatistics<TestType>* merged_statistics) {
                             EXPECT_TRUE(merged_statistics->HasNullCount());
                             EXPECT_EQ(0, merged_statistics->null_count());
                             auto encoded = merged_statistics->Encode();
                             EXPECT_TRUE(encoded.has_null_count);
                             EXPECT_EQ(0, encoded.null_count);
                           });

    // When loaded from thrift, might not have null count.
    std::shared_ptr<TypedStatistics<TestType>> statistics2;
    {
      EncodedStatistics encoded_statistics2;
      encoded_statistics2.has_null_count = false;
      statistics2 = std::dynamic_pointer_cast<TypedStatistics<TestType>>(
          Statistics::Make(this->schema_.Column(0), &encoded_statistics2,
                           /*num_values=*/1000));
      EXPECT_FALSE(statistics2->Encode().has_null_count);
      EXPECT_FALSE(statistics2->HasNullCount());
    }

    // Merge without null-count should not have null count
    VerifyMergedStatistics(*statistics1, *statistics2,
                           [](TypedStatistics<TestType>* merged_statistics) {
                             EXPECT_FALSE(merged_statistics->HasNullCount());
                             EXPECT_FALSE(merged_statistics->Encode().has_null_count);
                           });
  }

  // statistics.all_null_value is used to build the page index.
  // If statistics doesn't have null count, all_null_value should be false.