Method TestMergeMinMax

cpp/src/parquet/statistics_test.cc:706–743

If all values in a page are null or nan, its stats should not set min-max. Merging its stats with another page having good min-max stats should not drop the valid min-max from the latter page.

  // Merging its stats with another page having good min-max stats should not
  // drop the valid min-max from the latter page.
  void TestMergeMinMax() {
    this->GenerateData(1000);
    // Create a statistics object without min-max.
    std::shared_ptr<TypedStatistics<TestType>> statistics1;
    {
      statistics1 = MakeStatistics<TestType>(this->schema_.Column(0));
      statistics1->Update(this->values_ptr_, /*num_values=*/0,
                          /*null_count=*/this->values_.size());
      auto encoded_stats1 = statistics1->Encode();
      EXPECT_FALSE(statistics1->HasMinMax());
      EXPECT_FALSE(encoded_stats1.has_min);
      EXPECT_FALSE(encoded_stats1.has_max);
      EXPECT_EQ(encoded_stats1.is_max_value_exact, std::nullopt);
      EXPECT_EQ(encoded_stats1.is_min_value_exact, std::nullopt);
    }
    // Create a statistics object with min-max.
    std::shared_ptr<TypedStatistics<TestType>> statistics2;
    {
      statistics2 = MakeStatistics<TestType>(this->schema_.Column(0));
      statistics2->Update(this->values_ptr_, this->values_.size(), 0);
      auto encoded_stats2 = statistics2->Encode();
      EXPECT_TRUE(statistics2->HasMinMax());
      EXPECT_TRUE(encoded_stats2.has_min);
      EXPECT_TRUE(encoded_stats2.has_max);
      EXPECT_EQ(encoded_stats2.is_min_value_exact, std::make_optional(true));
      EXPECT_EQ(encoded_stats2.is_max_value_exact, std::make_optional(true));
    }
    VerifyMergedStatistics(*statistics1, *statistics2,
                           [](TypedStatistics<TestType>* merged_statistics) {
                             EXPECT_TRUE(merged_statistics->HasMinMax());
                             EXPECT_TRUE(merged_statistics->Encode().has_min);
                             EXPECT_TRUE(merged_statistics->Encode().has_max);
                             EXPECT_EQ(merged_statistics->Encode().is_min_value_exact,
                                       std::make_optional(true));
                             EXPECT_EQ(merged_statistics->Encode().is_max_value_exact,
                                       std::make_optional(true));
                           });
  }

  // Default statistics should have null_count even if no nulls is written.
  // However, if statistics is created from thrift message, it might not