Function TEST

cpp/src/parquet/metadata_test.cc:89–276

}

TEST(Metadata, TestBuildAccess) {
  parquet::schema::NodeVector fields;
  parquet::schema::NodePtr root;
  parquet::SchemaDescriptor schema;

  WriterProperties::Builder prop_builder;

  std::shared_ptr<WriterProperties> props =
      prop_builder.version(ParquetVersion::PARQUET_2_6)->build();

  fields.push_back(parquet::schema::Int32("int_col", Repetition::REQUIRED));
  fields.push_back(parquet::schema::Float("float_col", Repetition::REQUIRED));
  root = parquet::schema::GroupNode::Make("schema", Repetition::REPEATED, fields);
  schema.Init(root);

  int64_t nrows = 1000;
  int32_t int_min = 100, int_max = 200;
  EncodedStatistics stats_int;
  stats_int.set_null_count(0)
      .set_distinct_count(nrows)
      .set_min(std::string(reinterpret_cast<const char*>(&int_min), 4))
      .set_max(std::string(reinterpret_cast<const char*>(&int_max), 4));
  EncodedStatistics stats_float;
  float float_min = 100.100f, float_max = 200.200f;
  stats_float.set_null_count(0)
      .set_distinct_count(nrows)
      .set_min(std::string(reinterpret_cast<const char*>(&float_min), 4))
      .set_max(std::string(reinterpret_cast<const char*>(&float_max), 4));

  // Generate the metadata
  auto f_accessor = GenerateTableMetaData(schema, props, nrows, stats_int, stats_float);

  std::string f_accessor_serialized_metadata = f_accessor->SerializeToString();
  uint32_t expected_len = static_cast<uint32_t>(f_accessor_serialized_metadata.length());

  // decoded_len is an in-out parameter
  uint32_t decoded_len = expected_len;
  auto f_accessor_copy =
      FileMetaData::Make(f_accessor_serialized_metadata.data(), &decoded_len);

  // Check that all of the serialized data is consumed
  ASSERT_EQ(expected_len, decoded_len);

  // Run this block twice, one for f_accessor, one for f_accessor_copy.
  // To make sure SerializedMetadata was deserialized correctly.
  std::vector<FileMetaData*> f_accessors = {f_accessor.get(), f_accessor_copy.get()};
  for (int loop_index = 0; loop_index < 2; loop_index++) {
    // file metadata
    ASSERT_EQ(nrows, f_accessors[loop_index]->num_rows());
    ASSERT_LE(0, static_cast<int>(f_accessors[loop_index]->size()));
    ASSERT_EQ(2, f_accessors[loop_index]->num_row_groups());
    ASSERT_EQ(ParquetVersion::PARQUET_2_6, f_accessors[loop_index]->version());
    ASSERT_EQ(DEFAULT_CREATED_BY, f_accessors[loop_index]->created_by());
    ASSERT_EQ(3, f_accessors[loop_index]->num_schema_elements());

    // row group1 metadata
    auto rg1_accessor = f_accessors[loop_index]->RowGroup(0);
    ASSERT_EQ(2, rg1_accessor->num_columns());