Method TEST

cpp/src/arrow/dataset/dataset_test.cc:769–803

}

TEST(TestDictPartitionColumn, SelectPartitionColumnFilterPhysicalColumn) {
  auto partition_field = field("part", dictionary(int32(), utf8()));
  auto path = "/dataset/part=one/data.json";
  auto dictionary = ArrayFromJSON(utf8(), R"(["one"])");

  auto mock_fs = std::make_shared<fs::internal::MockFileSystem>(fs::kNoTime);
  ARROW_EXPECT_OK(mock_fs->CreateFile(path, R"([ {"phy_1": 111, "phy_2": 211} ])",
                                      /*recursive=*/true));

  auto physical_schema = SchemaFromNames({"phy_1", "phy_2"});
  auto format = std::make_shared<JSONRecordBatchFileFormat>(
      [=](const FileSource&) { return physical_schema; });

  FileSystemFactoryOptions options;
  options.partition_base_dir = "/dataset";
  options.partitioning = std::make_shared<HivePartitioning>(schema({partition_field}),
                                                            ArrayVector{dictionary});

  ASSERT_OK_AND_ASSIGN(auto factory,
                       FileSystemDatasetFactory::Make(mock_fs, {path}, format, options));

  ASSERT_OK_AND_ASSIGN(auto schema, factory->Inspect());

  ASSERT_OK_AND_ASSIGN(auto dataset, factory->Finish(schema));

  // Selects re-ordered virtual column with a filter on a physical column
  ASSERT_OK_AND_ASSIGN(auto scan_builder, dataset->NewScan());
  ASSERT_OK(scan_builder->Filter(equal(field_ref("phy_1"), literal(111))));
  ASSERT_OK(scan_builder->Project({"part"}));

  ASSERT_OK_AND_ASSIGN(auto scanner, scan_builder->Finish());
  ASSERT_OK_AND_ASSIGN(auto table, scanner->ToTable());
  AssertArraysEqual(*table->column(0)->chunk(0),
                    *ArrayFromJSON(partition_field->type(), R"(["one"])"));
}

}  // namespace dataset
}  // namespace arrow