Ignore row count when comparing datasets

In tests I would like to verify only small subset of existing rows in the table. Let’s say I have ‘users’ table and I need to insert common users rows for most of the tests:

common.yml

users:
  - id: 1
    name: User1
  - id: 2
    name: User2

Then in some test I want to check update of the user name. For example the test will update name of user with id = 2 from ‘User2’ to ‘Another user 2’. So I would like to have the following in expected dataset:

expected.yml

users:
  - id: 2
    name: Another user 2

In current implementation the test will fail because row count is obviously different - 1 row expected and 2 rows in actual dataset. I was able to implement a workaround using PrimaryKeyFilteredTableWrapper in DataSetAssertion class:

public class DataSetAssertion {
    private static final DataSetAssert INSTANCE = new DataSetAssert();

    private static ITable filterActualDataset(ITable expectedDataSet, ITable actualDataSet) throws DataSetException {
        String pkName = actualDataSet.getTableMetaData().getPrimaryKeys()[0].getColumnName();
        Set<Object> pkList = new HashSet<>();
        for (int i = 0; i < expectedDataSet.getRowCount(); i++) {
            pkList.add(expectedDataSet.getValue(i, pkName));
        }

        return new PrimaryKeyFilteredTableWrapper(actualDataSet, pkList);


    }

    public static void assertEqualsIgnoreCols(ITable expectedDataSet,
                                              ITable actualDataSet, String[] ignoreCols)
            throws DatabaseUnitException {
        INSTANCE.assertEqualsIgnoreCols(expectedDataSet, filterActualDataset(expectedDataSet, actualDataSet), ignoreCols);
    }
}

Can we add some flag/enum to com.github.database.rider.core.api.dataset.ExpectedDataSet annotation to change default checking logic?

[rmpestano]

Thank you very much @iceberk, I’ll review ASAP!

[iceberk]

Yes, that’s exactly what I’d like to see 😃 Created pull PR #115