Isolation levels act as correctness contracts between real-world applications and database systems. Incorrect implementations of isolation levels would compromise the data consistency and integrity in the database. Therefore, exploring a correctness testing approach for isolation levels is of significant research importance. Specifically, correctness testing approaches for isolation levels involves addressing the following two research questions. A workload generation approach is needed to generate effective workloads for testing isolation levels and inputting them into the tested database system. After that, a test oracle approach is required to verify whether the tested database system processes workloads according to the definition of isolation levels. However, existing studies for testing isolation levels in database systems lack an effective, non-redundant, customized workload generation approach, and they also are unable to accomplish the correctness verification task for isolation levels in a general, black-box, efficient manner. In our work, we conduct in-depth research and propose a simple yet effective framework to address the above two problems. From experimental results, our framework can generate various workloads to test isolation levels, which outperform existing approaches in testing effectiveness, efficiency, and coverage. Moreover, our framework has a superpower to verify various isolation levels for any workload running on black-box database systems, and its verification performance can even surpass the execution performance of the database systems. Practically, we have successfully found 33 bugs that cannot be found by other existing approaches in popular database systems.