Automatic refactoring of test clones

测试维护占了整个软件开发工作的很大一部分：在我们的基准测试套件中，测试占了195万行代码的21%，这些代码来自不同的领域，包含10个开源Java项目。此外，测试代码通常是重复的，因为它执行类似的功能。因此，测试用例通常由彼此非常相似的方法组成。在过去的工作中，我们发现基准套件中44%的测试与套件中至少一种其他方法相似或克隆。提出的研究将开发测试克隆的分类，并为这些克隆设计和实现新的重构技术。我们将开发一个工具，对测试套件中的测试克隆进行分类和排序，并自动应用所需的测试重构，提高测试套件的可扩展性和可维护性。

Test maintenance accounts for a substantial fraction of total software development effort: Tests account for 21% of the 1.95 million lines of code in our benchmark suite of 10 open-source Java projects drawn from a variety of domains. Moreover, test code is often repetitive, since it exercises similar functionality. Hence, test cases often consist of methods that are quite similar to each other. In past work, we have found that 44% of the tests in our benchmark suite are similar to-or clones of-at least one other method in the suite. The proposed research will develop a classification of test clones and design and implement novel refactoring techniques for such clones. We will develop a tool that classifies and ranks test clones in a test suite and automatically applies desired test refactorings, improving the extensibility and maintainability of test suites.