Automatic test case generation for multiple coverage criteria: towards more efficient bug localization

Software testing plays a vital role in ensuring the quality and reliability of software systems. The quality of test cases is particularly significant. However, manual test case creation requires substantial resources and may not consistently ensure high quality. Therefore, this paper proposes a test case generation method named MCTG (Multiple Coverage Test Case Generator), which enhances upon the symbolic execution tool KLEE. MCTG searches paths and solves constraint conditions within tested functions to generate test cases that meet various coverage criteria. Using a subset of programs from the Codenet dataset, this study recruited 27 experienced testers to manually write test cases as a baseline. Experimental results show that the test case suite generated by MCTG achieves 3.1% higher coverage than KLEE on average and 11.2% higher than manually written cases. Moreover, MCTG’s EXAM value is 20.6% lower than manual methods and 18.0% lower than KLEE, indicating that MCTG places bug-containing statements closer to the top, requiring inspection of less code to find bugs. This suggests that MCTG’s multi-coverage test case generation technique can improve bug localization effectiveness as proposed in this study.