Regression Test Generation Approach Based on Tree-Structured Analysis

Regression test generation is an important process to make sure that changes of program have no unintended side-effects. To achieve full confidence, many projects have to re-run all the test cases for entire program, which makes it a time consuming and expensive activity. In this paper, a code based regression testing approach is proposed to generate selected test suites for unit testing. The framework contains five phases: program change detection phase, logical verification phase, branch pruning phase, test case prioritization phase and test suite generation phase. These five phases can achieve detection of program´s modification, coding standard, test case pruning, test case prioritizing and inputs generation for regression test cases respectively. A prototype based on this framework is implemented using logical tree-structured analysis, and the preliminary experiment shows that proposed approach can provide efficient regression test suites.