Generation of all-paths unit test with function calls

摘要

Structural testing is usually restricted to unit tests andbased on some clear definition of source code coverage. Inparticular, the all-paths criterion, which requires at leastone test-case per feasible path of the function under test,is recognised as offering a high level of software reliabil-ity. This paper deals with the difficulties of using structuralunit testing to test functions which call other functions. Tolimit the resulting combinatorial explosion in the number ofpaths, we choose to abstract the called functions by theirspecification. We incorporate the functional information onthe called functions within the structural information on thefunction under test, given as a control flow graph (CFG).This representation combining functional and structural de-scriptions may be viewed as an extension of the classic CFGand allows us to characterise test selection criteria ensur-ing the coverage of the source code of the function undertest. Two new criteria will be proposed. The first criterioncorresponds to the coverage of all the paths of this new rep-resentation, including all the paths arising from the func-tional description of the called functions. The second cri-terion covers all the feasible paths of the function undertest only. We describe how we automate test-data genera-tion with respect to such grey-box (combinations of black-box and white-box) test selection strategies, and we applythe resulting extension of our PathCrawler tool to examplescoded in the C language.