Constraint-Based Automated Generation of Test Data

摘要

We present a novel method for automatically generating artificial test data that are particularly suited for testing form-centric software applications with several thousand input fields. The complex validation rules for user input are translated to a constraint satisfaction problem (CSP), which is solved using an off-the-shelf SMT-solver. In order to exert pressure onto the software under test, the generated test data have to incorporate extreme and special values (ESVs) for each field. The SMT-solver is aided by a sophisticated graph-based cluster algorithm and by other heuristic methods in order to reduce the complexity of the CSPs. With further optimizations, the test data generator now routinely generates a complete set of test data records for large form-centric applications within less than two hours. The test data generator described here is operationally being used for automated tests of form-centric Web-applications, within an iterative development process emphasizing very early testing of software applications.