Improving mutation testing process of python programs

Abstract

Mutation testing helps in evaluation of test suite quality and test development. It can be directed to programs of different languages. High cost of a mutation testing process limits its applicability. This paper focuses on mutation testing of Python programs, discussing several issues of mutant creation and execution. It was showed how they can be effectively handled in the Python environment. We discuss introduction of first and higher order mutation in an abstract syntax tree with use of generators, dealing with code coverage with AST, executing mutants via mutant injection into tests. The solutions were used in reengineering of MutPy - a mutation testing tool for Python programs. The improvements were positively verified in mutation experiments.