Static Slicing for Python First-Class Objects

Abstract

Program slicing is an important program analysis technique and now has been used in many fields of software engineering. However, most existing program slicing methods focus on static programming languages such as C/C++ and Java, and methods on dynamic languages like Python are rarely seen. Python, a typical dynamic object-oriented language, has been more and more widely used now. In Python, everything is a first-class object, including functions, classes, methods, and modules. Existing slicing methods cannot handle the issue of these first-class objects. Therefore, this paper proposes a static slicing method for Python first-class objects. By adding all the definitions of first-class objects into the dependence model and uniformly constructing the program dependence graphs for all the functions, classes, methods, and modules, this method can effectively solve the slicing problems caused by arbitrary definitions and uses of first-class objects in Python.