Program Refactoring in the Presence of Preprocessor Directives

Abstract

The C preprocessor is heavily used in C programs because it provides useful and even necessary additions to the C language. Since preprocessor directives are not part of C, they are removed before parsing and program analysis take place, during the phase called preprocessing. In the context of refactoring, it is inappropriate to remove preprocessor directives: if changes are applied on the preprocessed version of a program, it may not be possible to recover the un-preprocessed version. This means that after refactoring, all the source code would be contained in a single unit, targeted to a single configuration and without preprocessor macros.

This thesis describes a novel approach to preserve preprocessor directives during parsing and program analysis, and integrate them in the program representations. Furthermore, it illustrates how the program representations are used during refactoring and how transformations preserve preprocessor directives.

Additionally, the semantics of the C preprocessor are formally specified, and the results of implementing this approach in a refactoring tool for C, CRefactory, are presented.