Identification and extraction of objects from legacy code

This paper presents an effort to utilize the benefits of recent innovations of software engineering in legacy code. Because most systems currently in use are, on the average, over 10 years old and suffering from extensive maintenance, and because it is impractical to simply discard a working system, these concerns are addressed in the context of reverse engineering. A methodology is presented which identifies candidate objects in imperative code, specifically, FORTRAN-77. Candidate objects are identified from several aspects of the system including subprogram invocations, common block definitions and array definitions. To assist the object identification algorithms, a representation of the original imperative code is introduced which uses the concepts of data flow analysis to represent variable usage information