Analysis and manipulation of distributed multi-language software code

Abstract

The authors propose a formal model and a platform to deal with distributed multi-language software analysis. These provide a graph representation of the software codes (source codes and byte-codes), a change propagation process based on graph rewriting, and an automatic profiling tool to measure the contribution of any component to the global performance of the software. The program codes are structured by a multi-graph in which the nodes represent the software components linked by edges representing the meaningful relationships. The software components and their relationships are extracted from the byte-code files, using the mocha decompiler tool, and from the source codes files, using the Javacc tool. Javacc allows one to generate parsers, based on grammar specification files, which include features to produce an XML (eXtensible Markup Language) representation of the software components. Furthermore, a graph of the software components is constructed on the top of the XML files, providing program analysis. This is implemented by an integrated platform including the mocha decompiler, a multi-language parsing tool, a software change management module, and a profiling tool.