Perceive core logical blocks of a C program automatically for source code transformations

Abstract

Tool like Flawfinder, used to identify a security flaw in a source code, is too expensive to be procured for usage but it can also be accessed on public cloud as a software as a service. Since there is possibility of inside attackers in cloud service, to unveil the logical possessions out of the source code, there is a need to transform the source code by altering the semantics. In this paper, we have introduced a novel method to identify the core logical blocks of any C source code. It mainly consists of two steps: (i) source code entity identification (ii) dependency identification. The entities are identified based on programming language constructs like variables, simple entities and control structures. Variable dependency is in deep analyzed by generating a dependency graph using Neo4j graph database software. This graph is further traversed and weighted matrix of the variable dependencies is created from which the core logical blocks could be identified. Algorithms are designed for the above two steps. Cyclomatic complexity analysis and Time complexity analysis are carried out and experimentations are conducted to verify the same.