Love Johansson, Johan Larsson:
A Framework for Extracting Information From a Code Base,
summary, report, February 2006.

Abstract:
This thesis examines the need for an information extraction tool which can aid developers in finding information in a large code base, and for controlling architectural principles and guidelines.

The thesis has been written at Ericsson AB in Lund, and our tool has been tailored for the mobile platform software. The platform is written entirely in C, and contains a large number of build variants. This poses a series of problems as the information must be extracted without any form of pre-processing, macro expansions or modification, so that our tool functions on real production code.

Other problems not related to the structure of the code, are issues regarding the implementation of our tool prototype. What kind of information is important? Is it possible to extract the desired information? How should we store the extracted information? How do we use this stored data to solve real problems?

We have designed a tool framework prototype, using our conclusions from the research on both code related and non-code related problems. The prototype for the framework which we have developed handles parsing of the C code, database facilities for storing parsed data, and a client- side application for producing relevant output from the database.