Visualizing textual redundancy in legacy source

As a result of maintenance activity legacy systems contain repeated text in the form of large and small blocks that appear in more or less the same form in several places. These repetitions define a structure that can contribute information about the development history of the source different from the documented version or the current directory structure.A strategy based on fingerprinting is used to obtain raw matches indicating where repetitions occur. The information inherent in these matches is then reorganized for easier processing, leading to a natural clustering of substrings. Suppression of detail is usually necessary to make further progress and can be done in several different ways.For example, matches of blocks of text identify associations within groups of files. In cases with complex clusters of files involving multiple overlapping subsets of files, Hasse diagrams can support visualization. Techniques useful for understanding such graphs can then be employed to provide significant insights into the structure of the redundancy and hence the source.The paper discusses this approach and shows results obtained from an example of reasonable size (40 Mbytes of source based on two releases of the GNU gcc compiler).