Compression via guided parsing

Summary form only given. The reduction in storage size achieved by compressing a file translates directly into a reduction in transmission time when communicating the file. An increasingly common form of transmitted data is a computer program description. This paper examines the compression of source code, the high-level language representation of a program, using the language's context free grammar. We call the general technique guided parsing since it is a compression scheme based on predicting the behavior of a parser when it parses the source code and guiding its behavior by encoding its next action based on this prediction. In this paper, we describe the implementation and results of two very different forms of guided parsing. One is based on bottom-up parsing while the other is a top-down approach.