Compressing computer programs

Abstract

This paper describes a scheme for compressing programs written in a particular programming language—which can be any language that has a formal lexical and syntactic description—in such a way that they can be reproduced exactly. Only syntactically correct programs can be compressed. The scheme is illustrated on the Pascal language, and compression results are given for a corpus of Pascal programs; but it is by no means restricted to Pascal. In fact, we discuss how a "compressor-generator" program can be constructed that creates a compressor automatically from a formal specification of a programming language, in much the same way as a parser generator creates a syntactic parser from a formal language description.