A Better XML Parser through Functional Programming

Abstract

This paper demonstrates how a higher-level, declarative view of XML parsing as folding over XML documents has helped to design and implement a better XML parser. By better we mean a full-featured, algorithmically optimal, pure-functional parser, which can act as a stream processor. By better we mean an efficient SAX parser that is easy to use, a parser that does not burden an application with the maintenance of a global state across several callbacks, a parser that eliminates classes of possible application errors.

This paper describes such better XML parser, SSAX. We demonstrate that SSAX is a better parser by comparing it with several XML parsers written in various (functional) languages, as well as with the reference XML parser Expat. In the experience of the author the declarative approach has greatly helped in the development of SSAX. We argue that the more expressive, reliable and easier to use application interface is the outcome of implementing the parsing engine as an enhanced tree fold combinator, which fully captures the control pattern of the depth- first tree traversal.