A Run-time Adaptive and Code-size Efficient XML Parser

This paper proposes a new method for designing and implementing a manual XML parser named BNFParser, which is based on the mechanism of XML document matching against a BNF tree that built on XML formal grammar represented by extended Backus-Naur form (EBNF) notation. Compared with generic XML parsers, e.g. top-down Xerces/LibXML parser and bottom-up YACC-assisted parser, BNFParser is designed to be used under the circumstance of storage limitation and memory restriction, e.g. embedded devices. Another advantage of BNFParser is more flexible to be tailored and upgraded through only updating plain text BNF syntax rules instead of reediting and recompiling source code. Furthermore, we investigate the architectures of the generic XML parsers, and present the performance evaluation. Experiment shows that applicable scenario of BNFParser is parsing an XML document with size less than 100 KB