The Design And Realization Of A Hardware-Based XML Parsing Method

XML (Extensible Markup Language) is a simple method that it marks content with markers to transmission information.Due to its extensibility and cross-platform features,XML has been widely used in web server and database system, and it has become a data description and exchange standard. However, relative to the present enterprise server GB data throughput, software-based XML processing methods fail to deliver the required throughput, posing a severe bottleneck to network performance. The need for exploring alternative solutions for parsing XML data has thus assumed prime importance today.Analyzing the theore of the XML parsing and contrasting the four parsing models, this paper proposes a DOM-based XML hardware parsing technique. It includes four main modules:Lexical Analysis module, Well-formed Checking module,DOM Constructor module and Schema Verification module.The input XML characters are marked to identify the element name,the attribute name and text content in the Lexical Analysis.This module uses a FSM to implement. Well-formed checking is mainly to check syntax of the XML document, which including the element name checking and attribute name checking. A stack is used to match the element name characters, and an efficient random storage structure called bloom filter is used to check uniqueness of attribute name, which can rapidly judge whether the current attribute had appeared. DOM constructing module is independent on the Well-formed Checking module and the Schema Verification module, which uses four kinds of node-type to build a DOM tree of the input XML document,including the element name node,the attribute name node, the element content node and the attribute content node. Schema Verification is mainly to match the element and the attribute name with the rules defined in the Schema file. Due to the relatively small volatility of schema files, we first pre-compile the current schema file into a static table which be storaged in a custom local memory to lookup rules..At last, compared with software parsing method,through a large number of experiments, the hardware parsing XML method has a high speedup ratio,.the system throughput can be achieved1.2Gbps. Realizations are tested and simulated.under the FPGA environment.