Schema Parser

Representing the structure of XML documents

One of the challenges when handling XML documents is to represent their structure in a machine-readable form, in a schema. If there is a schema for an XML document it allows to automatically check documents for their validity, but there are also more advanced use cases.

In the context of the kode project XML schemas are used to generate code to represent, read and write XML documents in C++ (see XML Compiler, and to generate user interfaces to handle data available in XML (see KXForms).

As there are different ways to specify XML schemas, kode also contains a schema abstraction. This makes it possible to plug in different schema representations into the tools making use of XML schema information.

The schema representations, which are use by kode are W3C's XML Schema, Relax NG, and example XML.

Schema Parser

W3C XML Schema

kode includes a basic parser for W3C's XML Schema written by Tobias König. It's passed on a C++ port of the wsdlpull parser by Viviek Krishna doen by Michaël Larouche.

The parser is pretty limited. Qt 4.6 will include a XML Schema implementation, which is much more advanced and complete. Once this is released, we should investigate in making use of that instead of the own implementation in kode.

Relax NG

kode also includes a Relax NG parser. It's even more limited than the XML Schema parser. It would need a lot of work to be ready for production use.

Example XML

Often XML data is used, but no formal schema definition exists. There are ad-hoc implementations and documentation simply providing example documents to explain how the XML data is used. While this might be an unsatisfying aproach from a systematic point of view, it's a pragmatic way to quickly implement functionality, and it has proven to be a surprisingly good way to handle XML data.

To accommodate to this situation kode is able to create a XML schema representation from example XML files. This for example make it possible to just take some response data from an XML based REST service and create C++ classes for handling this data. This results in very little overhead for using XML to exchange data.