Constraining content: specification and processing

Abstract

SGML and XML are changing the way people think an act with documents. These standards have given rise to a new methodology and document processing model.

Independence between structure and format, structural validation, document longevity, software and hardware independence, all these and much more became possible.

However there is still a lack for content validation which relies beyond DTD scope.

Recently, some proposals were submitted to the W3C, attempting to solve the problem: DCD, Schema, ... XML Schema seems to be the final choice, and some implementations are already emerging.

However, in our opinion, we do not need any of those. We already have what we need we just must to give it a different use.

In order to format a document we need to select fragments and apply style to them or if we want to query the document content we need a combination of tree-walkers to select the wanted nodes in the document tree. We have XSLT to do that. The pattern language inside XSLT is powerful enough to select almost any kind of context inside a document tree.

In this paper we are stating that we can express content constraints using XSLT. The syntax and the processing model are there we only need to tweak a little bit the semantics.

We will present some examples of constraints small enough to fit in this document and we will express them using XSLT in a way that XSLT processors can act as constraint validators.

After that we will describe a simple operational architecture that enables users to specify constraints and test them inside a XML-XSL framework.

At the end we generalize the idea specifying a method to implement the constraint processing model with any tool supporting XSLT patterns.