XML Schema Home

XML::Schema

Table of Contents


Synopsis


Description

The XML::Schema module set implements the necessary functionality to construct, represent and utilise XML Schemata in Perl. From a schema, an XML parser can be generated which is dedicated to parsing and validating XML instance documents according to that schema specification. In addition, the modules support a powerful extension mechanism to allow custom processing actions to be scheduled for execution at different times during the parsing of instance documents, e.g. on a particular element, type, attribute, etc. By this simple annotation of a schema with Perl callbacks, it is possible to build extremely flexible, powerful and efficient XML processing and manipulation applications.

An XML Schema is essentially a formal specification of the structure (i.e. elements and attributes) and the data types (i.e. element content and attribute values) of a particular class of XML document. An XML document which conforms to a particular schema is then said to be a valid instance document of that schema.

A Document Type Definition (DTD) is one example of a simple schema specification. The W3C XML Schema specification is another example, boasting a far richer set of features and the additional complexity that goes with it. In fact there are currently more than a dozen different published standards for representing XML schemata, most of which are syntactic and semantic variations of the same theme.

The XML::Schema module set aims to implement a generic framework for parsing XML documents according to a schema definition of some kind. It is hoped that by providing the appropriate hooks through which different schema "backends" can be attached, we can extend the basic framework to build schema representations and parsers compatible with many of the different schema standards.

Much of the architecture is based around the W3C XML Schema specification and conformance with this standard in particular is a desired goal of the project. The simple types supported by these modules are a direct implementation of the W3C XML Schema Datatypes (although some are currently incomplete or not yet implemented). This provides a rich and powerful set of data types which, to the author's best knowledge, is a superset of the data types supported by other schema standards.

The schema structure is also based on that defined in the W3C XML Schema Structures specification although some of the more advanced features are not yet implemented. The modules implement a fairly generic set of objects for representing complex types, elements, attributes and content models. These can be extended through subclassing by implementations offering greater conformance to one or other schema standard that requires a more specific or different implementation.

There is currently no support for reading a schema specification from an XML file ("full conformance" in W3C terms). Schemata must be constructed "manually" by instantiating an XML::Schema object and calling its various methods to define types, elements, and so on. ("minimal conformance" in W3C terms). It is hoped that the modules will eventually implement sufficient conformance with the W3C XML Schema standard to allow the schema for W3C XML Schema documents to be encoded. This will allow a parser to be automatically generated which can read W3C XML Schema documents, validate them, and generate the approriate XML::Schema Perl objects to represent it. Thus, the minimally conformant parser should be able to bootstrap a fully conformant parser.

Once you have a schema representation in terms of Perl objects, whether constructed manually or by an automatic code generator as described above, the parser() method can be called to generate a validating parser (an XML::Parser object) for parsing and validating XML instance documents according to the schema.

In addition to "trivial" XML::Schema validation (we use the term lightly because such validation is anything but trivial), the XML::Schema module set implements a powerful extension mechanism by which the basic capability of the validating parser can be enhanced. We introduce the term "XML Schedule" to describe this feature.

An XML Schedule is a set of production rules which can be associated with an XML Schema. The rules specify what actions should be taken after the parser has identified, parsed and validated particular elements, attributes or character content within an instance document. In implementation terms, the rules are simply Perl callbacks which annotate the schema and are called either immediately before or after a particular item is parsed.

By this process, we can create the appropriate annotations to a schema to perform additional validation or post-processing of certain part(s) of the instance document. A schedule is effectively a "back-end" which sits behind the "front-end" parser. The parser parses an XML document, validates each attribute, element, etc., and then calls the schedule to perform any actions associated with that item. You can use this technique to annotation the schema with your own XML processing actions that perform whatever manipulation of the incoming data that you require. You can convert XML documents to different formats according to your own stylesheet rules or transformations, marshal XML data into Perl objects or database records, generate HTML pages, forms, and so on.

One of the good things about all this is that it happens in "streaming" mode. You don't have to build a big or complicated model in memory and then navigate it to find the bits that you want and throw away the rest. Because you're parsing to a schema which indicates a "known" document structure, you can specify in advance what you want to do with different parts of it.

As well as specifying what you want, you can also chose to selectively ignore certain parts of a document, effectively navigating only certain nodes of interest. It should be possible, for example, to implement an XPath facility (or rather a subset of XPath) which annotes the nodes in the schema required to navigate to the specified XPath and ignores all others.

This is a stable alpha release of these modules. They are complete in as much as they do what they advertise they can do. However, they are incomplete in places due to missing or inadequate documentation, tests, examples, etc. Also be advised that the modules do not yet support some of the more obscure features described in the W3C XML Schema specification.


Package Variables

Perl XML::Schema Documentation