This package provides Saxon's preferred Java API for XSLT, XQuery, XPath, and XML Schema processing. The interface is designed to hide as much as possible of the detail of the implementation. However, the API architecture faithfully reflects the internal architecture of the Saxon product, unlike standard APIs such as JAXP and XQJ which in many cases force compromises in the design and performance of the application.

An application starts by loading a {@link net.sf.saxon.s9api.Processor}, which allows configuration options to be set. As far as possible, an application should instantiate a single Processor.

The interfaces for XSLT, XQuery, and XPath processing all follow the same pattern. There is a three-stage execution model: first a compiler is created using a factory method in the Processor object. The compiler holds compile-time options and the static context information. Then the compiler's compile() method is called to create an executable, a representation of the compiled stylesheet, query, or expression. This is thread-safe and immutable once created. To run the query or transformation, first call the load() method to create a run-time object called variously an {@link net.sf.saxon.s9api.XsltTransformer}, {@link net.sf.saxon.s9api.XQueryEvaluator}, or {@link net.sf.saxon.s9api.XPathSelector}. This holds run-time context information such as parameter settings and the initial context node; the object is therefore not shareable and should only be run in a single thread; indeed it should normally only be used once. This object also provides methods allowing the transformation or query to be executed.

In Saxon-EE the Processor owns a {@link net.sf.saxon.s9api.SchemaManager} that holds the cache of schema components and can be used to load new schema components from source schema documents. It can also be used to create a {@link net.sf.saxon.s9api.SchemaValidator}, which in turn is used to validate instance documents.

Source documents can be constructed using a {@link net.sf.saxon.s9api.DocumentBuilder}, which holds all the options and parameters to control document building.

The output of a transformation, or of any other process that generates an XML tree, can be sent to a {@link net.sf.saxon.s9api.Destination}. There are a number of implementations of this interface, including a {@link net.sf.saxon.s9api.Serializer} which translates the XML document tree into lexical XML form.

There are classes to represent the objects of the XDM data model, including {@link net.sf.saxon.s9api.XdmValue}, {@link net.sf.saxon.s9api.XdmItem}, {@link net.sf.saxon.s9api.XdmNode}, and {@link net.sf.saxon.s9api.XdmAtomicValue}. These can be manipulated using methods based on the Java 8 streams processing model: for details see package {@link net.sf.saxon.s9api.streams}.