This class can be used by itself or as part of a SAX event stream: it takes as input a series of SAX2 ContentHandler events and uses the information in those events to write an XML document. Since this class is a filter, it can also pass the events on down a filter chain for further processing (you can use the XMLWriter to take a snapshot of the current state at any point in a filter chain), and it can be used directly as a ContentHandler for a SAX2 XMLReader.

The client creates a document by invoking the methods for standard SAX2 events, always beginning with the StartDocument() method and ending with the EndDocument() method. There are convenience methods provided so that clients to not have to create empty attribute lists or provide empty strings as parameters; for example, the method invocation

    w.StartElement("foo");

is equivalent to the regular SAX2 ContentHandler method

    w.StartElement("", "foo", "", new Attributes());

Except that it is more efficient because it does not allocate a new empty attribute list each time. The following code will send a simple XML document to standard output:

    XMLWriter w = new XMLWriter();
    w.StartDocument();
    w.StartElement("greeting");
    w.Characters("Hello, world!");
    w.EndElement("greeting");
    w.EndDocument();

The resulting document will look like this:

<?xml version="1.0" standalone="yes"?>
<greeting>Hello, world!</greeting>

In fact, there is an even simpler convenience method, DataElement(String, String), designed for writing elements that contain only character data, so the code to generate the document could be shortened to

    XMLWriter w = new XMLWriter();
    w.StartDocument();
    w.DataElement("greeting", "Hello, world!");
    w.EndDocument();

Whitespace

According to the XML Recommendation, all whitespace in an XML document is potentially significant to an application, so this class never adds newlines or indentation. If you insert three elements in a row, as in

    w.DataElement("item", "1");
    w.DataElement("item", "2");
    w.DataElement("item", "3");

you will end up with

    <item>1</item><item>3</item><item>3</item>

You need to invoke one of the Characters methods explicitly to add newlines or indentation. Alternatively, you can use DataWriter, which is derived from this class -- it is optimized for writing purely data-oriented (or field-oriented) XML, and does automatic linebreaks and indentation (but does not support mixed content properly).

Namespace Support

The writer contains extensive support for XML Namespaces, so that a client application does not have to keep track of prefixes and supply xmlns attributes. By default, the XML writer will generate Namespace declarations in the form _NS1, _NS2, etc., wherever they are needed, as in the following example:

    w.StartDocument();
    w.EmptyElement("http://www.foo.com/ns/", "foo");
    w.EndDocument();

The resulting document will look like this:

    <?xml version="1.0" standalone="yes"?>
    <_NS1:foo xmlns:_NS1="http://www.foo.com/ns/"/>

In many cases, document authors will prefer to choose their own prefixes rather than using the (ugly) default names. The XML writer allows two methods for selecting prefixes:

1. the qualified name
2. the SetPrefix(String, String) method.

Whenever the XML writer finds a new Namespace URI, it checks to see if a qualified (prefixed) name is also available; if so it attempts to use the name's prefix (as long as the prefix is not already in use for another Namespace URI).

Before writing a document, the client can also pre-map a prefix to a Namespace URI with the setPrefix method:

    w.SetPrefix("http://www.foo.com/ns/", "foo");
    w.StartDocument();
    w.EmptyElement("http://www.foo.com/ns/", "foo");
    w.EndDocument();

The resulting document will look like this:

    <?xml version="1.0" standalone="yes"?>
    <foo:foo xmlns:foo="http://www.foo.com/ns/"/>

The default Namespace simply uses an empty string as the prefix:

    w.SetPrefix("http://www.foo.com/ns/", "");
    w.StartDocument();
    w.EmptyElement("http://www.foo.com/ns/", "foo");
    w.EndDocument();

The resulting document will look like this:

    <?xml version="1.0" standalone="yes"?>
    <foo xmlns="http://www.foo.com/ns/"/>

By default, the XML writer will not declare a Namespace until it is actually used. Sometimes, this approach will create a large number of Namespace declarations, as in the following example:

    <xml version="1.0" standalone="yes"?>
    <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
    <rdf:Description about="http://www.foo.com/ids/books/12345">
    <dc:title xmlns:dc="http://www.purl.org/dc/">A Dark Night</dc:title>
    <dc:creator xmlns:dc="http://www.purl.org/dc/">Jane Smith</dc:title>
    <dc:date xmlns:dc="http://www.purl.org/dc/">2000-09-09</dc:title>
    </rdf:Description>
    </rdf:RDF>

The "rdf" prefix is declared only once, because the RDF Namespace is used by the root element and can be inherited by all of its descendants; the "dc" prefix, on the other hand, is declared three times, because no higher element uses the Namespace. To solve this problem, you can instruct the XML writer to predeclare Namespaces on the root element even if they are not used there:

    w.ForceNSDecl("http://www.purl.org/dc/");

Now, the "dc" prefix will be declared on the root element even though it's not needed there, and can be inherited by its descendants:

    <xml version="1.0" standalone="yes"?>
    <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
    xmlns:dc="http://www.purl.org/dc/">
    <rdf:Description about="http://www.foo.com/ids/books/12345">
    <dc:title>A Dark Night</dc:title>
    <dc:creator>Jane Smith</dc:title>
    <dc:date>2000-09-09</dc:title>
    </rdf:Description>
    </rdf:RDF>

This approach is also useful for declaring Namespace prefixes that be used by qualified names appearing in attribute values or character data.