You can also declare default namespaces in the prolog, using the syntax shown in Figure 10-2. There are two different default namespace declarations, one for elements and types, and the other for functions.

Figure 10-2. Syntax of a prolog default namespace declaration

For example:

declare default element namespace "http://datypic.com/cat";

would make http://datypic.com/cat the default namespace for elements and types. This means that anywhere an unprefixed element or type name is used in the query (for example, in element constructors and path expressions), it is assumed to be in that namespace.

This is shown in Example 10-7, which is the same as Example 10-6 except with an added default namespace declaration and a new step catalog in the path expression. The results are the same as those returned by Example 10-6. In this case, because catalog is unprefixed, the processor will look for a catalog element in the default element namespace, http://datypic.com/cat.

declare default element namespace "http://datypic.com/cat";
declare namespace rep = "http://datypic.com/report";
declare namespace prod = "http://datypic.com/prod";
<rep:report> {
  doc("cat_ns.xml")/catalog/prod:product
} </rep:report>

In Example 10-7, it just so happens that the namespace declarations in the query match the namespace declarations in the input document. The namespace http://datypic.com/prod is mapped to the prod prefix, and http://datypic.com/cat is the default, in both places.

However, it is not necessary for the prefixes to match. Example 10-8 shows an equivalent query where a different prefix is used for the namespace http://datypic.com/prod, and http://datypic.com/cat now has a prefix associated with it too. Note that the prefixes declared in the query (prod2 and cat) are used in the path expression, not those from the input document. In fact, it would be illegal to use the prod prefix in this query, because it is not declared in the query itself.

declare namespace rep = "http://datypic.com/report";
declare namespace cat = "http://datypic.com/cat";
declare namespace prod2 = "http://datypic.com/prod";
<rep:report> {
  doc("cat_ns.xml")/cat:catalog/prod2:product
} </rep:report>

Example 10-8 yields the same results as Example 10-7. It is worth noting that the prod prefix (from the input document) is used in the results rather than prod2 (from the query).

As with a regular XML default namespace declaration, a prolog default element namespace declaration does not apply to attribute names, nor does it apply to variable names or function names.

Only one default element namespace declaration may appear in the query prolog. If the string literal is a zero-length string, unprefixed element and type names are considered to be in no namespace.

You can also declare a default namespace for functions, using the same syntax, but with the keyword function. For example:

declare default function namespace "http://datypic.com/funclib";

This means that all unprefixed functions that are called or declared within that query (including type constructors) are assumed to be in that namespace. If you specify a zero-length string, the default will be "no namespace." Only one default function namespace declaration may appear in the query prolog.

If no default function namespace is declared, the default is the XPath Functions Namespace, http://www.w3.org/2005/xpath-functions. In general, it is best not to override this, because if you do, you are required to prefix all calls to the built-in functions such as substring and max.

For convenience, other prolog declarations can bind namespaces to prefixes, namely schema imports, module declarations, and module imports. For example:

import module namespace strings = "http://datypic.com/strings"
                        at "http://datypic.com/strings/lib.xq";

will bind the http://datypic.com/strings namespace to the strings prefix. These types of declarations are covered in Chapters 12 and 13.

The scope of a namespace declaration in an element constructor is the constructor itself (including any other attributes, whether they appear before or after the namespace declaration). Much like regular XML documents, namespace declarations in child element constructors can override the namespace declarations of the outer constructors. However, this is not recommended.

The scope of a prolog namespace declaration is the entire query module. However, prolog namespace declarations can also be overridden, by namespace declarations in an element constructor. For example, if in Example 10-7 you placed a default namespace declaration in the report start tag, it would be in scope until the report end tag (including any attributes of report), overriding the default namespace declaration in the prolog.

Whether namespace declarations appear in the prolog or in a direct element constructor, they have similar rules regarding the names on which they have an effect. Namespace declarations that specify a prefix (i.e., ones that are not default namespace declarations) allow that prefix to be used with any qualified name, namely:

Default element namespace declarations affect only element and type names. They do not affect attribute, variable, or function names. Attribute and variable names, when they appear unprefixed, are considered to be in no namespace, not the default element namespace. Unprefixed function names are in the default function namespace.

<abc xmlns="xyz">{$root/d/e}</abc>

the processor will look for the d and e elements in the xyz namespace. In XSLT:

<abc xmlns="xyz">
  <xsl:copy-of select="$root/d/e"/>
</abc>

the processor will look for the d and e elements in no namespace.

These rules governing the impact of namespace declarations on names are summarized in Table 10-2.

It is important to understand that namespace declarations in a query affect only names that are explicitly specified in that query, not those of any child elements that might be included from the input document. Example 10-10 shows a query that copies some elements from an input document into a newly constructed element. It uses the input document prod_ns.xml shown in Example 10-11. The input document elements (names and name) are in the http://datypic.com/prod namespace, while the constructed report element is in the http://datypic.com/report namespace.

Query
<report xmlns="http://datypic.com/report">
  <firstChild/>
  {doc("prod_ns.xml")/*}
</report>
Results
<report xmlns="http://datypic.com/report">
  <firstChild/>
  <prod:product xmlns:prod="http://datypic.com/prod">
    <prod:number>563</prod:number>
    <prod:name language="en">Floppy Sun Hat</prod:name>
  </prod:product>
</report>

In the results, report and firstChild are in the default namespace specified in the query. However, the product, number, and name elements are still in their original namespace from the input document. They do not become part of the default http://datypic.com/prodreport namespace. This would be true even if they were in no namespace in the input document. Copying an element never changes its namespace.

<prod:product xmlns:prod="http://datypic.com/prod">
  <prod:number>563</prod:number>
  <prod:name language="en">Floppy Sun Hat</prod:name>
</prod:product>