Introduction to Filtering with XPath Predicates

An XPath location path consists of one or more location steps, delimited by slashes. For example:

/child::book/child::chapter/attribute::date_written

is an XPath location path which returns a node-set consisting of all date_written attributes of those <chapter> elements which themselves are children of the root <book> element.

Note   This example could also be expressed using XPath's abbreviated syntax as:

/book/chapter/@date_written

Remaining examples in this topic will use the abbreviated syntax. For more information, see Using Abbreviations in XPath Expressions.

In general, the purpose of a location path is to return a set of one or more nodes—a node-set—meeting specified criteria. Each location step in the location path consists of an axis (explicit or defaulted to child::) relative to the context node, a node test, and an optional predicate. (The above example does not include a predicate in any of the three location steps.) All nodes which meet all criteria specified by the full location path are returned by the expression.

The general format of a predicate in a location step is:

[condition]

where the square brackets, [ and ], are required, and condition represents some test typically resulting in a Boolean true or false value. The condition usually takes the form:

value operator value

If a particular condition itself may be represented as a Boolean true or false value, the predicate can take a simpler form:

[value]

Note   This simplified form of the predicate is very important in limiting a location path only to nodes which have other nodes along a particular axis. If there are any nodes of the given name or type along the indicated axis, the result is true; if there are no such nodes, the result is false. Thus:

chapter[descendant::figure]

selects <chapter> elements which are children of the context node only if they have any descendant <figure> elements.

As an example, the set of nodes returned by the location path at the beginning of this topic could be further limited as follows:

/book/chapter[position()>1]/@date_written[.="2001-05-12"]

This location path can be read, from left to right:

1. Locate the root <book> element (/book).
2. Locate all <chapter> elements but the first, in document order, which are children of the root <book> element located by Step 1 (chapter[position()>1]).

   Note   A predicate, like everything else in an XML document, must adhere to the XML rules for well-formed documents. Therefore, operators such as > (in this example), <, and &, which have special meaning to XML parsers, must be escaped using entity references such as >, <, and & respectively.

3. For the <chapter> elements selected by Step 2, locate all of their date_written attributes whose values equal the string "2001-05-12" (@date_written[.="2001-05-12"]).

The result of evaluating this location path is a node-set consisting of the date_written attributes (if any) meeting all these criteria. If no date_written attributes meet the criteria, the result is an empty node-set.

White space between the value and operator terms in a predicate is not significant. Thus, the following two predicates produce the same result:

[position()   =   1]
[position()=1]

See Also

Use XPath Axes to Navigate through XML Data | Determining the Context Node | Constructing the Node-Test Portion of a Location Step