Apache Commons logo Commons JXPath

What's JXPath

JXPath provides APIs for traversal of graphs of JavaBeans, DOM and other types of objects using the XPath syntax.

If you are not familiar with the XPath syntax, start with XPath Tutorial by W3Schools.
Also see XML Path Language (XPath) Version 1.0 - that's the official standard.

XPath is the official expression language of XSLT. In XSLT, you mostly use XPath to access various elements of XML documents. You can do that with JXPath as well. In addition, you can read and write properties of JavaBeans, get and set elements of arrays, collections, maps, transparent containers, various context objects in Servlets etc. In other words, JXPath applies the concepts of XPath to alternate object models.

You can also have JXPath create new objects if needed.

The central class in the JXPath architecture is JXPathContext. Most of the APIs discussed in this document have to do with the JXPathContext class.

Object Graph Traversal

JXPath uses JavaBeans introspection to enumerate and access JavaBeans properties.

The interpretation of the XPath syntax in the context of Java object graphs is quite intuitive: the "child" axis of XPath is mapped to JavaBean properties. In fact, the "attribute:" axis is mapped exactly the same way, so the "child::" and "attribute:" axes can be used interchangeably with JavaBeans.

JavaBean Property Access

JXPath can be used to access properties of a JavaBean.

public class Employee {
    public String getFirstName(){

Employee emp = new Employee();

JXPathContext context = JXPathContext.newContext(emp);
String fName = (String)context.getValue("firstName");

In this example, we are using JXPath to access a property of the emp bean. In this simple case the invocation of JXPath is equivalent to invocation of getFirstName() on the bean.

Note that using the XPath "@firstName" instead of "firstName" would produce the same result, because the "child::" and "attribute::" axes are equivalent.

Lenient Mode

The context.getValue(xpath) method throws an exception if the supplied xpath does not map to an existing property. This constraint can be relaxed by calling context.setLenient(true). In the lenient mode the method merely returns null if the path maps to nothing.

Nested Bean Property Access

JXPath can traverse object graphs:

 public class Employee {
    public Address getHomeAddress(){
 public class Address {
    public String getStreetNumber(){

 Employee emp = new Employee();

 JXPathContext context = JXPathContext.newContext(emp);
 String sNumber = (String)context.getValue("homeAddress/streetNumber");

In this case XPath is used to access a property of a nested bean.

A property identified by the XPath does not have to be a "leaf" property. For instance, we can extract the whole Address object in above example:

    Address addr = (Address)context.getValue("homeAddress");

Collection Subscripts

JXPath can extract elements from arrays and collections.

 public class Integers {
    public int[] getNumbers(){

 Integers ints = new Integers();

 JXPathContext context = JXPathContext.newContext(ints);
 Integer thirdInt = (Integer)context.getValue("numbers[3]");

A collection can be an arbitrary array or an instance of java.util.Collection. JXPath also supports indexed properties according to the JavaBeans specification.

Note: in XPath the first element of a collection has index 1, not 0.

Retrieving Multiple Results

JXPath can retrieve multiple objects from a graph. Note that the method called in this case is not getValue, but iterate.

 public class Author {
    public Book[] getBooks(){

 Author auth = new Author();

 JXPathContext context = JXPathContext.newContext(auth);
 Iterator threeBooks = context.iterate("books[position() < 4]");

This returns an iterator over at most three books from the array of all books written by the author.

Map Element Access

JXPath supports maps. To get a value use its key as the name in a child::name construct.

 public class Employee {
    private Map addressMap = new HashMap();
        addressMap.put("home", new Address(...));
        addressMap.put("office", new Address(...));

    public Map getAddresses(){
       return addressMap;

 Employee emp = new Employee();
 JXPathContext context = JXPathContext.newContext(emp);
 String homeZipCode =

Often you will need to use the alternative syntax for accessing Map elements:

 String homeZipCode = (String)context.

Unlike a child name in XPath, the value of the "name" attribute does not have to be a properly formed identifier. Also, in this case the key can be an expression, e.g. a variable.

The attribute "name" can be used not only with Maps, but with JavaBeans as well. The value of this attribute represents the name of a property.

Note: At this point JXPath only supports Maps that use strings for keys.

Note: JXPath supports the extended notion of Map: any object similar to Map, i.e. having some kind of API for accessing values by key, can be handled by JXPath provided that its class is registered with the JXPathIntrospector. The term JXPath uses for such objects is "objects with Dynamic Properties".

DynaBean Access

JXPath supports DynaBeans as well. DynaBeans are treated exactly the same way as JavaBeans.

DOM/JDOM Document Access

JXPath supports access to DOM and JDOM Nodes. The DOM/JDOM node can be the context node of JXPathContext or it can be a value of a property, element of a collection, value of a variable etc. Let's say we have a path "$foo/bar/baz". It will find the desired node if, for instance, the value of the variable "foo" is a JavaBean, whose property "bar" contains a DOM/JDOM Node, which has a child element named "baz".

The intepretation of XPath over DOM/JDOM structures is implemented in accordance with the XPath specification.

Getting a Value vs. Selecting a Node

JXPathContext has two similar sets of APIs: getValue(xpath)/iterate(xpath) and selectSingleNode(xpath)/selectNodes(xpath). With JavaBeans and similar Java object models, these sets of APIs are effectively equivalent. However, with DOM/JDOM there is a difference: selectSingleNode(xpath) and selectNodes(xpath) return Nodes, while getValue() and iterate(xpath) return textual contents of those nodes.

Consider the following XML document:

    <?xml version="1.0" ?>
      <street>Orchard Road</street>

With the same XPath, getValue("/address/street"), will return the string "Orchard Road", while selectSingleNode("/address/street") - an object of type Element (DOM or JDOM, depending on the type of parser used). The returned Element is, of course, <street>Orchard Road</street>.

Registering Namespaces

When using namespaces, it is important to remember that XPath matches qualified names (QNames) based on the namespace URI, not on the prefix. Therefore the XPath "//foo:bar" may not find a node named "foo:bar" if the prefix "foo" in the context of the node and in the execution context of the XPath are mapped to different URIs. Conversely, "//foo:bar" will find the node named "biz:bar", if "foo" in the execution context and "biz" in the node context are mapped to the same URI.

In order to use a namespace prefix with JXPath, that prefix should be known to JXPathContext. JXPathContext knows about namespace prefixes declared on the document element of the context node (the one passed to JXPathContext.newContext(node)), as well as the ones explicitly registered using the JXPathContext.registerNamespace(prefix, namespaceURI) method.


A Container is an object implementing an indirection mechanism transparent to JXPath.

For example, if property "foo" of the context node has a Container as its value, the XPath "foo" will produce the contents of that Container, not the container itself.

An example of a useful container is XMLDocumentContainer. When you create an XMLDocumentContainer, you give it a pointer to an XML file (a URL or a javax.xml.transform.Source). It will read and parse the XML file only when it is accessed. You can create XMLDocumentContainers for various XML documents that may or may not be accessed by XPaths. If they are, they will be automatically read, parsed and traversed. If they are not- they won't be read at all. Of course, once XMLDocumentContainer has read its XML file, it will cache the parse results for a future use.

Let's say we have the the following XML file, which is stored as a Java resource.

    <?xml version="1.0" ?>
      <location id="store101">
          <street>Orchard Road</street>

      <location id="store102">
          <street>Tangerine Drive</street>

Here's the code that makes use of XMLDocumentContainer.

 class Company {
    private Container locations = null;

    public Container getLocations(){
        if (locations == null){
            URL url = getClass().getResource("Vendor.xml");
            locations = new XMLDocumentContainer(url);
        return locations;
 context = JXPathContext.newContext(new Company());
 String street = (String)context.getValue(
                "locations/vendor/location[@id = 'store102']//street");

Like was described before, this code will implicitly open and parse the XML file and find a value in it according to the XPath.

Functions id() and key()

Functions id() and key() can be used with JXPath, however most of the time that requires custom coding.

The only situation where no custom coding is needed is when you want to use the id() function and you have a DOM Node as the context node of the JXPathContext. In this case, JXPath will use the standard behavior of DOM.

In order to evaluate the id() function, JXPath calls a delegate object that should be implemented and installed on the JXPathContext. The object should implement the IdentityManager interface.

Similarly, the key() function relies on a custom implementation of the KeyManager interface.

XPath Axes And Object Graphs

The interpretation of XPath over XML models like DOM and JDOM is governed by the XPath standard. There is no official standard for the interpretation of XPaths on other types of models: beans, maps etc. This part describes how JXPath performs such interpretation.

Parent/child Relationship

In DOM/JDOM the definition of a node's parent is clear: a Node always points to its parent. XML is a strict tree, so there always exactly one parent for every node except the root.

With other models the situation is more complex. An general object model can not be described as a tree. In many cases it is a complicated graph with many paths to the same node and even referential cycles where node A is node B's child, but also node B is node A's child. Even if the graph is a strict tree, a node of that tree may not have a pointer to its parent.

Because of all these issues, JXPath abandons the static notion of a parent/child relationship in favor of a dynamic one. When an XPath is evaluated, the engine performs a series of searches and computations in so called evaluation contexts. For example, when the "/foo/bar" path is evaluated, JXPath first looks for a node named "foo" in the root evaluation context. If such a node is found, the interpreter forms a new context for the discovered node and searches for a node named "bar" in that context.

This chain of contexts is used in JXPath to define the parent-child relationship. Parent is the base node of the previous evaluation context in the chain. A more appropriate name for the "parent::" axis would then be "step back".

Consider this example. The evaluated path is "foo//bar/../baz". In the process of evaluating of this path, the engine will walk the graph forming chains of context like "/foo/a/b/c/bar". Once a node with the name "bar" is found, the engine will "step back": in our case it will go back to the "/foo/a/b/c" context and then look for the node with the name "baz" in that context.

Exercise: think about how the path "//foo[../@name='bar']" would be interpreted.


  1. Descend from the root of the graph looking for a node with the name "foo".
  2. Let's say the engine has found such a node in the context of a node called "biz". The "biz" node is the dynamic parent of the node "foo".
  3. Form a new context for the "foo" node.
  4. To evaluate the predicate "../@name='bar'", step back to the previous context, which is the context of the node "biz" to see if it has an attribute called "name". If so, compare the value of that attribute to "bar". If it is equal, include the current "foo" node in the result set.

The dynamic interpretation of the parent/child relationship affects most axes including "parent::", "ancestor::", "preceding::", "following::" etc.

Document Order

The XPath standard defines the term "document order" as the order in which pieces of XML follow each other in the textual representation. This definition is not applicable directly to non-XML models.

Results of many types of xpaths depend on the document order, so we cannot leave it as "unpredictable" or "undefined" for such nodes as JavaBeans or Maps. In order to have a predictable order, JXPath sorts properties of beans and keys of maps alphabetically.


For JavaBeans and Maps the "attribute::" axis is interpreted the same as the "child::" axis.

The only distinctions are "xml:lang", "xml:space", and "name".

Attribute xml:lang refers to the name of the locale associated with the node. In XML the xml:lang attribute can be specifed for an element explicitly. In non-XML models, the locale is associated with the whole JXPathContext. Unless explicitly set it is the application's default locale.

Since version 1.3, the xml:space attribute can be used in an XML model to direct JXPath's interpretation of embedded whitespace among XML content and nested text. In previous versions this data was trimmed, and this has been preserved as the default behavior for reasons of backward compatibility. Specifying xml:space="preserve" will cause JXPath to preserve whitespace. Keep in mind that it is possible to specify default attribute values using DTD or XML schema, so that there exists a straightforward and standards-based way to enable whitespace preservation by default at the document or element level.

The name attribute is primarily used when working with Maps. Often elements of a Map can be retrieved using the "child::" axis. For example, if "foo" in "foo/bar" refers to a Map then the path extracts from the map the value for the key "bar". The syntax of XPath requires that a child name be a properly formed identifier. Now, what if the key we are looking for is "?%$", which is not an identifier. In this case we can use the "name" attribute like this: "foo[@name='?%$']". This path is not interpreted as "find a 'foo' that has the name '?%$'". It is interpreted as "find a 'foo' and get the value for the key '?%$'" from it. This interpretation is used for maps and beans only. In the case of XML, "name" is treated like any other attribute.

Exceptions During XPath Evaluation

Exceptions thrown by accessor methods are treated differently depending on the evaluated xpath and the particular method used to do the evaluation.

The basic idea is that if JXPath is looking for something by iterating over all properties of a bean and during that iteration an accessor method for one of these properties throws an exception, JXPath ignores the exception and moves on to the next property. This could happen if the method is iterate() or if the path contains search axes like "descendant::", "ancestor::" etc.

In all other cases, an exception thrown by an accessor method is wrapped into a JXPathException and re-thrown.

Modifying Object Graphs

JXPath can also be used to modify parts of object graphs: property values, values for keys in Maps. It can in some cases create intermediate nodes in object graphs.

Setting Properties

JXPath can be used to modify property values.

 public class Employee {
    public Address getAddress() {

    public void setAddress(Address address) {

 Employee emp = new Employee();
 Address addr = new Address();

 JXPathContext context = JXPathContext.newContext(emp);
 context.setValue("address", addr);
 context.setValue("address/zipCode", "90190");

Creating Objects

JXPath can be used to create new objects. First, create a subclass of AbstractFactory and install it on the JXPathContext. Then call jxPathContext.createPath(xpath). JXPathContext will invoke your AbstractFactory when it discovers that an intermediate node of the path is null. It will not override existing nodes.

 public class AddressFactory extends AbstractFactory {
    public boolean createObject(JXPathContext context, Pointer pointer,
                                Object parent, String name, int index){
     if ((parent instanceof Employee) && name.equals("address"){
       ((Employee)parent).setAddress(new Address());
       return true;
     return false;

 JXPathContext context = JXPathContext.newContext(emp);
 context.setFactory(new AddressFactory());

You can also combine creating a path with setting the value of the leaf: the createPathAndSetValue(path, value) method is used for that.

 context.createPathAndSetValue("address/zipCode", "90190");

Note that it only makes sense to use the automatic creation of nodes with very simple paths. In fact, JXPath will not attempt to create intermediate nodes for paths that don't follow these three rules:

  • The only axes used are "child::" and "attribute::", e.g. "foo/bar/@baz"
  • The only two types of predicates used are context-independent indexing and the "[@name = expr]" construct, e.g. "map[@name='key1'][4/2]".
  • If a variable is used, it is the root of the path, e.g. "$object/child".


JXPath supports the notion of variables. The XPath syntax for accessing variables is "$varName".

 public class Author {
    public Book[] getBooks(){

 Author auth = new Author();

 JXPathContext context = JXPathContext.newContext(auth);
 context.getVariables().declareVariable("index", new Integer(2));

 Book secondBook = (Book)context.getValue("books[$index]");

You can also set variables using JXPath:

 context.setValue("$index", new Integer(3));

Note: generally speaking, you can only change the value of an existing variable this way, you cannot define a new variable. If you do want to be able to define a new variable dynamically, implement a defineVariable() method on your custom AbstractFactory and call createPathAndSetValue() rather than setValue(). The restrictions described in the "Creating Objects" section still apply.

When a variable contains a JavaBean or a collection, you can traverse the bean or collection as well:

 context.getVariables().declareVariable("book", myBook);
 String title = (String)context.getValue("$book/title);

 Book array[] = new Book[]{...};

 context.getVariables().declareVariable("books", array);

 String title = (String)context.getValue("$books[2]/title);

Custom Variable Pools

By default, JXPathContext creates a HashMap of variables. However, you can substitute a custom implementation of the Variables interface to make JXPath work with an alternative source of variables. For example, you can define implementations of Variables that cover a servlet context, HTTP request or any similar structure.

See the org.apache.commons.jxpath.servlet package for an example of just that.

Servlet Contexts

The org.apache.commons.jxpath.servlet package contains classes that make it easy to use XPath to access values in various sevlet contexts: "page" (for JSPs), "request", "session" and "application".

See static methods of the class JXPathServletContexts. They allocate various servlet-related JXPathContexts.

JSP Page Context

The JXPathContext returned by getPageContext(PageContext pageContext) provides access to all scopes via the PageContext.findAttribute() method. Thus, an expression like "foo" will first look for the attribute named "foo" in the "page" context, then the "request" context, then the "session" one and finally in the "application" context.

If you need to limit the attibute lookup to just one scope, you can use the pre-definded variables "page", "request", "session" and "application". For example, the expression "$session/foo" extracts the value of the session attribute named "foo".

Servlet Request Context

The getRequestContext(ServletRequest request, ServletContext servletContext) method will give you a context that checks the request scope first, then (if there is a session) the session context, then the application context.

HttpSession Context

The getSessionContext(HttpSession session, ServletContext servletContext) method will give you a context that checks the session context, then the application context.

ServletContext Context

Finally, getApplicationContext(ServletContext servletContext) method will give you a context that checks the application context.

All these methods cache the JXPathContexts they create within the corresponding scopes. Subsequent calls use the JXPathContexts created earlier.


Often, rather than getting a node in the object graph, you need to find out where in the graph that node is. In such situations you will need to employ Pointers. A Pointer is an object that represents the specific location in the object graph. Effectively, it is a simple XPath leading from the context root to the selected node. That simple XPath can be used to repeatedly acquire the same node of the graph without performing a costly search. Let's say, you invoke the JXPath search process by calling the getPointer() method:

Pointer ptr = context.getPointer("//address[zipCode='90190']")

This code will find the address with zipCode = 90190 and return a Pointer describing that node's location. The printed line will look something like this: /vendor[2]/location[1]/address[3]. It provides an unambiguous description of the node's location in the object graph and a fast XPath leading directly to that node.

Here's another example:

Pointer ptr = context.getPointer("employees[$i]/addresses[$j]")

Let's say, at the time of execution the value of the variable i is 1 and j = 3. If we call ptr.asPath(), it returns a simple XPath: "/employees[1]/addresses[3]"; this path does not have a dependency on the variables, it will remain the same when the variables change.

If you need to perform an exhaustive search for all nodes in the graph matching a certain XPath, you can get JXPath to produce an iterator returning pointers for all of discovered locations:

Iterator homeAddresses = context.iteratePointers("//employee/address[@name='home']");

Each Pointer returned by the iterator will represent a home address object in the graph.

It is a good idea to use pointers whenever you need to access the same node of a graph repeatedly.

JXPath is optimized to interpret XPaths produced by Pointers much faster than many other types of XPaths.

Relative Contexts

If you need to evaluate multiple paths relative to a certain node in the object graph, you might want to create a relative JXPathContext.

First, obtain the pointer for the location that is supposed to be the root the relative context. Then obtain the relative context by calling context.getRelativeContext(pointer).

 JXPathContext context = JXPathContext.newContext(bean);

 Pointer addressPtr = context.getPointer("/employees[1]/addresses[2]");

 JXPathContext relativeContext = 

 // Evaluate relative path 
 String zipCode = (String)relativeContext.getValue("zipCode");

 // Evaluate absolute path
 String name = (String)relativeContext.getValue("/employees[1]/name");

 // Use the parent axis to locate the employee for the current address
 Double salary = (Double)relativeContext.getValue("../salary");


Extension Functions

JXPath supports standard XPath functions right out of the box. It also supports "standard" extension functions, which are basically a bridge to Java, as well as entirely custom extension functions.

Standard Extension Functions

Using the standard extension functions, you can call methods on objects, static methods on classes and create objects using any constructors. All class names should be fully qualified.

Here's how you can create new objects:

 Book book = (Book)context.
   getValue("com.myco.books.Book.new('John Updike')");

Here's how you can call static methods:

 Book book = (Book)context.
   getValue("com.myco.books.Book.getBestBook('John Updike')");

Here's how you can call regular methods:

 String firstName = (String)context.

As you can see, the target of the method is specified as the first parameter of the function.

Custom Extension Functions

Collections of custom extension functions can be implemented as Functions objects or as Java classes, whose methods become extension functions.

Let's say the following class implements various formatting operations:

 public class Formats {
    public static String date(Date d, String pattern){
        return new SimpleDateFormat(pattern).format(d);

We can register this class with a JXPathContext:

 context.setFunctions(new ClassFunctions(Formats.class, "format"));

 context.getVariables().declareVariable("today", new Date());
 String today =
     (String)context.getValue("format:date($today, 'MM/dd/yyyy')");

You can also register whole packages of Java classes using PackageFunctions.

Also, see FunctionLibrary, which is a class that allows you to register multiple sets of extension functions with the same JXPathContext.

Expression Context

A custom function can get access to the context in which it is being evaluated. ClassFunctions and PackageFunctions have special support for methods and constructors that have ExpressionContext as the first argument. When such an extension function is invoked, it is passed an object that implements the ExpressionContext interface. The function can then gain access to the "current" object in the currently evaluated context.

public class MyExtensionFunctions {
   public static boolean isDate(ExpressionContext context){
      Pointer pointer = context.getContextNodePointer();
      if (pointer == null){
        return false;
      return pointer.getValue() instanceof Date;

You can then register this extension function using ClassFunctions and call it like this:


This expression will find all nodes of the graph that are dates.

Collections as Arguments

There are two ways a collection can be passed to an extension function: as a NodeSet or as a Collection proper. If the argument type is declared as NodeSet, JXPath will pass a NodeSet object, otherwise it will take values out of the node set and pass those to the function as a regular collection. NodeSet, in addition to providing access to the values, also provides access to pointers. Note that a collection is often passed to an extension function by value and cannot be modified.

public class MyExtensionFunctions {
   public static boolean contains(NodeSet nodeSet, Object value){
      Iterator iter = nodeSet.getPointers().iterator();
      while (iter.hasNext()) {
          Pointer item = (Pointer)iter.next();
          if (item.getValue().equals(value)){
            return true;
      return false;
   // Alternative implementation
   public static boolean contains(List list, Object value){
      Iterator iter = list.iterator();
      while (iter.hasNext()) {
          Object item = iter.next();
          if (item.getValue().equals(value)){
            return true;
      return false;

You can call this function to find all people who have a certain phone number:

  "/addressBook/contact[myext:contains(phoneNumbers, '555-5555']"

In JXPath version 1.1, a function argument declared as Object would be passed as a NodeSet. In version 1.2, this behavior was changed such that a declared argument type of Object triggers the conversion of the NodeSet to its list of values. The simplest way to avoid this conversion, thereby receiving the untouched NodeSet as the function argument, is to declare the argument as a NodeSet. For such times as this may prove impractical, however, a version 1.1-compatible TypeConverter implementation, (org.apache.commons.jxpath.util.JXPath11CompatibleTypeConverter), has been provided in version 1.3. To enable this:

TypeUtils.setTypeConverter(new JXPath11CompatibleTypeConverter());

Collection as the Return Value

A custom function can return a collection of arbitrary objects or a NodeSet. The simple implementation of NodeSet, BasicNodeSet, may come in handy.

Type Conversions

JXPath automatically performs the following type conversions:

From type To type Operation
null primitive false, zero
null string ""
any non-null object String Calls toString()
Boolean any Number True = 1, false = 0
any Number any other Number Truncates if needed
String any primitive type Parses the string
array array Creates a new array of the same size and converts every element
array Collection Creates a collection and adds to it all elements of the array. Note that it will only know how to create the collection if the type is a concrete class, List or Set
Collection array Creates a new array the same size as the collection, converts and copies every element of the collection into the array.
Collection Collection Creates a collection and copies the source collection into the new collection. Note that it will only know how to create the collection if the type is a concrete class, List or Set
non-empty array any Takes the first element of the array
and (recursively) converts it to the needed type
non-empty collection any Takes the first element of the array
and (recursively) converts it to the needed type
NodeSet any Extracts a list of values from the NodeSet and (recursively) converts the list to the needed type.


For DOM Documents JXPathContext supports internationalization XPath-style. A locale can be declared on an XML Element like this:

     <book xml:lang="fr">Les Miserables</book>

You can then use the lang function in XPath to find nodes for a specific language:


The "lang" boolean function is supported for non-DOM objects as well. It tests the Locale set on the JXPathContext (or the default locale). See JXPathContext.setLocale().

You can also utilize the xml:lang attribute, whose value is the name of the locale, whether in a DOM document or outside.

Nested Contexts

If you need to use the same configuration (variables, functions, abstract factories, locale, leniency etc.) while interpreting XPaths with different beans, it makes sense to put the configuration in a separate context and specify that context as a parent context every time you allocate a new JXPathContext for a JavaBean. This way you don't need to waste time fully configuring every context.

 JXPathContext sharedContext = JXPathContext.newContext(null);
 sharedContext.getVariables().declareVariable("title", "Java");
 sharedContext.setFunctions(new MyExtensionFunctions());
 sharedContext.setFactory(new MyFactory());


 JXPathContext context = JXPathContext.newContext(sharedContext, auth);

 Iterator javaBooks =
      context.iterate("books[preprocessTitle(title) = $title]");

Compiled Expressions

When JXPath is asked to evaluate an expression for the first time, it compiles it and caches its compiled representation. This mechanism reduces the overhead caused by compilation. However, in some cases JXPath's own caching may not be sufficient- JXPath caches have limited size and they are automatically cleared once in a while.

Here's how you can precompile an XPath expression:

     CompiledExpression expr = context.compile(xpath);
     Object value = expr.getValue(context);

Use compiled expressions if you need to satisfy any of the following requirements:

  • There is a relatively small number of XPaths your application works with, and it needs to evaluate those XPaths multiple times.
  • Some XPaths need to be precompiled at initialization time for speed.
  • The syntax of some XPaths needs to be checked before they are used for the first time.

Customizing JXPath

JXPath can be customized on several levels.

  • You can provide custom JXPathBeanInfo objects to customize lists of properties of JavaBeans available to JXPath.
  • You can easily add support for object types similar to Map. All you need to do is implement the DynamicPropertyHandler interface and register the implementation with JXPathIntrospector.
  • You can add support for types of object models JXPath does not support out of the box. An example of such model would be an alternative implementation of XML parse tree (e.g. DOM4J etc). You will need to implement two or three classes to allow JXPath to traverse properties of these custom objects.
  • The most dramatic customization of JXPath can be done at the level of JXPathContextFactory- you can transparently provide an alternative implementation of all top level APIs.

Custom JXPathBeanInfo

JXPath uses JavaBeans introspection to discover properties of JavaBeans. You can provide alternative property lists by supplying custom JXPathBeanInfo classes (see JXPathBeanInfo).

Custom DynamicPropertyHandler

JXPath uses various implementations of the DynamicPropertyHandler interface to access properties of objects similar to Map.

The org.apache.commons.jxpath.servlet package has several examples of custom DynamicPropertyHandlers.

Custom Pointers and Iterators

Architecturally, multiple model support is made possible by the notions of a NodePointer and NodeIterator, which are simple abstract classes that are extended in different ways to traverse graphs of objects of different kinds. The NodePointer/NodeIterator APIs are designed with models like JavaBeans in mind. They directly support indexed collections. As a result, XPaths like "foo[10]" can be executed as "getFoo(9)" or "getFoo()[9]", or "getFoo().get(9)", depending on the type of collection. This flexibility is disguised well enough by the APIs of the abstract classes, so we can still have a natural implementation of traversal of object models, such as DOM, that do not have the same notion of collection.

To add support for a new object model, build custom implementations of NodePointer and NodeIterator as well as NodePointerFactory. Then register the new factory with JXPathContextReferenceImpl.

See existing NodePointerFactories for examples of how that's done:

  • BeanPointerFactory works with JavaBeans
  • DynamicPointerFactory works with Dynamic beans like Map, HttpRequest and such
  • ContainerPointerFactory works with Container objects like XMLDocumentContainer
  • DOMPointerFactory works with DOM Nodes

Alternative JXPath Implementation

The core JXPath class, JXPathContext, allows for alternative implementations. This is why instead of allocating JXPathContext directly, you should call a static newContext method. This method will utilize the JXPathContextFactory API to locate a suitable implementation of JXPath. JXPath comes bundled with a default implementation called Reference Implementation.