net.sf.saxon.xpath.XPathExpressionImpl Maven / Gradle / Ivy
Show all versions of saxon-he Show documentation
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
// Copyright (c) 2013 Saxonica Limited.
// This Source Code Form is subject to the terms of the Mozilla Public License, v. 2.0.
// If a copy of the MPL was not distributed with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
// This Source Code Form is "Incompatible With Secondary Licenses", as defined by the Mozilla Public License, v. 2.0.
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
package net.sf.saxon.xpath;
import net.sf.saxon.Configuration;
import net.sf.saxon.expr.*;
import net.sf.saxon.expr.instruct.Executable;
import net.sf.saxon.expr.instruct.SlotManager;
import net.sf.saxon.expr.sort.AtomicComparer;
import net.sf.saxon.expr.sort.SortKeyDefinition;
import net.sf.saxon.expr.sort.SortKeyEvaluator;
import net.sf.saxon.expr.sort.SortedIterator;
import net.sf.saxon.functions.NumberFn;
import net.sf.saxon.om.*;
import net.sf.saxon.pattern.NodeTest;
import net.sf.saxon.trans.XPathException;
import net.sf.saxon.tree.wrapper.VirtualNode;
import net.sf.saxon.value.*;
import org.xml.sax.InputSource;
import javax.xml.namespace.QName;
import javax.xml.transform.sax.SAXSource;
import javax.xml.xpath.XPathConstants;
import javax.xml.xpath.XPathExpression;
import javax.xml.xpath.XPathExpressionException;
import java.util.List;
/**
* The JAXP XPathExpression interface represents a compiled XPath expression that can be repeatedly
* evaluated. This class is Saxon's implementation of that interface.
*
* The class also includes some methods retained from Saxon's original XPath API. When these methods
* are used, the object contains the context node and other state, so it is not thread-safe.
*
* @author Michael H. Kay
*/
public class XPathExpressionImpl implements XPathExpression, SortKeyEvaluator {
private Configuration config;
private Executable executable;
private Expression expression;
private Expression atomizer;
/*@Nullable*/ private NodeInfo contextNode;
private SlotManager stackFrameMap;
/*@Nullable*/ private XPathExpressionImpl sortKey = null;
/**
* The constructor is protected, to ensure that instances can only be
* created using the createExpression() method of XPathEvaluator
* @param exp the compiled expression
* @param exec the executable
*/
protected XPathExpressionImpl(Expression exp, /*@NotNull*/ Executable exec) {
expression = exp;
executable = exec;
config = exec.getConfiguration();
}
/**
* Define the number of slots needed for local variables within the expression.
* This method is for internal use only.
* @param map description of the stack frame
*/
protected void setStackFrameMap(SlotManager map) {
stackFrameMap = map;
}
/**
* Get the stack frame map. This holds information about the allocation of slots to variables.
* This is needed by applications using low-level interfaces for evaluating the expression
* @return a description of the stack frame
*/
public SlotManager getStackFrameMap() {
return stackFrameMap;
}
/**
* Get the Configuration under which this XPath expression was compiled
* @return the Saxon configuration
*/
public Configuration getConfiguration() {
return config;
}
/**
* Define the sort order for the results of the expression. If this method is called, then
* the list returned by a subsequent call on the evaluate() method will first be sorted.
* @param sortKey an XPathExpression, which will be applied to each item in the sequence;
* the result of this expression determines the ordering of the list returned by the evaluate()
* method. The sortKey can be null, to clear a previous sort key. Note that the expression is
* not automatically atomized; if it selects nodes, these should be explicitly converted to
* atomic values by calling the string() or data() functions.
* @deprecated since 9.0. This method is not present in the JAXP interface. The recommended
* way to get a sorted result is to use XQuery instead of XPath.
*/
public void setSortKey(XPathExpressionImpl sortKey) {
this.sortKey = sortKey;
}
/**
* Set the context node for evaluating the expression. If this method is not called,
* the context node will be the root of the document to which the prepared expression is
* bound.
* @param node the context node
* @deprecated since 9.0. Using this method is not thread-safe. Use a method instead
* such as {@link #evaluate(Object, QName)} that allows the context node to be specified
* as a parameter to the call.
*/
public void setContextNode(/*@Nullable*/ NodeInfo node) {
if (node==null) {
throw new NullPointerException("Context node cannot be null");
}
if (node.getConfiguration() != config) {
throw new IllegalArgumentException("Supplied node uses the wrong Configuration");
}
contextNode = node;
}
/**
* Protected, undeprecated version of setContextNode() for use by deprecated paths within the package
* (exists to avoid deprecation warnings when compiling Saxon)
* @param node the context node
*/
protected void privatelySetContextNode(/*@Nullable*/ NodeInfo node) {
if (node==null) {
throw new NullPointerException("Context node cannot be null");
}
if (node.getConfiguration() != config) {
throw new IllegalArgumentException("Supplied node uses the wrong Configuration");
}
contextNode = node;
}
/**
* Execute a prepared XPath expression, returning the results as a List. The context
* node must have been set previously using {@link #setContextNode(net.sf.saxon.om.NodeInfo)}.
* @return The results of the expression, as a List. The List represents the sequence
* of items returned by the expression. Each item in the list will either be an instance
* of net.sf.saxon.om.NodeInfo, representing a node, or a Java object representing an atomic value.
* For the types of Java object that may be returned, see {@link #evaluate(Object, javax.xml.namespace.QName)}
* with the second argument set to NODESET.
* @deprecated since 9.0. This method is not present in the JAXP interface. Either use
* the JAXP methods such as {@link #evaluate(Object, QName)}, or use the Saxon XPath
* API instead of JAXP.
*/
/*@NotNull*/ public List evaluate() throws XPathException {
XPathContextMajor context = new XPathContextMajor(contextNode, executable);
context.openStackFrame(stackFrameMap);
SequenceIterator iter = expression.iterate(context);
SequenceExtent extent = new SequenceExtent(iter);
return (List)PJConverter.ToCollection.INSTANCE.convert(extent, List.class, context);
}
/**
* Execute a prepared XPath expression, returning the first item in the result.
* This is useful where it is known that the expression will only return
* a singleton value (for example, a single node, or a boolean). The context node
* must be set previously using {@link #setContextNode(net.sf.saxon.om.NodeInfo)}.
* @return The first item in the sequence returned by the expression. If the expression
* returns an empty sequence, this method returns null. Otherwise, it returns the first
* item in the result sequence, represented as a Java object using the same mapping as for
* the evaluate() method
* @deprecated since 9.0. This method is not present in the JAXP interface. Either use
* the JAXP methods such as {@link #evaluate(Object, QName)}, or use the Saxon XPath
* API instead of JAXP.
*/
/*@Nullable*/ public Object evaluateSingle() throws XPathException {
XPathContextMajor context = new XPathContextMajor(contextNode, executable);
context.openStackFrame(stackFrameMap);
SequenceIterator iterator = expression.iterate(context);
Item item = iterator.next();
if (item == null) {
return null;
} else {
return SequenceTool.convertToJava(item);
}
}
/**
* Get a raw iterator over the results of the expression. This returns results without
* any conversion of the returned items to "native" Java classes. This method is intended
* for use by applications that need to process the results of the expression using
* internal Saxon interfaces.
* @param contextItem the context item for evaluating the expression
* @return an iterator over the results of the expression, with no conversion of returned items
* @since 9.0
*/
/*@Nullable*/ public SequenceIterator rawIterator(Item contextItem) throws XPathException {
XPathContextMajor context = new XPathContextMajor(contextItem, executable);
return rawIterator(context);
}
/*@Nullable*/ private SequenceIterator rawIterator(/*@NotNull*/ XPathContextMajor context) throws XPathException {
context.openStackFrame(stackFrameMap);
SequenceIterator iterator = expression.iterate(context);
if (sortKey != null) {
Expression key = sortKey.expression;
if (key.getItemType(config.getTypeHierarchy()) instanceof NodeTest) {
sortKey.expression = Atomizer.makeAtomizer(key);
}
SortKeyDefinition sk = new SortKeyDefinition();
sk.setSortKey(sortKey.expression, true);
AtomicComparer comp = sk.makeComparator(context);
AtomicComparer[] comps = {comp};
iterator = new SortedIterator(context, iterator, this, comps, true);
((SortedIterator)iterator).setHostLanguage(Configuration.XPATH);
}
return iterator;
}
/**
* JAXP 1.3 evaluate() method
* @param node The context node. This must use a representation of nodes that this implementation understands.
* This may be a Saxon NodeInfo, or a node in one of the external object models supported, for example
* DOM, DOM4J, JDOM, or XOM, provided the support module for that object model is loaded.
*
* Contrary to the interface specification, Saxon does not supply an empty
* document when the value is null. This is because Saxon supports multiple object models,
* and it's unclear what kind of document node would be appropriate. Instead, Saxon uses
* the node supplied to the {@link #setContextNode} method if available, and if none
* is available, executes the XPath expression with the context item undefined.
* Saxon does not allow a NodeList to be supplied for this parameter. It's not clear
* what this would be intended to mean.
* @param qName Indicates the type of result required. This must be one of the constants defined in
* the JAXP {@link XPathConstants} class.
* Saxon will attempt to convert the actual result of the expression to the required type using the
* XPath 1.0 conversion rules.
* @return the result of the evaluation, as a Java object of the appropriate type. Saxon interprets the
* rules as follows:
*
* QName Return Value
*
* BOOLEAN
* The effective boolean value of the actual result,
* as a Java Boolean object
* STRING
* The result of applying the string() function to the actual result,
* as a Java String object
* NUMBER
* The result of applying the number() function to the actual result,
* as a Java Double object
* NODE
* A single node, in the native data model supplied as input. If the
* expression returns more than one node, the first is returned. If
* the expression returns an empty sequence, null is returned. If the
* expression returns an atomic value, or if the first item in the
* result sequence is an atomic value, an exception is thrown.
* NODESET
* This is interpreted as allowing any sequence, of nodes or atomic values.
* If the first argument is a wrapper around a DOM Node, then the result is
* returned as a DOM NodeList, and an exception is then thrown if the result sequence
* contains a value that is not a DOM Node. In all other cases
* the result is returned as a Java List object, unless it is empty, in which
* case null is returned. The contents of the list may be node objects (in the
* native data model supplied as input), or Java objects representing the XPath
* atomic values in the actual result: String for an xs:string, Double for a xs:double,
* Long for an xs:integer, and so on. (For safety, cast the values to a type
* such as xs:string within the XPath expression).
*
* @throws XPathExpressionException if evaluation of the expression fails or if the
* result cannot be converted to the requested type.
*/
/*@Nullable*/ public Object evaluate(/*@Nullable*/ Object node, /*@NotNull*/ QName qName) throws XPathExpressionException {
NodeInfo contextNode = this.contextNode;
if (node != null) {
if (node instanceof SingletonItem) {
node = ((SingletonItem)node).asItem();
}
if (node instanceof NodeInfo) {
if (!((NodeInfo)node).getConfiguration().isCompatible(config)) {
throw new XPathExpressionException(
"Supplied node must be built using the same or a compatible Configuration");
}
if (node instanceof DocumentInfo && ((DocumentInfo)node).isTyped() && !executable.isSchemaAware()) {
throw new XPathExpressionException(
"The expression was compiled to handled untyped data, but the input is typed");
}
contextNode = ((NodeInfo)node);
} else {
JPConverter converter = JPConverter.allocate(node.getClass(), config);
Sequence val;
try {
val = converter.convert(node, new EarlyEvaluationContext(config, null));
} catch (XPathException e) {
throw new XPathExpressionException(
"Failure converting a node of class " + node.getClass().getName() +
": " + e.getMessage());
}
if (val instanceof NodeInfo) {
if (!((NodeInfo)val).getConfiguration().isCompatible(config)) {
throw new XPathExpressionException(
"Supplied node must be built using the same or a compatible Configuration");
}
if (node instanceof DocumentInfo && ((DocumentInfo)node).isTyped() && !executable.isSchemaAware()) {
throw new XPathExpressionException(
"The expression was compiled to handled untyped data, but the input is typed");
}
contextNode = (NodeInfo)val;
} else {
throw new XPathExpressionException(
"Cannot locate an object model implementation for nodes of class "
+ node.getClass().getName());
}
}
}
XPathContextMajor context = new XPathContextMajor(contextNode, executable);
context.openStackFrame(stackFrameMap);
try {
if (qName.equals(XPathConstants.BOOLEAN)) {
return Boolean.valueOf(expression.effectiveBooleanValue(context));
} else if (qName.equals(XPathConstants.STRING)) {
SequenceIterator iter = expression.iterate(context);
Item first = iter.next();
if (first == null) {
return "";
}
return first.getStringValue();
} else if (qName.equals(XPathConstants.NUMBER)) {
if (atomizer == null) {
atomizer = Atomizer.makeAtomizer(expression);
}
SequenceIterator iter = atomizer.iterate(context);
Item first = iter.next();
if (first == null) {
return new Double(Double.NaN);
}
if (first instanceof NumericValue) {
return new Double(((NumericValue)first).getDoubleValue());
} else {
DoubleValue v = NumberFn.convert((AtomicValue)first, getConfiguration());
return new Double(v.getDoubleValue());
}
} else if (qName.equals(XPathConstants.NODE)) {
SequenceIterator iter = expression.iterate(context);
Item first = iter.next();
if (first instanceof VirtualNode) {
return ((VirtualNode)first).getRealNode();
}
if (first == null || first instanceof NodeInfo) {
return first;
}
throw new XPathExpressionException("Expression result is not a node");
} else if (qName.equals(XPathConstants.NODESET)) {
//SequenceIterator iter = expression.iterate(context);
SequenceIterator iter = rawIterator(context);
SequenceExtent extent = new SequenceExtent(iter);
PJConverter converter = PJConverter.allocateNodeListCreator(config, node);
return converter.convert(extent, Object.class, context);
} else {
throw new IllegalArgumentException("qName: Unknown type for expected result");
}
} catch (XPathException e) {
throw new XPathExpressionException(e);
}
}
/**
* Evaluate the expression to return a string value
* @param node the initial context node. This must be either an instance of NodeInfo or a node
* recognized by a known external object model.
* Contrary to the interface specification, Saxon does not supply an empty
* document when the value is null. This is because Saxon supports multiple object models,
* and it's unclear what kind of document node would be appropriate. Instead, Saxon uses
* the node supplied to the {@link #setContextNode} method if available, and if none
* is available, executes the XPath expression with the context item undefined.
* @return the results of the expression, converted to a String
* @throws XPathExpressionException if evaluation fails
*/
/*@NotNull*/ public String evaluate(Object node) throws XPathExpressionException {
return (String)evaluate(node, XPathConstants.STRING);
}
/**
* Evaluate the XPath expression against an input source to obtain a result of a specified type
* @param inputSource The input source document against which the expression is evaluated.
* (Note that there is no caching. This will be parsed, and the parsed result will be discarded.)
* If the supplied value is null then (contrary to the JAXP specifications), the XPath expression
* is evaluated with the context item undefined.
* @param qName The type required, identified by a constant in {@link XPathConstants}
* @return the result of the evaluation, as a Java object of the appropriate type:
* see {@link #evaluate(Object, javax.xml.namespace.QName)}
* @throws XPathExpressionException
*/
/*@Nullable*/ public Object evaluate(/*@Nullable*/ InputSource inputSource, /*@Nullable*/ QName qName) throws XPathExpressionException {
if (qName == null) {
throw new NullPointerException("qName");
}
try {
NodeInfo doc = null;
if (inputSource != null) {
doc = config.buildDocument(new SAXSource(inputSource));
}
return evaluate(doc, qName);
} catch (XPathException e) {
throw new XPathExpressionException(e);
}
}
/**
* Evaluate the XPath expression against an input source to obtain a string result
* @param inputSource The input source document against which the expression is evaluated.
* (Note that there is no caching. This will be parsed, and the parsed result will be discarded.)
* @return the result of the evaluation, converted to a String
* @throws XPathExpressionException in the event of an XPath dynamic error
* @throws NullPointerException If inputSource
is null
.
*/
/*@NotNull*/ public String evaluate(/*@Nullable*/ InputSource inputSource) throws XPathExpressionException {
if (inputSource == null) {
throw new NullPointerException("inputSource");
}
try {
NodeInfo doc = config.buildDocument(new SAXSource(inputSource));
return (String)evaluate(doc, XPathConstants.STRING);
} catch (XPathException e) {
throw new XPathExpressionException(e);
}
}
/**
* Callback for evaluating the sort keys. For internal use only.
*/
/*@Nullable*/ public AtomicValue evaluateSortKey(int n, XPathContext c) throws XPathException {
return (AtomicValue)sortKey.getInternalExpression().evaluateItem(c);
}
/**
* Low-level method to get the internal Saxon expression object. This exposes a wide range of
* internal methods that may be needed by specialized applications, and allows greater control
* over the dynamic context for evaluating the expression.
* @return the underlying Saxon expression object.
*/
public Expression getInternalExpression() {
return expression;
}
}