net.sf.saxon.pull.VirtualTreeWalker Maven / Gradle / Ivy
Show all versions of saxon9 Show documentation
package net.sf.saxon.pull;
import net.sf.saxon.Configuration;
import net.sf.saxon.sort.IntArraySet;
import net.sf.saxon.event.PipelineConfiguration;
import net.sf.saxon.expr.XPathContext;
import net.sf.saxon.instruct.DocumentInstr;
import net.sf.saxon.instruct.ElementCreator;
import net.sf.saxon.instruct.ParentNodeConstructor;
import net.sf.saxon.om.*;
import net.sf.saxon.trans.XPathException;
import net.sf.saxon.type.Type;
import net.sf.saxon.value.AtomicValue;
import javax.xml.transform.SourceLocator;
import java.util.ArrayList;
import java.util.List;
import java.util.Stack;
/**
* A virtual tree walker provides a sequence of pull events describing the structure and content of a tree
* that is conceptually being constructed by expressions in a query or stylesheet; in fact the tree is
* not necessarily constructed in memory, and exists only as this stream of pull events.
*
* The tree is physically constructed if operations are requested that depend on the identity of the nodes
* in the tree, or that navigate within the tree. Operations such as copying or atomizing the tree can be done
* directly, without building it in memory. (Note however that if such operations are done more than once, the
* underlying instructions may be evaluated repeatedly.)
*/
public class VirtualTreeWalker implements PullProvider, NamespaceDeclarations {
private PipelineConfiguration pipe;
private int currentEvent = START_OF_INPUT;
private int nameCode;
private int nextNameCode;
private ParentNodeConstructor instruction;
private XPathContext context;
private Stack constructorStack = new Stack();
private Stack iteratorStack = new Stack();
private PullProvider subordinateTreeWalker = null;
// The subordinateTreeWalker is used if the tree construction expression pulls in references
// to document or element nodes in an existing source document. When this happens, tree walking
// events generated by walking the source document are copied into to the stream of events
// generated by this class.
private boolean alreadyRead = false;
private boolean allowAttributes = false;
private int stripDepth = -1;
// If this is >0, it indicates that instructions on the constructor stack with depth < stripDepth
// specify validation=preserve, while those at greater depth effectively specify validation=strip
// (It doesn't matter if they actually say validation=preserve, the stripping takes priority)
private AttributeCollectionImpl attributes;
private boolean foundAttributes;
private int[] activeNamespaces;
private ArrayList additionalNamespaces = new ArrayList(10); // array of namespace nodes
/**
* Create a VirtualTreeWalker to navigate the tree constructed by evaluating a given instruction
* in a given dyamic context
* @param instruction the instruction (this will always be an instruction that creates element or
* document nodes)
* @param context the dynamic evaluation context
*/
public VirtualTreeWalker(ParentNodeConstructor instruction, XPathContext context) {
this.instruction = instruction;
this.context = context;
}
/**
* Set configuration information. This must only be called before any events
* have been read.
*/
public void setPipelineConfiguration(PipelineConfiguration pipe) {
this.pipe = pipe;
}
/**
* Get configuration information.
* @return the pipeline configuration
*/
public PipelineConfiguration getPipelineConfiguration() {
return pipe;
}
/**
* Get the namepool
* @return the NamePool
*/
public NamePool getNamePool() {
return pipe.getConfiguration().getNamePool();
}
/**
* Get the next event
*
* @return an integer code indicating the type of event. The code
* {@link #END_OF_INPUT} is returned at the end of the sequence.
*/
public int next() throws XPathException {
try {
// First see if we are currently walking some other tree that has been logically
// copied into this tree.
if (subordinateTreeWalker != null) {
currentEvent = subordinateTreeWalker.next();
if (currentEvent == END_OF_INPUT) {
subordinateTreeWalker = null;
return next();
}
return currentEvent;
}
// On the first call, produce a START_ELEMENT or START_DOCUMENT event depending on the instruction
if (currentEvent == START_OF_INPUT) {
constructorStack.push(instruction);
if (stripDepth < 0 && instruction.getValidationMode() == Validation.STRIP) {
stripDepth = constructorStack.size();
}
SequenceIterator content = instruction.getContentExpression().iterate(context);
iteratorStack.push(content);
if (instruction instanceof DocumentInstr) {
currentEvent = START_DOCUMENT;
nameCode = -1;
} else {
currentEvent = START_ELEMENT;
nameCode = ((ElementCreator)instruction).getNameCode(context);
allowAttributes = true;
// look ahead to generate all the attributes and namespaces of the element
processAttributesAndNamespaces((ElementCreator)instruction, content);
// remember that we've read one event too many
allowAttributes = false;
alreadyRead = true;
}
return currentEvent;
}
if (iteratorStack.isEmpty()) {
// if we're at the top level, see if we've just started, or are about to finish
if (currentEvent == START_DOCUMENT || currentEvent == START_ELEMENT) {
// we've just started: start processing the content of the instruction
SequenceIterator iter = instruction.getContentExpression().iterate(context);
constructorStack.push(instruction);
if (stripDepth < 0 && instruction.getValidationMode() == Validation.STRIP) {
stripDepth = constructorStack.size();
}
iteratorStack.push(iter);
} else if (currentEvent == END_DOCUMENT || currentEvent == END_ELEMENT) {
// we're about to finish
currentEvent = END_OF_INPUT;
return currentEvent;
} else {
// we're going to finish soon, but must first generate the last END_DOCUMENT or END_ELEMENT event
currentEvent =
((instruction instanceof DocumentInstr) ? END_DOCUMENT : END_ELEMENT);
return currentEvent;
}
}
// Read the next item from the current content iterator
SequenceIterator iterator = ((SequenceIterator)iteratorStack.peek());
if (alreadyRead) {
Item item = iterator.current();
alreadyRead = false;
nameCode = nextNameCode;
return processItem(iterator, item);
} else {
return processItem(iterator, iterator.next());
}
} catch (XPathException e) {
// report any dynamic errors unless already reported
context.getController().reportFatalError(e);
throw e;
}
}
private FastStringBuffer textNodeBuffer = new FastStringBuffer(100);
private CharSequence currentTextValue = null;
/**
* Process an item in the content of an element or document node
* @param iterator the iterator over the contents of the element or document
* @param item the current item to be processed, or null if the end of the content
* iterator has been reached
* @return the next event code
* @throws XPathException if a dynamic error occurs
*/
private int processItem(SequenceIterator iterator, Item item) throws XPathException {
if (item == null) {
// we've reached the end of the children
if (stripDepth == constructorStack.size()) {
stripDepth = -1;
}
ParentNodeConstructor inst = (ParentNodeConstructor)constructorStack.pop();
if (inst instanceof DocumentInstr) {
iteratorStack.pop();
if (iteratorStack.isEmpty()) {
currentEvent = END_DOCUMENT;
nameCode = -1;
return currentEvent;
}
// skip the END_DOCUMENT event for a nested document node
return next();
} else {
currentEvent = END_ELEMENT;
nameCode = -1;
iteratorStack.pop();
return currentEvent;
}
} else if (item instanceof UnconstructedParent) {
// this represents a nested element or document node constructor
UnconstructedParent parent = (UnconstructedParent)item;
ParentNodeConstructor inst = parent.getInstruction();
constructorStack.push(inst);
if (stripDepth < 0 && inst.getValidationMode() == Validation.STRIP) {
stripDepth = constructorStack.size();
}
SequenceIterator content = inst.getContentExpression().iterate(parent.getXPathContext());
if (inst instanceof DocumentInstr) {
iteratorStack.push(content);
// skip the START_DOCUMENT event
return next();
} else {
currentEvent = START_ELEMENT;
nameCode = ((UnconstructedElement)item).getNameCode();
processAttributesAndNamespaces((ElementCreator)inst, content);
alreadyRead = true;
iteratorStack.push(content);
return currentEvent;
}
} else if (item instanceof AtomicValue) {
currentTextValue = textNodeBuffer;
textNodeBuffer.setLength(0);
textNodeBuffer.append(item.getStringValueCS());
while (true) {
Item next = iterator.next();
if (next instanceof AtomicValue) {
textNodeBuffer.append(' ');
textNodeBuffer.append(next.getStringValueCS());
//continue;
} else {
currentEvent = TEXT;
nameCode = -1;
alreadyRead = true;
return currentEvent;
}
}
} else {
nameCode = ((NodeInfo)item).getNameCode();
switch (((NodeInfo)item).getNodeKind()) {
case Type.TEXT:
currentEvent = TEXT;
currentTextValue = item.getStringValueCS();
return currentEvent;
case Type.COMMENT:
currentEvent = COMMENT;
return currentEvent;
case Type.PROCESSING_INSTRUCTION:
currentEvent = PROCESSING_INSTRUCTION;
return currentEvent;
case Type.ATTRIBUTE:
if (!allowAttributes) {
XPathException de;
if (constructorStack.peek() instanceof DocumentInstr) {
de = new XPathException(
"Attributes cannot be attached to a document node");
if (pipe.getHostLanguage() == Configuration.XQUERY) {
de.setErrorCode("XQTY0004");
} else {
de.setErrorCode("XTDE0420");
}
} else {
de = new XPathException(
"Attributes in the content of an element must come before the child nodes");
if (pipe.getHostLanguage() == Configuration.XQUERY) {
de.setErrorCode("XQDY0024");
} else {
de.setErrorCode("XTDE0410");
}
}
de.setXPathContext(context);
de.setLocator(getSourceLocator());
throw de;
}
currentEvent = ATTRIBUTE;
return currentEvent;
case Type.NAMESPACE:
if (!allowAttributes) {
XPathException de = new XPathException("Namespace nodes in the content of an element must come before the child nodes");
de.setErrorCode("XTDE0410");
de.setXPathContext(context);
de.setLocator(getSourceLocator());
throw de;
}
currentEvent = NAMESPACE;
return currentEvent;
case Type.ELEMENT:
subordinateTreeWalker = TreeWalker.makeTreeWalker((NodeInfo)item);
subordinateTreeWalker.setPipelineConfiguration(pipe);
currentEvent = subordinateTreeWalker.next();
nameCode = subordinateTreeWalker.getNameCode();
return currentEvent;
case Type.DOCUMENT:
subordinateTreeWalker = TreeWalker.makeTreeWalker((NodeInfo)item);
subordinateTreeWalker.setPipelineConfiguration(pipe);
subordinateTreeWalker = new DocumentEventIgnorer(subordinateTreeWalker);
subordinateTreeWalker.setPipelineConfiguration(pipe);
currentEvent = subordinateTreeWalker.next();
nameCode = -1;
return currentEvent;
default:
throw new IllegalStateException();
}
}
}
/**
* Following a START_ELEMENT event, evaluate the contents of the element to obtain all attributes and namespaces.
* This process stops when the first event other than an attribute or namespace is read. We then remember the
* extra event, which will be the next event returned in the normal sequence. Note that the relative order
* of attributes and namespaces is undefined.
* @param inst The instruction that creates the element node
* @param content Iterator over the expression that generates the attributes, namespaces, and content of the
* element
* @throws XPathException if any dynamic error occurs
*/
private void processAttributesAndNamespaces(ElementCreator inst, SequenceIterator content) throws XPathException {
foundAttributes = false;
additionalNamespaces.clear();
activeNamespaces = inst.getActiveNamespaces();
if (activeNamespaces == null) {
activeNamespaces = IntArraySet.EMPTY_INT_ARRAY;
}
// if the namespace of the element name itself is not one of the active namespaces, make sure
// a namespace node is created for it
// NamePool pool = context.getNamePool();
// int nscode = pool.allocateNamespaceCode(nameCode);
// boolean found = false;
// for (int i=0; i= 0) {
// Attribute already exists. In XQuery this is an error. In XSLT, the last attribute wins
if (context.getController().getExecutable().getHostLanguage() == Configuration.XSLT) {
attributes.setAttribute(index,
node.getNameCode(),
preserve ? node.getTypeAnnotation() : StandardNames.XS_UNTYPED_ATOMIC,
node.getStringValue(), 0, 0);
} else {
XPathException de = new XPathException("The attributes of an element must have distinct names");
de.setErrorCode("XQDY0025");
de.setXPathContext(context);
de.setLocator(getSourceLocator());
throw de;
}
} else {
attributes.addAttribute(
node.getNameCode(),
preserve ? node.getTypeAnnotation() : StandardNames.XS_UNTYPED_ATOMIC,
node.getStringValue(),
0, 0);
}
// if the namespace of the attribute name itself has not already been declared, make sure
// a namespace node is created for it
// int anc = node.getNameCode();
// if ((anc >> 20) != 0) {
// // the attribute name is prefixed
//
// int anscode = pool.allocateNamespaceCode(anc);
// boolean afound = false;
// for (int i=0; iAttributes may be read before or after reading the namespaces of an element,
* but must not be read after the first child node has been read, or after calling
* one of the methods skipToEnd(), getStringValue(), or getTypedValue().
*
* @return an AttributeCollection representing the attributes of the element
* that has just been notified.
*/
public AttributeCollection getAttributes() throws XPathException {
if (subordinateTreeWalker != null) {
return subordinateTreeWalker.getAttributes();
} else {
if (foundAttributes) {
return attributes;
} else {
return AttributeCollectionImpl.EMPTY_ATTRIBUTE_COLLECTION;
}
}
}
/**
* Get the namespace declarations associated with the current element. This method must
* be called only after a START_ELEMENT event has been notified. In the case of a top-level
* START_ELEMENT event (that is, an element that either has no parent node, or whose parent
* is not included in the sequence being read), the NamespaceDeclarations object returned
* will contain a namespace declaration for each namespace that is in-scope for this element
* node. In the case of a non-top-level element, the NamespaceDeclarations will contain
* a set of namespace declarations and undeclarations, representing the differences between
* this element and its parent.
*
* It is permissible for this method to return namespace declarations that are redundant.
*
* The NamespaceDeclarations object is guaranteed to remain unchanged until the next START_ELEMENT
* event, but may then be overwritten. The object should not be modified by the client.
*
* Namespaces may be read before or after reading the attributes of an element,
* but must not be read after the first child node has been read, or after calling
* one of the methods skipToEnd(), getStringValue(), or getTypedValue().
*
*/
public NamespaceDeclarations getNamespaceDeclarations() throws XPathException {
if (subordinateTreeWalker != null) {
return subordinateTreeWalker.getNamespaceDeclarations();
} else {
return this;
}
}
/**
* Skip the current subtree. This method may be called only immediately after
* a START_DOCUMENT or START_ELEMENT event. This call returns the matching
* END_DOCUMENT or END_ELEMENT event; the next call on next() will return
* the event following the END_DOCUMENT or END_ELEMENT.
* @throws IllegalStateException if the method is called at any time other than
* immediately after a START_DOCUMENT or START_ELEMENT event.
*/
public int skipToMatchingEnd() throws XPathException {
if (currentEvent != START_DOCUMENT && currentEvent != START_ELEMENT) {
throw new IllegalStateException();
}
if (subordinateTreeWalker != null) {
return subordinateTreeWalker.skipToMatchingEnd();
} else {
SequenceIterator content = (SequenceIterator)iteratorStack.peek();
if (alreadyRead) {
alreadyRead = false;
}
while (true) {
Item next = content.next();
if (next == null) {
break;
}
}
return (currentEvent == START_DOCUMENT ? END_DOCUMENT : END_ELEMENT);
}
}
/**
* Close the event reader. This indicates that no further events are required.
* It is not necessary to close an event reader after {@link #END_OF_INPUT} has
* been reported, but it is recommended to close it if reading terminates
* prematurely. Once an event reader has been closed, the effect of further
* calls on next() is undefined.
*/
public void close() {
if (subordinateTreeWalker != null) {
subordinateTreeWalker.close();
} else {
// do nothing
}
}
/**
* Set the initial nameCode
* @param nameCode the nameCode of the node at the root of the tree being walked
*/
public void setNameCode(int nameCode) {
this.nameCode = nameCode;
}
/**
* Get the nameCode identifying the name of the current node. This method
* can be used after the {@link #START_ELEMENT}, {@link #END_ELEMENT}, {@link #PROCESSING_INSTRUCTION},
* {@link #ATTRIBUTE}, or {@link #NAMESPACE} events. With some PullProvider implementations,
* but not this one, it can also be used after {@link #END_ELEMENT}.
* If called at other times, the result is undefined and may result in an IllegalStateException.
* If called when the current node is an unnamed namespace node (a node representing the default namespace)
* the returned value is -1.
*
* @return the nameCode. The nameCode can be used to obtain the prefix, local name,
* and namespace URI from the name pool.
*/
public int getNameCode() {
if (subordinateTreeWalker != null) {
return subordinateTreeWalker.getNameCode();
}
return nameCode;
}
/**
* Get the fingerprint of the name of the element. This is similar to the nameCode, except that
* it does not contain any information about the prefix: so two elements with the same fingerprint
* have the same name, excluding prefix. This method
* can be used after the {@link #START_ELEMENT}, {@link #END_ELEMENT}, {@link #PROCESSING_INSTRUCTION},
* {@link #ATTRIBUTE}, or {@link #NAMESPACE} events.
* If called at other times, the result is undefined and may result in an IllegalStateException.
* If called when the current node is an unnamed namespace node (a node representing the default namespace)
* the returned value is -1.
*
* @return the fingerprint. The fingerprint can be used to obtain the local name
* and namespace URI from the name pool.
*/
public int getFingerprint() {
int nc = getNameCode();
if (nc == -1) {
return -1;
} else {
return nc & NamePool.FP_MASK;
}
}
/**
* Get the string value of the current element, text node, processing-instruction,
* or top-level attribute or namespace node, or atomic value.
*
* In other situations the result is undefined and may result in an IllegalStateException.
*
* If the most recent event was a {@link #START_ELEMENT}, this method causes the content
* of the element to be read. The current event on completion of this method will be the
* corresponding {@link #END_ELEMENT}. The next call of next() will return the event following
* the END_ELEMENT event.
*
* @return the String Value of the node in question, defined according to the rules in the
* XPath data model.
*/
public CharSequence getStringValue() throws XPathException {
if (subordinateTreeWalker != null) {
return subordinateTreeWalker.getStringValue();
} else if (currentEvent == TEXT) {
return currentTextValue;
} else if (currentEvent != START_ELEMENT && currentEvent != START_DOCUMENT) {
SequenceIterator content = (SequenceIterator)iteratorStack.peek();
if (content.current() == null) {
return "";
}
return content.current().getStringValue();
} else {
FastStringBuffer sb = new FastStringBuffer(100);
SequenceIterator content = (SequenceIterator)iteratorStack.peek();
boolean previousAtomic = false;
if (alreadyRead) {
Item current = content.current();
if (current == null) {
return "";
}
processText(current, sb);
previousAtomic = (current instanceof AtomicValue);
alreadyRead = false;
}
while (true) {
Item next = content.next();
if (next == null) {
break;
}
boolean atomic = (next instanceof AtomicValue);
if (atomic && previousAtomic) {
sb.append(' ');
}
processText(next, sb);
previousAtomic = atomic;
}
return sb;
}
}
/**
* Add the string value of a child item to a string buffer that is used to accumulate the
* string value of a document or element node
* @param item the child item
* @param sb the string buffer where the content is accumulated
*/
private void processText(Item item, FastStringBuffer sb) {
if (item instanceof UnconstructedParent) {
sb.append(item.getStringValueCS());
} else if (item instanceof AtomicValue) {
sb.append(item.getStringValueCS());
} else {
NodeInfo node = (NodeInfo)item;
switch (node.getNodeKind()) {
case Type.DOCUMENT:
case Type.ELEMENT:
case Type.TEXT:
sb.append(node.getStringValueCS());
default:
// do nothing
}
}
}
/**
* Get an atomic value. This call may be used only when the last event reported was
* ATOMIC_VALUE. This indicates that the PullProvider is reading a sequence that contains
* a free-standing atomic value; it is never used when reading the content of a node.
*/
public AtomicValue getAtomicValue() {
throw new IllegalStateException();
}
/**
* Get the type annotation of the current attribute or element node, or atomic value.
* The result of this method is undefined unless the most recent event was START_ELEMENT,
* ATTRIBUTE, or ATOMIC_VALUE.
*
* @return the type annotation. This code is the fingerprint of a type name, which may be
* resolved to a {@link net.sf.saxon.type.SchemaType} by access to the Configuration.
*/
public int getTypeAnnotation() {
if (subordinateTreeWalker != null && stripDepth < 0) {
return subordinateTreeWalker.getTypeAnnotation();
} else {
return -1;
}
}
/**
* Get the location of the current event.
* For an event stream representing a real document, the location information
* should identify the location in the lexical XML source. For a constructed document, it should
* identify the location in the query or stylesheet that caused the node to be created.
* A value of null can be returned if no location information is available.
*/
public SourceLocator getSourceLocator() {
return instruction;
}
/**
* Get the number of declarations (and undeclarations) in this list.
*/
public int getNumberOfNamespaces() {
return activeNamespaces.length + additionalNamespaces.size();
}
/**
* Get the prefix of the n'th declaration (or undeclaration) in the list,
* counting from zero.
*
* @param index the index identifying which declaration is required.
* @return the namespace prefix. For a declaration or undeclaration of the
* default namespace, this is the zero-length string.
* @throws IndexOutOfBoundsException if the index is out of range.
*/
public String getPrefix(int index) {
if (index < activeNamespaces.length) {
return getNamePool().getPrefixFromNamespaceCode(activeNamespaces[index]);
} else {
return ((NodeInfo)additionalNamespaces.get(index - activeNamespaces.length)).getLocalPart();
}
}
/**
* Get the namespace URI of the n'th declaration (or undeclaration) in the list,
* counting from zero.
*
* @param index the index identifying which declaration is required.
* @return the namespace URI. For a namespace undeclaration, this is the
* zero-length string.
* @throws IndexOutOfBoundsException if the index is out of range.
*/
public String getURI(int index) {
if (index < activeNamespaces.length) {
return getNamePool().getURIFromNamespaceCode(activeNamespaces[index]);
} else {
return ((NodeInfo)additionalNamespaces.get(index - activeNamespaces.length)).getStringValue();
}
}
/**
* Get the n'th declaration in the list in the form of a namespace code. Namespace
* codes can be translated into a prefix and URI by means of methods in the
* NamePool
*
* @param index the index identifying which declaration is required.
* @return the namespace code. This is an integer whose upper half indicates
* the prefix (0 represents the default namespace), and whose lower half indicates
* the URI (0 represents an undeclaration).
* @throws IndexOutOfBoundsException if the index is out of range.
* @see net.sf.saxon.om.NamePool#getPrefixFromNamespaceCode(int)
* @see net.sf.saxon.om.NamePool#getURIFromNamespaceCode(int)
*/
public int getNamespaceCode(int index) {
if (index < activeNamespaces.length) {
return activeNamespaces[index];
} else {
return getNamePool().allocateNamespaceCode(getPrefix(index), getURI(index));
}
}
/**
* Get all the namespace codes, as an array.
*
* @param buffer a sacrificial array that the method is free to use to contain the result.
* May be null.
* @return an integer array containing namespace codes. The array may be filled completely
* with namespace codes, or it may be incompletely filled, in which case a -1 integer acts
* as a terminator.
*/
public int[] getNamespaceCodes(int[] buffer) {
if (buffer.length < getNumberOfNamespaces()) {
buffer = new int[getNumberOfNamespaces()];
} else {
buffer[getNumberOfNamespaces()] = -1;
}
for (int i=0; i