domconsumer.java

来自「kaffe Java 解释器语言,源码,Java的子集系统,开放源代码」· Java 代码 · 共 944 行 · 第 1/2 页

/*
 * Copyright (C) 1999-2001 David Brownell
 * 
 * This file is part of GNU JAXP, a library.
 *
 * GNU JAXP is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 * 
 * GNU JAXP is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 * 
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 *
 * As a special exception, if you link this library with other files to
 * produce an executable, this library does not by itself cause the
 * resulting executable to be covered by the GNU General Public License.
 * This exception does not however invalidate any other reasons why the
 * executable file might be covered by the GNU General Public License. 
 */

package gnu.xml.pipeline;

import java.util.Hashtable;

import org.w3c.dom.*;
import org.xml.sax.*;
import org.xml.sax.ext.DeclHandler;
import org.xml.sax.ext.LexicalHandler;
import org.xml.sax.helpers.AttributesImpl;

import gnu.xml.util.DomParser;


/**
 * This consumer builds a DOM Document from its input, acting either as a
 * pipeline terminus or as an intermediate buffer.  When a document's worth
 * of events has been delivered to this consumer, that document is read with
 * a {@link DomParser} and sent to the next consumer.  It is also available
 * as a read-once property.
 *
 * <p>The DOM tree is constructed as faithfully as possible.  There are some
 * complications since a DOM should expose behaviors that can't be implemented
 * without API backdoors into that DOM, and because some SAX parsers don't
 * report all the information that DOM permits to be exposed.  The general
 * problem areas involve information from the Document Type Declaration (DTD).
 * DOM only represents a limited subset, but has some behaviors that depend
 * on much deeper knowledge of a document's DTD.  You shouldn't have much to
 * worry about unless you change handling of "noise" nodes from its default
 * setting (which ignores them all); note if you use JAXP to populate your
 * DOM trees, it wants to save "noise" nodes by default.  (Such nodes include
 * ignorable whitespace, comments, entity references and CDATA boundaries.)
 * Otherwise, your
 * main worry will be if you use a SAX parser that doesn't flag ignorable
 * whitespace unless it's validating (few don't).
 *
 * <p> The SAX2 events used as input must contain XML Names for elements
 * and attributes, with original prefixes.  In SAX2,
 * this is optional unless the "namespace-prefixes" parser feature is set.
 * Moreover, many application components won't provide completely correct
 * structures anyway.  <em>Before you convert a DOM to an output document,
 * you should plan to postprocess it to create or repair such namespace
 * information.</em> The {@link NSFilter} pipeline stage does such work.
 *
 * <p> <em>Note:  changes late in DOM L2 process made it impractical to
 * attempt to create the DocumentType node in any implementation-neutral way,
 * much less to populate it (L1 didn't support even creating such nodes).
 * To create and populate such a node, subclass the inner
 * {@link DomConsumer.Handler} class and teach it about the backdoors into
 * whatever DOM implementation you want.  It's possible that some revised
 * DOM API (L3?) will make this problem solvable again. </em>
 *
 * @see DomParser
 *
 * @author David Brownell
 */
public class DomConsumer implements EventConsumer
{
    private Class		domImpl;

    private boolean		hidingCDATA = true;
    private boolean		hidingComments = true;
    private boolean		hidingWhitespace = true;
    private boolean		hidingReferences = true;

    private Handler		handler;
    private ErrorHandler	errHandler;

    private EventConsumer	next;

    // FIXME:  this can't be a generic pipeline stage just now,
    // since its input became a Class not a String (to be turned
    // into a class, using the right class loader)


    /**
     * Configures this pipeline terminus to use the specified implementation
     * of DOM when constructing its result value.
     *
     * @param impl class implementing {@link org.w3c.dom.Document Document}
     *	which publicly exposes a default constructor
     *
     * @exception SAXException when there is a problem creating an
     *	empty DOM document using the specified implementation
     */
    public DomConsumer (Class impl)
    throws SAXException
    {
	domImpl = impl;
	handler = new Handler (this);
    }

    /**
     * This is the hook through which a subclass provides a handler
     * which knows how to access DOM extensions, specific to some
     * implementation, to record additional data in a DOM.
     * Treat this as part of construction; don't call it except
     * before (or between) parses.
     */
    protected void setHandler (Handler h)
    {
	handler = h;
    }


    private Document emptyDocument ()
    throws SAXException
    {
	try {
	    return (Document) domImpl.newInstance ();
	} catch (IllegalAccessException e) {
	    throw new SAXException ("can't access constructor: "
		    + e.getMessage ());
	} catch (InstantiationException e) {
	    throw new SAXException ("can't instantiate Document: "
		    + e.getMessage ());
	}
    }


    /**
     * Configures this consumer as a buffer/filter, using the specified
     * DOM implementation when constructing its result value.
     *
     * <p> This event consumer acts as a buffer and filter, in that it
     * builds a DOM tree and then writes it out when <em>endDocument</em>
     * is invoked.  Because of the limitations of DOM, much information
     * will as a rule not be seen in that replay.  To get a full fidelity
     * copy of the input event stream, use a {@link TeeConsumer}.
     *
     * @param impl class implementing {@link org.w3c.dom.Document Document}
     *	which publicly exposes a default constructor
     * @param next receives a "replayed" sequence of parse events when
     *	the <em>endDocument</em> method is invoked.
     *
     * @exception SAXException when there is a problem creating an
     *	empty DOM document using the specified DOM implementation
     */
    public DomConsumer (Class impl, EventConsumer n)
    throws SAXException
    {
	this (impl);
	next = n;
    }


    /**
     * Returns the document constructed from the preceding
     * sequence of events.  This method should not be
     * used again until another sequence of events has been
     * given to this EventConsumer.  
     */
    final public Document getDocument ()
    {
	return handler.clearDocument ();
    }

    public void setErrorHandler (ErrorHandler handler)
    {
	errHandler = handler;
    }


    /**
     * Returns true if the consumer is hiding entity references nodes
     * (the default), and false if EntityReference nodes should
     * instead be created.  Such EntityReference nodes will normally be
     * empty, unless an implementation arranges to populate them and then
     * turn them back into readonly objects.
     *
     * @see #setHidingReferences
     */
    final public boolean	isHidingReferences ()
	{ return hidingReferences; }

    /**
     * Controls whether the consumer will hide entity expansions,
     * or will instead mark them with entity reference nodes.
     *
     * @see #isHidingReferences
     * @param flag False if entity reference nodes will appear
     */
    final public void		setHidingReferences (boolean flag)
	{ hidingReferences = flag; }
    

    /**
     * Returns true if the consumer is hiding comments (the default),
     * and false if they should be placed into the output document.
     *
     * @see #setHidingComments
     */
    public final boolean isHidingComments ()
	{ return hidingComments; }

    /**
     * Controls whether the consumer is hiding comments.
     *
     * @see #isHidingComments
     */
    public final void setHidingComments (boolean flag)
	{ hidingComments = flag; }


    /**
     * Returns true if the consumer is hiding ignorable whitespace
     * (the default), and false if such whitespace should be placed
     * into the output document as children of element nodes.
     *
     * @see #setHidingWhitespace
     */
    public final boolean isHidingWhitespace ()
	{ return hidingWhitespace; }

    /**
     * Controls whether the consumer hides ignorable whitespace
     *
     * @see #isHidingComments
     */
    public final void setHidingWhitespace (boolean flag)
	{ hidingWhitespace = flag; }


    /**
     * Returns true if the consumer is saving CDATA boundaries, or
     * false (the default) otherwise.
     *
     * @see #setHidingCDATA
     */
    final public boolean	isHidingCDATA ()
	{ return hidingCDATA; }

    /**
     * Controls whether the consumer will save CDATA boundaries.
     *
     * @see #isHidingCDATA
     * @param flag True to treat CDATA text differently from other
     *	text nodes
     */
    final public void		setHidingCDATA (boolean flag)
	{ hidingCDATA = flag; }
    


    /** Returns the document handler being used. */
    final public ContentHandler getContentHandler ()
	{ return handler; }

    /** Returns the DTD handler being used. */
    final public DTDHandler getDTDHandler ()
	{ return handler; }

    /**
     * Returns the lexical handler being used.
     * (DOM construction can't really use declaration handlers.)
     */
    final public Object getProperty (String id)
    throws SAXNotRecognizedException
    {
	if ("http://xml.org/sax/properties/lexical-handler".equals (id))
	    return handler;
	if ("http://xml.org/sax/properties/declaration-handler".equals (id))
	    return handler;
	throw new SAXNotRecognizedException (id);
    }

    EventConsumer getNext () { return next; }

    ErrorHandler getErrorHandler () { return errHandler; }

    /**
     * Class used to intercept various parsing events and use them to
     * populate a DOM document.  Subclasses would typically know and use
     * backdoors into specific DOM implementations, used to implement 
     * DTD-related functionality.
     *
     * <p> Note that if this ever throws a DOMException (runtime exception)
     * that will indicate a bug in the DOM (e.g. doesn't support something
     * per specification) or the parser (e.g. emitted an illegal name, or
     * accepted illegal input data). </p>
     */
    public static class Handler
	implements ContentHandler, LexicalHandler,
	    DTDHandler, DeclHandler
    {
	protected DomConsumer		consumer;

	private DOMImplementation	impl;
	private Document 		document;
	private boolean		isL2;

	private Locator		locator;
	private Node		top;
	private boolean		inCDATA;
	private boolean		mergeCDATA;
	private boolean		inDTD;
	private String		currentEntity;

	private boolean		recreatedAttrs;
	private AttributesImpl	attributes = new AttributesImpl ();

	/**
	 * Subclasses may use SAX2 events to provide additional
	 * behaviors in the resulting DOM.
	 */
	protected Handler (DomConsumer consumer)
	throws SAXException
	{
	    this.consumer = consumer;
	    document = consumer.emptyDocument ();
	    impl = document.getImplementation ();
	    isL2 = impl.hasFeature ("XML", "2.0");
	}

	private void fatal (String message, Exception x)
	throws SAXException
	{
	    SAXParseException	e;
	    ErrorHandler	errHandler = consumer.getErrorHandler ();;

	    if (locator == null)
		e = new SAXParseException (message, null, null, -1, -1, x);
	    else
		e = new SAXParseException (message, locator, x);
	    if (errHandler != null)
		errHandler.fatalError (e);
	    throw e;
	}

	/**
	 * Returns and forgets the document produced.  If the handler is
	 * reused, a new document may be created.
	 */
	Document clearDocument ()
	{
	    Document retval = document;
	    document = null;
	    locator = null;
	    return retval;
	}

	/**
	 * Returns the document under construction.
	 */
	protected Document getDocument ()
	    { return document; }
	
	/**
	 * Returns the current node being populated.  This is usually
	 * an Element or Document, but it might be an EntityReference
	 * node if some implementation-specific code knows how to put
	 * those into the result tree and later mark them as readonly.
	 */
	protected Node getTop ()
	    { return top; }


	// SAX1
	public void setDocumentLocator (Locator locator)
	{
	    this.locator = locator;
	}

	// SAX1
	public void startDocument ()
	throws SAXException
	{
	    if (document == null)
		try {
		    if (isL2) {
			// couple to original implementation
			document = impl.createDocument (null, "foo", null);
			document.removeChild (document.getFirstChild ());
		    } else {
			document = consumer.emptyDocument ();
		    }
		} catch (Exception e) {
		    fatal ("DOM create document", e);
		}
	    top = document;
	}

	// SAX1
	public void endDocument ()
	throws SAXException
	{
	    try {
		if (consumer.getNext () != null && document != null) {
		    DomParser	parser = new DomParser (document);

		    EventFilter.bind (parser, consumer.getNext ());
		    parser.parse ("ignored");
		}
	    } finally {
		top = null;
	    }
	}

	// SAX1
	public void processingInstruction (String target, String data)
	throws SAXException
	{
	    // we can't create populated entity ref nodes using
	    // only public DOM APIs (they've got to be readonly)
	    if (currentEntity != null)
		return;

	    ProcessingInstruction	pi;

	    if (isL2
		    // && consumer.isUsingNamespaces ()
		    && target.indexOf (':') != -1)
		namespaceError (
		    "PI target name is namespace nonconformant: "
			+ target);
	    if (inDTD)
		return;
	    pi = document.createProcessingInstruction (target, data);
	    top.appendChild (pi);
	}

	/**
	 * Subclasses may overrride this method to provide a more efficient
	 * way to construct text nodes.
	 * Typically, copying the text into a single character array will
	 * be more efficient than doing that as well as allocating other
	 * needed for a String, including an internal StringBuffer.
	 * Those additional memory and CPU costs can be incurred later,
	 * if ever needed.
	 * Unfortunately the standard DOM factory APIs encourage those costs
	 * to be incurred early.
	 */
	protected Text createText (
	    boolean	isCDATA,
	    char	ch [],
	    int		start,
	    int		length
	) {
	    String	value = new String (ch, start, length);

	    if (isCDATA)
		return document.createCDATASection (value);
	    else
		return document.createTextNode (value);
	}

	// SAX1