domhelper.java

来自「JAVA 所有包」· Java 代码 · 共 1,333 行 · 第 1/3 页

/*
 * Copyright 1999-2004 The Apache Software Foundation.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
/*
 * $Id: DOMHelper.java,v 1.2.4.1 2005/09/15 08:15:40 suresh_emailid Exp $
 */
package com.sun.org.apache.xml.internal.utils;

import java.util.Hashtable;
import java.util.Vector;

import javax.xml.XMLConstants;
import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.parsers.ParserConfigurationException;

import com.sun.org.apache.xml.internal.dtm.ref.DTMNodeProxy;
import com.sun.org.apache.xml.internal.res.XMLErrorResources;
import com.sun.org.apache.xml.internal.res.XMLMessages;

import org.w3c.dom.Attr;
import org.w3c.dom.DOMImplementation;
import org.w3c.dom.Document;
import org.w3c.dom.DocumentType;
import org.w3c.dom.Element;
import org.w3c.dom.Entity;
import org.w3c.dom.NamedNodeMap;
import org.w3c.dom.Node;
import org.w3c.dom.Text;

/**
 * @deprecated Since the introduction of the DTM, this class will be removed.
 * This class provides a front-end to DOM implementations, providing
 * a number of utility functions that either aren't yet standardized
 * by the DOM spec or that are defined in optional DOM modules and
 * hence may not be present in all DOMs.
 */
public class DOMHelper
{

  /**
   * DOM Level 1 did not have a standard mechanism for creating a new
   * Document object. This function provides a DOM-implementation-independent
   * abstraction for that for that concept. It's typically used when 
   * outputting a new DOM as the result of an operation.
   * <p>
   * TODO: This isn't directly compatable with DOM Level 2. 
   * The Level 2 createDocument call also creates the root 
   * element, and thus requires that you know what that element will be
   * before creating the Document. We should think about whether we want
   * to change this code, and the callers, so we can use the DOM's own 
   * method. (It's also possible that DOM Level 3 may relax this
   * sequence, but you may give up some intelligence in the DOM by
   * doing so; the intent was that knowing the document type and root
   * element might let the DOM automatically switch to a specialized
   * subclass for particular kinds of documents.)
   *
   * @param isSecureProcessing state of the secure processing feature.
   * @return The newly created DOM Document object, with no children, or
   * null if we can't find a DOM implementation that permits creating
   * new empty Documents.
   */
  public static Document createDocument(boolean isSecureProcessing)
  {

    try
    {

      // Use an implementation of the JAVA API for XML Parsing 1.0 to
      // create a DOM Document node to contain the result.
      DocumentBuilderFactory dfactory = DocumentBuilderFactory.newInstance();

      dfactory.setNamespaceAware(true);
      dfactory.setValidating(true);
      
      if (isSecureProcessing)
      {
        try
        {
          dfactory.setFeature(XMLConstants.FEATURE_SECURE_PROCESSING, true);
        }
        catch (ParserConfigurationException pce) {}
      }
      
      DocumentBuilder docBuilder = dfactory.newDocumentBuilder();
      Document outNode = docBuilder.newDocument();

      return outNode;
    }
    catch (ParserConfigurationException pce)
    {
      throw new RuntimeException(
        XMLMessages.createXMLMessage(
          XMLErrorResources.ER_CREATEDOCUMENT_NOT_SUPPORTED, null));  //"createDocument() not supported in XPathContext!");

      // return null;
    }
  }

  /**
   * DOM Level 1 did not have a standard mechanism for creating a new
   * Document object. This function provides a DOM-implementation-independent
   * abstraction for that for that concept. It's typically used when 
   * outputting a new DOM as the result of an operation.
   *
   * @return The newly created DOM Document object, with no children, or
   * null if we can't find a DOM implementation that permits creating
   * new empty Documents.
   */
  public static Document createDocument()
  {
    return createDocument(false);
  }
  
  /**
   * Tells, through the combination of the default-space attribute
   * on xsl:stylesheet, xsl:strip-space, xsl:preserve-space, and the
   * xml:space attribute, whether or not extra whitespace should be stripped
   * from the node.  Literal elements from template elements should
   * <em>not</em> be tested with this function.
   * @param textNode A text node from the source tree.
   * @return true if the text node should be stripped of extra whitespace.
   *
   * @throws javax.xml.transform.TransformerException
   * @xsl.usage advanced
   */
  public boolean shouldStripSourceNode(Node textNode)
          throws javax.xml.transform.TransformerException
  {

    // return (null == m_envSupport) ? false : m_envSupport.shouldStripSourceNode(textNode);
    return false;
  }

  /**
   * Supports the XPath function GenerateID by returning a unique
   * identifier string for any given DOM Node.
   * <p>
   * Warning: The base implementation uses the Node object's hashCode(),
   * which is NOT guaranteed to be unique. If that method hasn't been
   * overridden in this DOM ipmlementation, most Java implementions will
   * derive it from the object's address and should be OK... but if
   * your DOM uses a different definition of hashCode (eg hashing the
   * contents of the subtree), or if your DOM may have multiple objects
   * that represent a single Node in the data structure (eg via proxying),
   * you may need to find another way to assign a unique identifier.
   * <p>
   * Also, be aware that if nodes are destroyed and recreated, there is
   * an open issue regarding whether an ID may be reused. Currently
   * we're assuming that the input document is stable for the duration
   * of the XPath/XSLT operation, so this shouldn't arise in this context.
   * <p>
   * (DOM Level 3 is investigating providing a unique node "key", but
   * that won't help Level 1 and Level 2 implementations.)
   *
   * @param node whose identifier you want to obtain
   *
   * @return a string which should be different for every Node object.
   */
  public String getUniqueID(Node node)
  {
    return "N" + Integer.toHexString(node.hashCode()).toUpperCase();
  }

  /**
   * Figure out whether node2 should be considered as being later
   * in the document than node1, in Document Order as defined
   * by the XPath model. This may not agree with the ordering defined
   * by other XML applications.
   * <p>
   * There are some cases where ordering isn't defined, and neither are
   * the results of this function -- though we'll generally return true.
   * 
   * TODO: Make sure this does the right thing with attribute nodes!!!
   *
   * @param node1 DOM Node to perform position comparison on.
   * @param node2 DOM Node to perform position comparison on .
   * 
   * @return false if node2 comes before node1, otherwise return true.
   * You can think of this as 
   * <code>(node1.documentOrderPosition &lt;= node2.documentOrderPosition)</code>.
   */
  public static boolean isNodeAfter(Node node1, Node node2)
  {
    if (node1 == node2 || isNodeTheSame(node1, node2))
      return true;

        // Default return value, if there is no defined ordering
    boolean isNodeAfter = true;
        
    Node parent1 = getParentOfNode(node1);
    Node parent2 = getParentOfNode(node2);          

    // Optimize for most common case
    if (parent1 == parent2 || isNodeTheSame(parent1, parent2))  // then we know they are siblings
    {
      if (null != parent1)
        isNodeAfter = isNodeAfterSibling(parent1, node1, node2);
      else
      {
                  // If both parents are null, ordering is not defined.
                  // We're returning a value in lieu of throwing an exception.
                  // Not a case we expect to arise in XPath, but beware if you
                  // try to reuse this method.
                  
                  // We can just fall through in this case, which allows us
                  // to hit the debugging code at the end of the function.
          //return isNodeAfter;
      }
    }
    else
    {

      // General strategy: Figure out the lengths of the two 
      // ancestor chains, reconcile the lengths, and look for
          // the lowest common ancestor. If that ancestor is one of
          // the nodes being compared, it comes before the other.
      // Otherwise perform a sibling compare. 
                //
                // NOTE: If no common ancestor is found, ordering is undefined
                // and we return the default value of isNodeAfter.
                
      // Count parents in each ancestor chain
      int nParents1 = 2, nParents2 = 2;  // include node & parent obtained above

      while (parent1 != null)
      {
        nParents1++;

        parent1 = getParentOfNode(parent1);
      }

      while (parent2 != null)
      {
        nParents2++;

        parent2 = getParentOfNode(parent2);
      }

          // Initially assume scan for common ancestor starts with
          // the input nodes.
      Node startNode1 = node1, startNode2 = node2;

      // If one ancestor chain is longer, adjust its start point
          // so we're comparing at the same depths
      if (nParents1 < nParents2)
      {
        // Adjust startNode2 to depth of startNode1
        int adjust = nParents2 - nParents1;

        for (int i = 0; i < adjust; i++)
        {
          startNode2 = getParentOfNode(startNode2);
        }
      }
      else if (nParents1 > nParents2)
      {
        // adjust startNode1 to depth of startNode2
        int adjust = nParents1 - nParents2;

        for (int i = 0; i < adjust; i++)
        {
          startNode1 = getParentOfNode(startNode1);
        }
      }

      Node prevChild1 = null, prevChild2 = null;  // so we can "back up"

      // Loop up the ancestor chain looking for common parent
      while (null != startNode1)
      {
        if (startNode1 == startNode2 || isNodeTheSame(startNode1, startNode2))  // common parent?
        {
          if (null == prevChild1)  // first time in loop?
          {

            // Edge condition: one is the ancestor of the other.
            isNodeAfter = (nParents1 < nParents2) ? true : false;

            break;  // from while loop
          }
          else 
          {
                        // Compare ancestors below lowest-common as siblings
            isNodeAfter = isNodeAfterSibling(startNode1, prevChild1,
                                             prevChild2);

            break;  // from while loop
          }
        }  // end if(startNode1 == startNode2)

                // Move up one level and try again
        prevChild1 = startNode1;
        startNode1 = getParentOfNode(startNode1);
        prevChild2 = startNode2;
        startNode2 = getParentOfNode(startNode2);
      }  // end while(parents exist to examine)
    }  // end big else (not immediate siblings)
        
        // WARNING: The following diagnostic won't report the early
        // "same node" case. Fix if/when needed.
        
    /* -- please do not remove... very useful for diagnostics --
    System.out.println("node1 = "+node1.getNodeName()+"("+node1.getNodeType()+")"+
    ", node2 = "+node2.getNodeName()
    +"("+node2.getNodeType()+")"+
    ", isNodeAfter = "+isNodeAfter); */
    return isNodeAfter;
  }  // end isNodeAfter(Node node1, Node node2)

  /**
   * Use DTMNodeProxy to determine whether two nodes are the same.
   * 
   * @param node1 The first DOM node to compare.
   * @param node2 The second DOM node to compare.
   * @return true if the two nodes are the same.
   */
  public static boolean isNodeTheSame(Node node1, Node node2)
  {
    if (node1 instanceof DTMNodeProxy && node2 instanceof DTMNodeProxy)
      return ((DTMNodeProxy)node1).equals((DTMNodeProxy)node2);
    else
      return (node1 == node2);
  }

  /**
   * Figure out if child2 is after child1 in document order.
   * <p>
   * Warning: Some aspects of "document order" are not well defined.
   * For example, the order of attributes is considered
   * meaningless in XML, and the order reported by our model will
   * be consistant for a given invocation but may not 
   * match that of either the source file or the serialized output.
   * 
   * @param parent Must be the parent of both child1 and child2.
   * @param child1 Must be the child of parent and not equal to child2.
   * @param child2 Must be the child of parent and not equal to child1.
   * @return true if child 2 is after child1 in document order.
   */
  private static boolean isNodeAfterSibling(Node parent, Node child1,
                                            Node child2)
  {

    boolean isNodeAfterSibling = false;
    short child1type = child1.getNodeType();
    short child2type = child2.getNodeType();

    if ((Node.ATTRIBUTE_NODE != child1type)
            && (Node.ATTRIBUTE_NODE == child2type))
    {

      // always sort attributes before non-attributes.
      isNodeAfterSibling = false;
    }
    else if ((Node.ATTRIBUTE_NODE == child1type)
             && (Node.ATTRIBUTE_NODE != child2type))
    {

      // always sort attributes before non-attributes.
      isNodeAfterSibling = true;
    }
    else if (Node.ATTRIBUTE_NODE == child1type)
    {
      NamedNodeMap children = parent.getAttributes();
      int nNodes = children.getLength();
      boolean found1 = false, found2 = false;

          // Count from the start until we find one or the other.
      for (int i = 0; i < nNodes; i++)
      {
        Node child = children.item(i);

        if (child1 == child || isNodeTheSame(child1, child))
        {
          if (found2)
          {
            isNodeAfterSibling = false;

            break;
          }

          found1 = true;
        }
        else if (child2 == child || isNodeTheSame(child2, child))
        {
          if (found1)
          {
            isNodeAfterSibling = true;

            break;
          }

          found2 = true;
        }
      }
    }
    else
    {
                // TODO: Check performance of alternate solution:
                // There are two choices here: Count from the start of
                // the document until we find one or the other, or count
                // from one until we find or fail to find the other.
                // Either can wind up scanning all the siblings in the worst
                // case, which on a wide document can be a lot of work but
                // is more typically is a short list. 
                // Scanning from the start involves two tests per iteration,
                // but it isn't clear that scanning from the middle doesn't
                // yield more iterations on average. 
                // We should run some testcases.
      Node child = parent.getFirstChild();
      boolean found1 = false, found2 = false;

      while (null != child)
      {

        // Node child = children.item(i);
        if (child1 == child || isNodeTheSame(child1, child))
        {
          if (found2)
          {
            isNodeAfterSibling = false;

            break;
          }

          found1 = true;
        }
        else if (child2 == child || isNodeTheSame(child2, child))
        {
          if (found1)
          {
            isNodeAfterSibling = true;