gecko-dev/content/base/src/nsXMLContentSerializer.cpp

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* ***** BEGIN LICENSE BLOCK *****
 * Version: MPL 1.1/GPL 2.0/LGPL 2.1
 *
 * The contents of this file are subject to the Mozilla Public License Version
 * 1.1 (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.mozilla.org/MPL/
 *
 * Software distributed under the License is distributed on an "AS IS" basis,
 * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
 * for the specific language governing rights and limitations under the
 * License.
 *
 * The Original Code is mozilla.org code.
 *
 * The Initial Developer of the Original Code is
 * Netscape Communications Corporation.
 * Portions created by the Initial Developer are Copyright (C) 1998
 * the Initial Developer. All Rights Reserved.
 *
 * Contributor(s):
 *   Laurent Jouanneau <laurent.jouanneau@disruptive-innovations.com>
 *
 * Alternatively, the contents of this file may be used under the terms of
 * either of the GNU General Public License Version 2 or later (the "GPL"),
 * or the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
 * in which case the provisions of the GPL or the LGPL are applicable instead
 * of those above. If you wish to allow use of your version of this file only
 * under the terms of either the GPL or the LGPL, and not to allow others to
 * use your version of this file under the terms of the MPL, indicate your
 * decision by deleting the provisions above and replace them with the notice
 * and other provisions required by the GPL or the LGPL. If you do not delete
 * the provisions above, a recipient may use your version of this file under
 * the terms of any one of the MPL, the GPL or the LGPL.
 *
 * ***** END LICENSE BLOCK ***** */

/*
 * nsIContentSerializer implementation that can be used with an
 * nsIDocumentEncoder to convert an XML DOM to an XML string that
 * could be parsed into more or less the original DOM.
 */

#include "nsXMLContentSerializer.h"

#include "nsGkAtoms.h"
#include "nsIDOMText.h"
#include "nsIDOMCDATASection.h"
#include "nsIDOMProcessingInstruction.h"
#include "nsIDOMComment.h"
#include "nsIDOMDocument.h"
#include "nsIDOMDocumentType.h"
#include "nsIDOMElement.h"
#include "nsIContent.h"
#include "nsIDocument.h"
#include "nsIDocumentEncoder.h"
#include "nsINameSpaceManager.h"
#include "nsTextFragment.h"
#include "nsString.h"
#include "prprf.h"
#include "nsUnicharUtils.h"
#include "nsCRT.h"
#include "nsContentUtils.h"
#include "nsAttrName.h"

nsresult NS_NewXMLContentSerializer(nsIContentSerializer** aSerializer)
{
  nsXMLContentSerializer* it = new nsXMLContentSerializer();
  if (!it) {
    return NS_ERROR_OUT_OF_MEMORY;
  }

  return CallQueryInterface(it, aSerializer);
}

nsXMLContentSerializer::nsXMLContentSerializer()
  : mPrefixIndex(0),
    mColPos(0),
    mInAttribute(PR_FALSE),
    mAddNewline(PR_FALSE)
{
}
 
nsXMLContentSerializer::~nsXMLContentSerializer()
{
}

NS_IMPL_ISUPPORTS1(nsXMLContentSerializer, nsIContentSerializer)

NS_IMETHODIMP 
nsXMLContentSerializer::Init(PRUint32 aFlags, PRUint32 aWrapColumn,
                             const char* aCharSet, PRBool aIsCopying,
                             PRBool aIsWholeDocument)
{
  mCharset = aCharSet;
  mFlags = aFlags;

  // Set the line break character:
  if ((mFlags & nsIDocumentEncoder::OutputCRLineBreak)
      && (mFlags & nsIDocumentEncoder::OutputLFLineBreak)) { // Windows
    mLineBreak.AssignLiteral("\r\n");
  }
  else if (mFlags & nsIDocumentEncoder::OutputCRLineBreak) { // Mac
    mLineBreak.AssignLiteral("\r");
  }
  else if (mFlags & nsIDocumentEncoder::OutputLFLineBreak) { // Unix/DOM
    mLineBreak.AssignLiteral("\n");
  }
  else {
    mLineBreak.AssignLiteral(NS_LINEBREAK);         // Platform/default
  }
  return NS_OK;
}

nsresult
nsXMLContentSerializer::AppendTextData(nsIDOMNode* aNode, 
                                       PRInt32 aStartOffset,
                                       PRInt32 aEndOffset,
                                       nsAString& aStr,
                                       PRBool aTranslateEntities,
                                       PRBool aIncrColumn)
{
  nsCOMPtr<nsIContent> content = do_QueryInterface(aNode);
  const nsTextFragment* frag;
  if (!content || !(frag = content->GetText())) {
    return NS_ERROR_FAILURE;
  }

  PRInt32 endoffset = (aEndOffset == -1) ? frag->GetLength() : aEndOffset;
  PRInt32 length = endoffset - aStartOffset;

  NS_ASSERTION(aStartOffset >= 0, "Negative start offset for text fragment!");
  NS_ASSERTION(aStartOffset <= endoffset, "A start offset is beyond the end of the text fragment!");

  if (length <= 0) {
    // XXX Zero is a legal value, maybe non-zero values should be an
    // error.
    return NS_OK;
  }
    
  if (frag->Is2b()) {
    const PRUnichar *strStart = frag->Get2b() + aStartOffset;
    AppendToString(Substring(strStart, strStart + length), aStr,
                   aTranslateEntities, aIncrColumn);
  }
  else {
    AppendToString(NS_ConvertASCIItoUTF16(frag->Get1b()+aStartOffset, length),
                   aStr, aTranslateEntities, aIncrColumn);
  }

  return NS_OK;
}

NS_IMETHODIMP 
nsXMLContentSerializer::AppendText(nsIDOMText* aText, 
                                   PRInt32 aStartOffset,
                                   PRInt32 aEndOffset,
                                   nsAString& aStr)
{
  NS_ENSURE_ARG(aText);

  nsAutoString data;
  nsresult rv;

  rv = AppendTextData(aText, aStartOffset, aEndOffset, data, PR_TRUE, PR_TRUE);
  if (NS_FAILED(rv))
    return NS_ERROR_FAILURE;

  AppendToStringConvertLF(data, aStr);

  return NS_OK;
}

NS_IMETHODIMP 
nsXMLContentSerializer::AppendCDATASection(nsIDOMCDATASection* aCDATASection,
                                           PRInt32 aStartOffset,
                                           PRInt32 aEndOffset,
                                           nsAString& aStr)
{
  NS_ENSURE_ARG(aCDATASection);
  nsresult rv;

  AppendToString(NS_LITERAL_STRING("<![CDATA["), aStr);
  rv = AppendTextData(aCDATASection, aStartOffset, aEndOffset, aStr, PR_FALSE, PR_TRUE);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;  
  AppendToString(NS_LITERAL_STRING("]]>"), aStr);

  return NS_OK;
}

NS_IMETHODIMP 
nsXMLContentSerializer::AppendProcessingInstruction(nsIDOMProcessingInstruction* aPI,
                                                    PRInt32 aStartOffset,
                                                    PRInt32 aEndOffset,
                                                    nsAString& aStr)
{
  NS_ENSURE_ARG(aPI);
  nsresult rv;
  nsAutoString target, data;

  MaybeAddNewline(aStr);

  rv = aPI->GetTarget(target);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;

  rv = aPI->GetData(data);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;

  AppendToString(NS_LITERAL_STRING("<?"), aStr);
  AppendToString(target, aStr);
  if (!data.IsEmpty()) {
    AppendToString(NS_LITERAL_STRING(" "), aStr);
    AppendToStringConvertLF(data, aStr);
  }
  AppendToString(NS_LITERAL_STRING("?>"), aStr);
  MaybeFlagNewline(aPI);
  
  return NS_OK;
}

NS_IMETHODIMP 
nsXMLContentSerializer::AppendComment(nsIDOMComment* aComment,
                                      PRInt32 aStartOffset,
                                      PRInt32 aEndOffset,
                                      nsAString& aStr)
{
  NS_ENSURE_ARG(aComment);
  nsresult rv;
  nsAutoString data;

  rv = aComment->GetData(data);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;

  MaybeAddNewline(aStr);

  AppendToString(NS_LITERAL_STRING("<!--"), aStr);
  if (aStartOffset || (aEndOffset != -1)) {
    PRInt32 length = (aEndOffset == -1) ? data.Length() : aEndOffset;
    length -= aStartOffset;

    nsAutoString frag;
    data.Mid(frag, aStartOffset, length);
    AppendToStringConvertLF(frag, aStr);
  }
  else {
    AppendToStringConvertLF(data, aStr);
  }
  AppendToString(NS_LITERAL_STRING("-->"), aStr);
  MaybeFlagNewline(aComment);
  
  return NS_OK;
}

NS_IMETHODIMP 
nsXMLContentSerializer::AppendDoctype(nsIDOMDocumentType *aDoctype,
                                      nsAString& aStr)
{
  NS_ENSURE_ARG(aDoctype);
  nsresult rv;
  nsAutoString name, publicId, systemId, internalSubset;

  rv = aDoctype->GetName(name);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;
  rv = aDoctype->GetPublicId(publicId);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;
  rv = aDoctype->GetSystemId(systemId);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;
  rv = aDoctype->GetInternalSubset(internalSubset);
  if (NS_FAILED(rv)) return NS_ERROR_FAILURE;

  MaybeAddNewline(aStr);

  AppendToString(NS_LITERAL_STRING("<!DOCTYPE "), aStr);
  AppendToString(name, aStr);
  PRUnichar quote;
  if (!publicId.IsEmpty()) {
    AppendToString(NS_LITERAL_STRING(" PUBLIC "), aStr);
    if (publicId.FindChar(PRUnichar('"')) == -1) {
      quote = PRUnichar('"');
    }
    else {
      quote = PRUnichar('\'');
    }
    AppendToString(quote, aStr);
    AppendToString(publicId, aStr);
    AppendToString(quote, aStr);

    if (!systemId.IsEmpty()) {
      AppendToString(PRUnichar(' '), aStr);
      if (systemId.FindChar(PRUnichar('"')) == -1) {
        quote = PRUnichar('"');
      }
      else {
        quote = PRUnichar('\'');
      }
      AppendToString(quote, aStr);
      AppendToString(systemId, aStr);
      AppendToString(quote, aStr);
    }
  }
  else if (!systemId.IsEmpty()) {
    if (systemId.FindChar(PRUnichar('"')) == -1) {
      quote = PRUnichar('"');
    }
    else {
      quote = PRUnichar('\'');
    }
    AppendToString(NS_LITERAL_STRING(" SYSTEM "), aStr);
    AppendToString(quote, aStr);
    AppendToString(systemId, aStr);
    AppendToString(quote, aStr);
  }
  
  if (!internalSubset.IsEmpty()) {
    AppendToString(NS_LITERAL_STRING(" ["), aStr);
    AppendToString(internalSubset, aStr);
    AppendToString(PRUnichar(']'), aStr);
  }
    
  AppendToString(PRUnichar('>'), aStr);
  MaybeFlagNewline(aDoctype);

  return NS_OK;
}

#define kXMLNS "xmlns"

nsresult
nsXMLContentSerializer::PushNameSpaceDecl(const nsAString& aPrefix,
                                          const nsAString& aURI,
                                          nsIDOMElement* aOwner)
{
  NameSpaceDecl* decl = mNameSpaceStack.AppendElement();
  if (!decl) return NS_ERROR_OUT_OF_MEMORY;

  decl->mPrefix.Assign(aPrefix);
  decl->mURI.Assign(aURI);
  // Don't addref - this weak reference will be removed when
  // we pop the stack
  decl->mOwner = aOwner;
  return NS_OK;
}

void
nsXMLContentSerializer::PopNameSpaceDeclsFor(nsIDOMElement* aOwner)
{
  PRInt32 index, count;

  count = mNameSpaceStack.Length();
  for (index = count - 1; index >= 0; index--) {
    if (mNameSpaceStack[index].mOwner != aOwner) {
      break;
    }
    mNameSpaceStack.RemoveElementAt(index);
  }
}

PRBool
nsXMLContentSerializer::ConfirmPrefix(nsAString& aPrefix,
                                      const nsAString& aURI,
                                      nsIDOMElement* aElement,
                                      PRBool aIsAttribute)
{
  if (aPrefix.EqualsLiteral(kXMLNS)) {
    return PR_FALSE;
  }

  if (aURI.EqualsLiteral("http://www.w3.org/XML/1998/namespace")) {
    // The prefix must be xml for this namespace. We don't need to declare it,
    // so always just set the prefix to xml.
    aPrefix.AssignLiteral("xml");

    return PR_FALSE;
  }

  PRBool mustHavePrefix;
  if (aIsAttribute) {
    if (aURI.IsEmpty()) {
      // Attribute in the null namespace.  This just shouldn't have a prefix.
      // And there's no need to push any namespace decls
      aPrefix.Truncate();
      return PR_FALSE;
    }

    // Attribute not in the null namespace -- must have a prefix
    mustHavePrefix = PR_TRUE;
  } else {
    // Not an attribute, so doesn't _have_ to have a prefix
    mustHavePrefix = PR_FALSE;
  }

  // Keep track of the closest prefix that's bound to aURI and whether we've
  // found such a thing.  closestURIMatch holds the prefix, and uriMatch
  // indicates whether we actually have one.
  nsAutoString closestURIMatch;
  PRBool uriMatch = PR_FALSE;

  // Also keep track of whether we've seen aPrefix already.  If we have, that
  // means that it's already bound to a URI different from aURI, so even if we
  // later (so in a more outer scope) see it bound to aURI we can't reuse it.
  PRBool haveSeenOurPrefix = PR_FALSE;

  PRInt32 count = mNameSpaceStack.Length();
  PRInt32 index = count - 1;
  while (index >= 0) {
    NameSpaceDecl& decl = mNameSpaceStack.ElementAt(index);
    // Check if we've found a prefix match
    if (aPrefix.Equals(decl.mPrefix)) {

      // If the URIs match and aPrefix is not bound to any other URI, we can
      // use aPrefix
      if (!haveSeenOurPrefix && aURI.Equals(decl.mURI)) {
        // Just use our uriMatch stuff.  That will deal with an empty aPrefix
        // the right way.  We can break out of the loop now, though.
        uriMatch = PR_TRUE;
        closestURIMatch = aPrefix;
        break;
      }

      haveSeenOurPrefix = PR_TRUE;      

      // If they don't, and either:
      // 1) We have a prefix (so we'd be redeclaring this prefix to point to a
      //    different namespace) or
      // 2) We're looking at an existing default namespace decl on aElement (so
      //    we can't create a new default namespace decl for this URI)
      // then generate a new prefix.  Note that we do NOT generate new prefixes
      // if we happen to have aPrefix == decl->mPrefix == "" and mismatching
      // URIs when |decl| doesn't have aElement as its owner.  In that case we
      // can simply push the new namespace URI as the default namespace for
      // aElement.
      if (!aPrefix.IsEmpty() || decl.mOwner == aElement) {
        NS_ASSERTION(!aURI.IsEmpty(),
                     "Not allowed to add a xmlns attribute with an empty "
                     "namespace name unless it declares the default "
                     "namespace.");

        GenerateNewPrefix(aPrefix);
        // Now we need to validate our new prefix/uri combination; check it
        // against the full namespace stack again.  Note that just restarting
        // the while loop is ok, since we haven't changed aURI, so the
        // closestURIMatch and uriMatch state is not affected.
        index = count - 1;
        haveSeenOurPrefix = PR_FALSE;
        continue;
      }
    }
    
    // If we've found a URI match, then record the first one
    if (!uriMatch && aURI.Equals(decl.mURI)) {
      // Need to check that decl->mPrefix is not declared anywhere closer to
      // us.  If it is, we can't use it.
      PRBool prefixOK = PR_TRUE;
      PRInt32 index2;
      for (index2 = count-1; index2 > index && prefixOK; --index2) {
        prefixOK = (mNameSpaceStack[index2].mPrefix != decl.mPrefix);
      }
      
      if (prefixOK) {
        uriMatch = PR_TRUE;
        closestURIMatch.Assign(decl.mPrefix);
      }
    }
    
    --index;
  }

  // At this point the following invariants hold:
  // 1) The prefix in closestURIMatch is mapped to aURI in our scope if
  //    uriMatch is set.
  // 2) There is nothing on the namespace stack that has aPrefix as the prefix
  //    and a _different_ URI, except for the case aPrefix.IsEmpty (and
  //    possible default namespaces on ancestors)
  
  // So if uriMatch is set it's OK to use the closestURIMatch prefix.  The one
  // exception is when closestURIMatch is actually empty (default namespace
  // decl) and we must have a prefix.
  if (uriMatch && (!mustHavePrefix || !closestURIMatch.IsEmpty())) {
    aPrefix.Assign(closestURIMatch);
    return PR_FALSE;
  }
  
  if (aPrefix.IsEmpty()) {
    // At this point, aPrefix is empty (which means we never had a prefix to
    // start with).  If we must have a prefix, just generate a new prefix and
    // then send it back through the namespace stack checks to make sure it's
    // OK.
    if (mustHavePrefix) {
      GenerateNewPrefix(aPrefix);
      return ConfirmPrefix(aPrefix, aURI, aElement, aIsAttribute);
    }

    // One final special case.  If aPrefix is empty and we never saw an empty
    // prefix (default namespace decl) on the namespace stack and we're in the
    // null namespace there is no reason to output an |xmlns=""| here.  It just
    // makes the output less readable.
    if (!haveSeenOurPrefix && aURI.IsEmpty()) {
      return PR_FALSE;
    }
  }

  // Now just set aURI as the new default namespace URI.  Indicate that we need
  // to create a namespace decl for the final prefix
  return PR_TRUE;
}

void
nsXMLContentSerializer::GenerateNewPrefix(nsAString& aPrefix)
{
  aPrefix.AssignLiteral("a");
  char buf[128];
  PR_snprintf(buf, sizeof(buf), "%d", mPrefixIndex++);
  AppendASCIItoUTF16(buf, aPrefix);
}

void
nsXMLContentSerializer::SerializeAttr(const nsAString& aPrefix,
                                      const nsAString& aName,
                                      const nsAString& aValue,
                                      nsAString& aStr,
                                      PRBool aDoEscapeEntities)
{
  AppendToString(PRUnichar(' '), aStr);
  if (!aPrefix.IsEmpty()) {
    AppendToString(aPrefix, aStr);
    AppendToString(PRUnichar(':'), aStr);
  }
  AppendToString(aName, aStr);
  
  if ( aDoEscapeEntities ) {
    // if problem characters are turned into character entity references
    // then there will be no problem with the value delimiter characters
    AppendToString(NS_LITERAL_STRING("=\""), aStr);

    mInAttribute = PR_TRUE;
    AppendToString(aValue, aStr, PR_TRUE);
    mInAttribute = PR_FALSE;

    AppendToString(PRUnichar('"'), aStr);
  }
  else {
    // Depending on whether the attribute value contains quotes or apostrophes we
    // need to select the delimiter character and escape characters using
    // character entity references, ignoring the value of aDoEscapeEntities.
    // See http://www.w3.org/TR/REC-html40/appendix/notes.html#h-B.3.2.2 for
    // the standard on character entity references in values.  We also have to
    // make sure to escape any '&' characters.
    
    PRBool bIncludesSingle = PR_FALSE;
    PRBool bIncludesDouble = PR_FALSE;
    nsAString::const_iterator iCurr, iEnd;
    PRUint32 uiSize, i;
    aValue.BeginReading(iCurr);
    aValue.EndReading(iEnd);
    for ( ; iCurr != iEnd; iCurr.advance(uiSize) ) {
      const PRUnichar * buf = iCurr.get();
      uiSize = iCurr.size_forward();
      for ( i = 0; i < uiSize; i++, buf++ ) {
        if ( *buf == PRUnichar('\'') )
        {
          bIncludesSingle = PR_TRUE;
          if ( bIncludesDouble ) break;
        }
        else if ( *buf == PRUnichar('"') )
        {
          bIncludesDouble = PR_TRUE;
          if ( bIncludesSingle ) break;
        }
      }
      // if both have been found we don't need to search further
      if ( bIncludesDouble && bIncludesSingle ) break;
    }

    // Delimiter and escaping is according to the following table
    //    bIncludesDouble     bIncludesSingle     Delimiter       Escape Double Quote
    //    FALSE               FALSE               "               FALSE
    //    FALSE               TRUE                "               FALSE
    //    TRUE                FALSE               '               FALSE
    //    TRUE                TRUE                "               TRUE
    PRUnichar cDelimiter = 
        (bIncludesDouble && !bIncludesSingle) ? PRUnichar('\'') : PRUnichar('"');
    AppendToString(PRUnichar('='), aStr);
    AppendToString(cDelimiter, aStr);
    nsAutoString sValue(aValue);
    sValue.ReplaceSubstring(NS_LITERAL_STRING("&"),
                            NS_LITERAL_STRING("&amp;"));
    if (bIncludesDouble && bIncludesSingle) {
      sValue.ReplaceSubstring(NS_LITERAL_STRING("\""),
                              NS_LITERAL_STRING("&quot;"));
    }
    mInAttribute = PR_TRUE;
    AppendToString(sValue, aStr, PR_FALSE);
    mInAttribute = PR_FALSE;
    AppendToString(cDelimiter, aStr);
  }
}

NS_IMETHODIMP 
nsXMLContentSerializer::AppendElementStart(nsIDOMElement *aElement,
                                           nsIDOMElement *aOriginalElement,
                                           nsAString& aStr)
{
  NS_ENSURE_ARG(aElement);

  nsAutoString tagPrefix, tagLocalName, tagNamespaceURI;
  nsAutoString xmlnsStr;
  xmlnsStr.AssignLiteral(kXMLNS);

  nsCOMPtr<nsIContent> content(do_QueryInterface(aElement));
  if (!content) return NS_ERROR_FAILURE;

  aElement->GetPrefix(tagPrefix);
  aElement->GetLocalName(tagLocalName);
  aElement->GetNamespaceURI(tagNamespaceURI);

  PRUint32 index, count;
  nsAutoString nameStr, prefixStr, uriStr, valueStr;

  count = content->GetAttrCount();

  // First scan for namespace declarations, pushing each on the stack
  PRUint32 skipAttr = count;
  for (index = 0; index < count; index++) {
    
    const nsAttrName* name = content->GetAttrNameAt(index);
    PRInt32 namespaceID = name->NamespaceID();
    nsIAtom *attrName = name->LocalName();
    
    if (namespaceID == kNameSpaceID_XMLNS ||
        // Also push on the stack attrs named "xmlns" in the null
        // namespace... because once we serialize those out they'll look like
        // namespace decls.  :(
        // XXXbz what if we have both "xmlns" in the null namespace and "xmlns"
        // in the xmlns namespace?
        (namespaceID == kNameSpaceID_None &&
         attrName == nsGkAtoms::xmlns)) {
      content->GetAttr(namespaceID, attrName, uriStr);

      if (!name->GetPrefix()) {
        if (tagNamespaceURI.IsEmpty() && !uriStr.IsEmpty()) {
          // If the element is in no namespace we need to add a xmlns
          // attribute to declare that. That xmlns attribute must not have a
          // prefix (see http://www.w3.org/TR/REC-xml-names/#dt-prefix), ie it
          // must declare the default namespace. We just found an xmlns
          // attribute that declares the default namespace to something
          // non-empty. We're going to ignore this attribute, for children we
          // will detect that we need to add it again and attributes aren't
          // affected by the default namespace.
          skipAttr = index;
        }
        else {
          // Default NS attribute does not have prefix (and the name is "xmlns")
          PushNameSpaceDecl(EmptyString(), uriStr, aOriginalElement);
        }
      }
      else {
        attrName->ToString(nameStr);
        PushNameSpaceDecl(nameStr, uriStr, aOriginalElement);
      }
    }
  }

  PRBool addNSAttr;
    
  MaybeAddNewline(aStr);

  addNSAttr = ConfirmPrefix(tagPrefix, tagNamespaceURI, aOriginalElement,
                            PR_FALSE);
  // Serialize the qualified name of the element
  AppendToString(NS_LITERAL_STRING("<"), aStr);
  if (!tagPrefix.IsEmpty()) {
    AppendToString(tagPrefix, aStr);
    AppendToString(NS_LITERAL_STRING(":"), aStr);
  }
  AppendToString(tagLocalName, aStr);
    
  // If we had to add a new namespace declaration, serialize
  // and push it on the namespace stack
  if (addNSAttr) {
    if (tagPrefix.IsEmpty()) {
      // Serialize default namespace decl
      SerializeAttr(EmptyString(), xmlnsStr, tagNamespaceURI, aStr, PR_TRUE);
    } else {
      // Serialize namespace decl
      SerializeAttr(xmlnsStr, tagPrefix, tagNamespaceURI, aStr, PR_TRUE);
    }
    PushNameSpaceDecl(tagPrefix, tagNamespaceURI, aOriginalElement);
  }

  // Now serialize each of the attributes
  // XXX Unfortunately we need a namespace manager to get
  // attribute URIs.
  for (index = 0; index < count; index++) {
    if (skipAttr == index) {
        continue;
    }

    const nsAttrName* name = content->GetAttrNameAt(index);
    PRInt32 namespaceID = name->NamespaceID();
    nsIAtom* attrName = name->LocalName();
    nsIAtom* attrPrefix = name->GetPrefix();

    if (attrPrefix) {
      attrPrefix->ToString(prefixStr);
    }
    else {
      prefixStr.Truncate();
    }

    addNSAttr = PR_FALSE;
    if (kNameSpaceID_XMLNS != namespaceID) {
      nsContentUtils::NameSpaceManager()->GetNameSpaceURI(namespaceID, uriStr);
      addNSAttr = ConfirmPrefix(prefixStr, uriStr, aOriginalElement, PR_TRUE);
    }
    
    content->GetAttr(namespaceID, attrName, valueStr);
    attrName->ToString(nameStr);

    // XXX Hack to get around the fact that MathML can add
    //     attributes starting with '-', which makes them
    //     invalid XML.
    if (!nameStr.IsEmpty() && nameStr.First() == '-')
      continue;

    if (namespaceID == kNameSpaceID_None) {
      if (content->GetNameSpaceID() == kNameSpaceID_XHTML) {
        if (IsShorthandAttr(attrName, content->Tag()) &&
            valueStr.IsEmpty()) {
          valueStr = nameStr;
        }
      }
    }
    SerializeAttr(prefixStr, nameStr, valueStr, aStr, PR_TRUE);
    
    if (addNSAttr) {
      NS_ASSERTION(!prefixStr.IsEmpty(),
                   "Namespaced attributes must have a prefix");
      SerializeAttr(xmlnsStr, prefixStr, uriStr, aStr, PR_TRUE);
      PushNameSpaceDecl(prefixStr, uriStr, aOriginalElement);
    }
  }

  // We don't output a separate end tag for empty element
  PRBool hasChildren;
  if (NS_FAILED(aOriginalElement->HasChildNodes(&hasChildren)) ||
      !hasChildren) {
    AppendToString(NS_LITERAL_STRING("/>"), aStr);    
    MaybeFlagNewline(aElement);
  } else {
    AppendToString(NS_LITERAL_STRING(">"), aStr);    
  }
  
  return NS_OK;
}

NS_IMETHODIMP 
nsXMLContentSerializer::AppendElementEnd(nsIDOMElement *aElement,
                                         nsAString& aStr)
{
  NS_ENSURE_ARG(aElement);

  // We don't output a separate end tag for empty element
  nsCOMPtr<nsIDOMNode> node(do_QueryInterface(aElement));
  PRBool hasChildren;
  if (NS_SUCCEEDED(node->HasChildNodes(&hasChildren)) && !hasChildren) {
    PopNameSpaceDeclsFor(aElement);
  
    return NS_OK;
  }
  
  nsCOMPtr<nsIContent> content(do_QueryInterface(aElement));
  if (!content) return NS_ERROR_FAILURE;

  nsAutoString tagPrefix, tagLocalName, tagNamespaceURI;
  
  aElement->GetPrefix(tagPrefix);
  aElement->GetLocalName(tagLocalName);
  aElement->GetNamespaceURI(tagNamespaceURI);

#ifdef DEBUG
  PRBool debugNeedToPushNamespace =
#endif
  ConfirmPrefix(tagPrefix, tagNamespaceURI, aElement, PR_FALSE);
  NS_ASSERTION(!debugNeedToPushNamespace, "Can't push namespaces in closing tag!");

  AppendToString(NS_LITERAL_STRING("</"), aStr);
  if (!tagPrefix.IsEmpty()) {
    AppendToString(tagPrefix, aStr);
    AppendToString(NS_LITERAL_STRING(":"), aStr);
  }
  AppendToString(tagLocalName, aStr);
  AppendToString(NS_LITERAL_STRING(">"), aStr);
  MaybeFlagNewline(aElement);
  
  PopNameSpaceDeclsFor(aElement);
  
  return NS_OK;
}

void
nsXMLContentSerializer::AppendToString(const PRUnichar* aStr,
                                       PRInt32 aLength,
                                       nsAString& aOutputStr)
{
  PRInt32 length = (aLength == -1) ? nsCRT::strlen(aStr) : aLength;
  
  aOutputStr.Append(aStr, length);
}

void 
nsXMLContentSerializer::AppendToString(const PRUnichar aChar,
                                       nsAString& aOutputStr)
{
  aOutputStr.Append(aChar);
}

static const PRUint16 kGTVal = 62;
static const char* kEntities[] = {
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "", "", "", "", "&amp;", "",
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "", "", "", "", "", "",
  "&lt;", "", "&gt;"
};

static const char* kAttrEntities[] = {
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "&quot;", "", "", "", "&amp;", "",
  "", "", "", "", "", "", "", "", "", "",
  "", "", "", "", "", "", "", "", "", "",
  "&lt;", "", "&gt;"
};

void
nsXMLContentSerializer::AppendToString(const nsAString& aStr,
                                       nsAString& aOutputStr,
                                       PRBool aTranslateEntities,
                                       PRBool aIncrColumn)
{
  if (aTranslateEntities) {
    nsReadingIterator<PRUnichar> done_reading;
    aStr.EndReading(done_reading);

    // for each chunk of |aString|...
    PRUint32 advanceLength = 0;
    nsReadingIterator<PRUnichar> iter;

    const char **entityTable = mInAttribute ? kAttrEntities : kEntities;

    for (aStr.BeginReading(iter); 
         iter != done_reading; 
         iter.advance(PRInt32(advanceLength))) {
      PRUint32 fragmentLength = iter.size_forward();
      const PRUnichar* c = iter.get();
      const PRUnichar* fragmentStart = c;
      const PRUnichar* fragmentEnd = c + fragmentLength;
      const char* entityText = nsnull;

      advanceLength = 0;
      // for each character in this chunk, check if it
      // needs to be replaced
      for (; c < fragmentEnd; c++, advanceLength++) {
        PRUnichar val = *c;
        if ((val <= kGTVal) && (entityTable[val][0] != 0)) {
          entityText = entityTable[val];
          break;
        }
      }

      aOutputStr.Append(fragmentStart, advanceLength);
      if (entityText) {
        AppendASCIItoUTF16(entityText, aOutputStr);
        advanceLength++;
      }
    }

    return;
  }
  
  aOutputStr.Append(aStr);
}

PRBool
nsXMLContentSerializer::IsShorthandAttr(const nsIAtom* aAttrName,
                                        const nsIAtom* aElementName)
{
  // checked
  if ((aAttrName == nsGkAtoms::checked) &&
      (aElementName == nsGkAtoms::input)) {
    return PR_TRUE;
  }

  // compact
  if ((aAttrName == nsGkAtoms::compact) &&
      (aElementName == nsGkAtoms::dir || 
       aElementName == nsGkAtoms::dl ||
       aElementName == nsGkAtoms::menu ||
       aElementName == nsGkAtoms::ol ||
       aElementName == nsGkAtoms::ul)) {
    return PR_TRUE;
  }

  // declare
  if ((aAttrName == nsGkAtoms::declare) &&
      (aElementName == nsGkAtoms::object)) {
    return PR_TRUE;
  }

  // defer
  if ((aAttrName == nsGkAtoms::defer) &&
      (aElementName == nsGkAtoms::script)) {
    return PR_TRUE;
  }

  // disabled
  if ((aAttrName == nsGkAtoms::disabled) &&
      (aElementName == nsGkAtoms::button ||
       aElementName == nsGkAtoms::input ||
       aElementName == nsGkAtoms::optgroup ||
       aElementName == nsGkAtoms::option ||
       aElementName == nsGkAtoms::select ||
       aElementName == nsGkAtoms::textarea)) {
    return PR_TRUE;
  }

  // ismap
  if ((aAttrName == nsGkAtoms::ismap) &&
      (aElementName == nsGkAtoms::img ||
       aElementName == nsGkAtoms::input)) {
    return PR_TRUE;
  }

  // multiple
  if ((aAttrName == nsGkAtoms::multiple) &&
      (aElementName == nsGkAtoms::select)) {
    return PR_TRUE;
  }

  // noresize
  if ((aAttrName == nsGkAtoms::noresize) &&
      (aElementName == nsGkAtoms::frame)) {
    return PR_TRUE;
  }

  // noshade
  if ((aAttrName == nsGkAtoms::noshade) &&
      (aElementName == nsGkAtoms::hr)) {
    return PR_TRUE;
  }

  // nowrap
  if ((aAttrName == nsGkAtoms::nowrap) &&
      (aElementName == nsGkAtoms::td ||
       aElementName == nsGkAtoms::th)) {
    return PR_TRUE;
  }

  // readonly
  if ((aAttrName == nsGkAtoms::readonly) &&
      (aElementName == nsGkAtoms::input ||
       aElementName == nsGkAtoms::textarea)) {
    return PR_TRUE;
  }

  // selected
  if ((aAttrName == nsGkAtoms::selected) &&
      (aElementName == nsGkAtoms::option)) {
    return PR_TRUE;
  }

  return PR_FALSE;
}

void
nsXMLContentSerializer::MaybeAddNewline(nsAString& aStr)
{
  if (mAddNewline) {
    aStr.Append(mLineBreak);
    mAddNewline = PR_FALSE;
  }
}

void
nsXMLContentSerializer::MaybeFlagNewline(nsIDOMNode* aNode)
{
  nsCOMPtr<nsIDOMNode> parent;
  aNode->GetParentNode(getter_AddRefs(parent));
  if (parent) {
    PRUint16 type;
    parent->GetNodeType(&type);
    mAddNewline = type == nsIDOMNode::DOCUMENT_NODE;
  }
}

NS_IMETHODIMP
nsXMLContentSerializer::AppendDocumentStart(nsIDOMDocument *aDocument,
                                            nsAString& aStr)
{
  NS_ENSURE_ARG_POINTER(aDocument);

  nsCOMPtr<nsIDocument> doc(do_QueryInterface(aDocument));
  if (!doc) {
    return NS_OK;
  }

  nsAutoString version, encoding, standalone;
  doc->GetXMLDeclaration(version, encoding, standalone);

  if (version.IsEmpty())
    return NS_OK; // A declaration must have version, or there is no decl

  NS_NAMED_LITERAL_STRING(endQuote, "\"");

  aStr += NS_LITERAL_STRING("<?xml version=\"") + version + endQuote;
  
  if (!mCharset.IsEmpty()) {
    aStr += NS_LITERAL_STRING(" encoding=\"") +
      NS_ConvertASCIItoUTF16(mCharset) + endQuote;
  }
  // Otherwise just don't output an encoding attr.  Not that we expect
  // mCharset to ever be empty.
#ifdef DEBUG
  else {
    NS_WARNING("Empty mCharset?  How come?");
  }
#endif

  if (!standalone.IsEmpty()) {
    aStr += NS_LITERAL_STRING(" standalone=\"") + standalone + endQuote;
  }

  aStr.AppendLiteral("?>");
  mAddNewline = PR_TRUE;

  return NS_OK;
}

void
nsXMLContentSerializer::AppendToStringConvertLF(const nsAString& aStr,
                                                 nsAString& aOutputStr)
{
  // Convert line-endings to mLineBreak
  PRUint32 start = 0;
  PRUint32 theLen = aStr.Length();
  while (start < theLen) {
    PRInt32 eol = aStr.FindChar('\n', start);
    if (eol == kNotFound) {
      nsDependentSubstring dataSubstring(aStr, start, theLen - start);
      AppendToString(dataSubstring, aOutputStr);
      start = theLen;
    }
    else {
      nsDependentSubstring dataSubstring(aStr, start, eol - start);
      AppendToString(dataSubstring, aOutputStr);
      AppendToString(mLineBreak, aOutputStr);
      start = eol + 1;
      if (start == theLen)
        mColPos = 0;
    }
  }
}