content/base/src/nsXHTMLContentSerializer.cpp
author Ehsan Akhgari <ehsan@mozilla.com>
Wed, 22 Aug 2012 11:56:38 -0400
changeset 102997 a16372ce30b5f6b747246b01fcd215a4bf3b6342
parent 100884 6e30103582c44ea1ee8c25747bd3bed6671c0011
child 106706 c4f83d9d8243f3f853a5356188164a5fddee2b5a
permissions -rw-r--r--
Bug 579517 - Part 1: Automated conversion of NSPR numeric types to stdint types in Gecko; r=bsmedberg This patch was generated by a script. Here's the source of the script for future reference: function convert() { echo "Converting $1 to $2..." find . ! -wholename "*nsprpub*" \ ! -wholename "*security/nss*" \ ! -wholename "*/.hg*" \ ! -wholename "obj-ff-dbg*" \ ! -name nsXPCOMCID.h \ ! -name prtypes.h \ -type f \ \( -iname "*.cpp" \ -o -iname "*.h" \ -o -iname "*.c" \ -o -iname "*.cc" \ -o -iname "*.idl" \ -o -iname "*.ipdl" \ -o -iname "*.ipdlh" \ -o -iname "*.mm" \) | \ xargs -n 1 sed -i -e "s/\b$1\b/$2/g" } convert PRInt8 int8_t convert PRUint8 uint8_t convert PRInt16 int16_t convert PRUint16 uint16_t convert PRInt32 int32_t convert PRUint32 uint32_t convert PRInt64 int64_t convert PRUint64 uint64_t convert PRIntn int convert PRUintn unsigned convert PRSize size_t convert PROffset32 int32_t convert PROffset64 int64_t convert PRPtrdiff ptrdiff_t convert PRFloat64 double

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim: set ts=2 sw=2 et tw=80: */
/* 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/. */

/*
 * nsIContentSerializer implementation that can be used with an
 * nsIDocumentEncoder to convert an XHTML (not HTML!) DOM to an XHTML
 * string that could be parsed into more or less the original DOM.
 */

#include "nsXHTMLContentSerializer.h"

#include "nsIDOMElement.h"
#include "nsIContent.h"
#include "nsIDocument.h"
#include "nsINameSpaceManager.h"
#include "nsString.h"
#include "nsUnicharUtils.h"
#include "nsXPIDLString.h"
#include "nsIServiceManager.h"
#include "nsIDocumentEncoder.h"
#include "nsGkAtoms.h"
#include "nsIURI.h"
#include "nsNetUtil.h"
#include "nsEscape.h"
#include "nsITextToSubURI.h"
#include "nsCRT.h"
#include "nsIParserService.h"
#include "nsContentUtils.h"
#include "nsLWBrkCIID.h"
#include "nsIScriptElement.h"
#include "nsAttrName.h"
#include "nsParserConstants.h"

static const char kMozStr[] = "moz";

static const int32_t kLongLineLen = 128;

#define kXMLNS "xmlns"

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

  return CallQueryInterface(it, aSerializer);
}

nsXHTMLContentSerializer::nsXHTMLContentSerializer()
  : mIsHTMLSerializer(false)
{
}

nsXHTMLContentSerializer::~nsXHTMLContentSerializer()
{
  NS_ASSERTION(mOLStateStack.IsEmpty(), "Expected OL State stack to be empty");
}

NS_IMETHODIMP
nsXHTMLContentSerializer::Init(uint32_t aFlags, uint32_t aWrapColumn,
                              const char* aCharSet, bool aIsCopying,
                              bool aRewriteEncodingDeclaration)
{
  // The previous version of the HTML serializer did implicit wrapping
  // when there is no flags, so we keep wrapping in order to keep
  // compatibility with the existing calling code
  // XXXLJ perhaps should we remove this default settings later ?
  if (aFlags & nsIDocumentEncoder::OutputFormatted ) {
      aFlags = aFlags | nsIDocumentEncoder::OutputWrap;
  }

  nsresult rv;
  rv = nsXMLContentSerializer::Init(aFlags, aWrapColumn, aCharSet, aIsCopying, aRewriteEncodingDeclaration);
  NS_ENSURE_SUCCESS(rv, rv);

  mRewriteEncodingDeclaration = aRewriteEncodingDeclaration;
  mIsCopying = aIsCopying;
  mIsFirstChildOfOL = false;
  mInBody = 0;
  mDisableEntityEncoding = 0;
  mBodyOnly = (mFlags & nsIDocumentEncoder::OutputBodyOnly) ? true
                                                            : false;

  // set up entity converter if we are going to need it
  if (mFlags & nsIDocumentEncoder::OutputEncodeW3CEntities) {
    mEntityConverter = do_CreateInstance(NS_ENTITYCONVERTER_CONTRACTID);
  }
  return NS_OK;
}


// See if the string has any lines longer than longLineLen:
// if so, we presume formatting is wonky (e.g. the node has been edited)
// and we'd better rewrap the whole text node.
bool
nsXHTMLContentSerializer::HasLongLines(const nsString& text, int32_t& aLastNewlineOffset)
{
  uint32_t start=0;
  uint32_t theLen = text.Length();
  bool rv = false;
  aLastNewlineOffset = kNotFound;
  for (start = 0; start < theLen; ) {
    int32_t eol = text.FindChar('\n', start);
    if (eol < 0) {
      eol = text.Length();
    }
    else {
      aLastNewlineOffset = eol;
    }
    if (int32_t(eol - start) > kLongLineLen)
      rv = true;
    start = eol + 1;
  }
  return rv;
}

NS_IMETHODIMP
nsXHTMLContentSerializer::AppendText(nsIContent* aText,
                                     int32_t aStartOffset,
                                     int32_t aEndOffset,
                                     nsAString& aStr)
{
  NS_ENSURE_ARG(aText);

  nsAutoString data;
  nsresult rv;

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

  if (mPreLevel > 0 || mDoRaw) {
    AppendToStringConvertLF(data, aStr);
  }
  else if (mDoFormat) {
    AppendToStringFormatedWrapped(data, aStr);
  }
  else if (mDoWrap) {
    AppendToStringWrapped(data, aStr);
  }
  else {
    int32_t lastNewlineOffset = kNotFound;
    if (HasLongLines(data, lastNewlineOffset)) {
      // We have long lines, rewrap
      mDoWrap = true;
      AppendToStringWrapped(data, aStr);
      mDoWrap = false;
    }
    else {
      AppendToStringConvertLF(data, aStr);
    }
  }

  return NS_OK;
}

nsresult
nsXHTMLContentSerializer::EscapeURI(nsIContent* aContent, const nsAString& aURI, nsAString& aEscapedURI)
{
  // URL escape %xx cannot be used in JS.
  // No escaping if the scheme is 'javascript'.
  if (IsJavaScript(aContent, nsGkAtoms::href, kNameSpaceID_None, aURI)) {
    aEscapedURI = aURI;
    return NS_OK;
  }

  // nsITextToSubURI does charset convert plus uri escape
  // This is needed to convert to a document charset which is needed to support existing browsers.
  // But we eventually want to use UTF-8 instead of a document charset, then the code would be much simpler.
  // See HTML 4.01 spec, "Appendix B.2.1 Non-ASCII characters in URI attribute values"
  nsCOMPtr<nsITextToSubURI> textToSubURI;
  nsAutoString uri(aURI); // in order to use FindCharInSet()
  nsresult rv = NS_OK;

  if (!mCharset.IsEmpty() && !IsASCII(uri)) {
    textToSubURI = do_GetService(NS_ITEXTTOSUBURI_CONTRACTID, &rv);
    NS_ENSURE_SUCCESS(rv, rv);
  }

  int32_t start = 0;
  int32_t end;
  nsAutoString part;
  nsXPIDLCString escapedURI;
  aEscapedURI.Truncate(0);

  // Loop and escape parts by avoiding escaping reserved characters
  // (and '%', '#', as well as '[' and ']' for IPv6 address literals).
  while ((end = uri.FindCharInSet("%#;/?:@&=+$,[]", start)) != -1) {
    part = Substring(aURI, start, (end-start));
    if (textToSubURI && !IsASCII(part)) {
      rv = textToSubURI->ConvertAndEscape(mCharset.get(), part.get(), getter_Copies(escapedURI));
      NS_ENSURE_SUCCESS(rv, rv);
    }
    else {
      escapedURI.Adopt(nsEscape(NS_ConvertUTF16toUTF8(part).get(), url_Path));
    }
    AppendASCIItoUTF16(escapedURI, aEscapedURI);

    // Append a reserved character without escaping.
    part = Substring(aURI, end, 1);
    aEscapedURI.Append(part);
    start = end + 1;
  }

  if (start < (int32_t) aURI.Length()) {
    // Escape the remaining part.
    part = Substring(aURI, start, aURI.Length()-start);
    if (textToSubURI) {
      rv = textToSubURI->ConvertAndEscape(mCharset.get(), part.get(), getter_Copies(escapedURI));
      NS_ENSURE_SUCCESS(rv, rv);
    }
    else {
      escapedURI.Adopt(nsEscape(NS_ConvertUTF16toUTF8(part).get(), url_Path));
    }
    AppendASCIItoUTF16(escapedURI, aEscapedURI);
  }

  return rv;
}

void
nsXHTMLContentSerializer::SerializeAttributes(nsIContent* aContent,
                                              nsIContent *aOriginalElement,
                                              nsAString& aTagPrefix,
                                              const nsAString& aTagNamespaceURI,
                                              nsIAtom* aTagName,
                                              nsAString& aStr,
                                              uint32_t aSkipAttr,
                                              bool aAddNSAttr)
{
  nsresult rv;
  uint32_t index, count;
  nsAutoString prefixStr, uriStr, valueStr;
  nsAutoString xmlnsStr;
  xmlnsStr.AssignLiteral(kXMLNS);

  int32_t contentNamespaceID = aContent->GetNameSpaceID();

  // this method is not called by nsHTMLContentSerializer
  // so we don't have to check HTML element, just XHTML

  if (mIsCopying && kNameSpaceID_XHTML == contentNamespaceID) {

    // Need to keep track of OL and LI elements in order to get ordinal number 
    // for the LI.
    if (aTagName == nsGkAtoms::ol) {
      // We are copying and current node is an OL;
      // Store its start attribute value in olState->startVal.
      nsAutoString start;
      int32_t startAttrVal = 0;
      aContent->GetAttr(kNameSpaceID_None, nsGkAtoms::start, start);
      if (!start.IsEmpty()) {
        nsresult rv = NS_OK;
        startAttrVal = start.ToInteger(&rv);
        //If OL has "start" attribute, first LI element has to start with that value
        //Therefore subtracting 1 as all the LI elements are incrementing it before using it;
        //In failure of ToInteger(), default StartAttrValue to 0.
        if (NS_SUCCEEDED(rv))
          --startAttrVal;
        else
          startAttrVal = 0;
      }
      olState state (startAttrVal, true);
      mOLStateStack.AppendElement(state);
    }
    else if (aTagName == nsGkAtoms::li) {
      mIsFirstChildOfOL = IsFirstChildOfOL(aOriginalElement);
      if (mIsFirstChildOfOL) {
        // If OL is parent of this LI, serialize attributes in different manner.
        SerializeLIValueAttribute(aContent, aStr);
      }
    }
  }

  // If we had to add a new namespace declaration, serialize
  // and push it on the namespace stack
  if (aAddNSAttr) {
    if (aTagPrefix.IsEmpty()) {
      // Serialize default namespace decl
      SerializeAttr(EmptyString(), xmlnsStr, aTagNamespaceURI, aStr, true);
    } else {
      // Serialize namespace decl
      SerializeAttr(xmlnsStr, aTagPrefix, aTagNamespaceURI, aStr, true);
    }
    PushNameSpaceDecl(aTagPrefix, aTagNamespaceURI, aOriginalElement);
  }

  NS_NAMED_LITERAL_STRING(_mozStr, "_moz");

  count = aContent->GetAttrCount();

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

    if (aSkipAttr == index) {
        continue;
    }

    const nsAttrName* name = aContent->GetAttrNameAt(index);
    int32_t namespaceID = name->NamespaceID();
    nsIAtom* attrName = name->LocalName();
    nsIAtom* attrPrefix = name->GetPrefix();

    // Filter out any attribute starting with [-|_]moz
    nsDependentAtomString attrNameStr(attrName);
    if (StringBeginsWith(attrNameStr, NS_LITERAL_STRING("_moz")) ||
        StringBeginsWith(attrNameStr, NS_LITERAL_STRING("-moz"))) {
      continue;
    }

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

    bool addNSAttr = false;
    if (kNameSpaceID_XMLNS != namespaceID) {
      nsContentUtils::NameSpaceManager()->GetNameSpaceURI(namespaceID, uriStr);
      addNSAttr = ConfirmPrefix(prefixStr, uriStr, aOriginalElement, true);
    }

    aContent->GetAttr(namespaceID, attrName, valueStr);

    nsDependentAtomString nameStr(attrName);
    bool isJS = false;

    if (kNameSpaceID_XHTML == contentNamespaceID) {
      //
      // Filter out special case of <br type="_moz"> or <br _moz*>,
      // used by the editor.  Bug 16988.  Yuck.
      //
      if (namespaceID == kNameSpaceID_None && aTagName == nsGkAtoms::br && attrName == nsGkAtoms::type
          && StringBeginsWith(valueStr, _mozStr)) {
        continue;
      }

      if (mIsCopying && mIsFirstChildOfOL && (aTagName == nsGkAtoms::li)
          && (attrName == nsGkAtoms::value)) {
        // This is handled separately in SerializeLIValueAttribute()
        continue;
      }

      isJS = IsJavaScript(aContent, attrName, namespaceID, valueStr);

      if (namespaceID == kNameSpaceID_None && 
          ((attrName == nsGkAtoms::href) ||
          (attrName == nsGkAtoms::src))) {
        // Make all links absolute when converting only the selection:
        if (mFlags & nsIDocumentEncoder::OutputAbsoluteLinks) {
          // Would be nice to handle OBJECT and APPLET tags,
          // but that gets more complicated since we have to
          // search the tag list for CODEBASE as well.
          // For now, just leave them relative.
          nsCOMPtr<nsIURI> uri = aContent->GetBaseURI();
          if (uri) {
            nsAutoString absURI;
            rv = NS_MakeAbsoluteURI(absURI, valueStr, uri);
            if (NS_SUCCEEDED(rv)) {
              valueStr = absURI;
            }
          }
        }
        // Need to escape URI.
        nsAutoString tempURI(valueStr);
        if (!isJS && NS_FAILED(EscapeURI(aContent, tempURI, valueStr)))
          valueStr = tempURI;
      }

      if (mRewriteEncodingDeclaration && aTagName == nsGkAtoms::meta &&
          attrName == nsGkAtoms::content) {
        // If we're serializing a <meta http-equiv="content-type">,
        // use the proper value, rather than what's in the document.
        nsAutoString header;
        aContent->GetAttr(kNameSpaceID_None, nsGkAtoms::httpEquiv, header);
        if (header.LowerCaseEqualsLiteral("content-type")) {
          valueStr = NS_LITERAL_STRING("text/html; charset=") +
            NS_ConvertASCIItoUTF16(mCharset);
        }
      }

      // Expand shorthand attribute.
      if (namespaceID == kNameSpaceID_None && IsShorthandAttr(attrName, aTagName) && valueStr.IsEmpty()) {
        valueStr = nameStr;
      }
    }
    else {
      isJS = IsJavaScript(aContent, attrName, namespaceID, valueStr);
    }

    SerializeAttr(prefixStr, nameStr, valueStr, aStr, !isJS);

    if (addNSAttr) {
      NS_ASSERTION(!prefixStr.IsEmpty(),
                   "Namespaced attributes must have a prefix");
      SerializeAttr(xmlnsStr, prefixStr, uriStr, aStr, true);
      PushNameSpaceDecl(prefixStr, uriStr, aOriginalElement);
    }
  }
}


void 
nsXHTMLContentSerializer::AppendEndOfElementStart(nsIContent *aOriginalElement,
                                                  nsIAtom * aName,
                                                  int32_t aNamespaceID,
                                                  nsAString& aStr)
{
  // this method is not called by nsHTMLContentSerializer
  // so we don't have to check HTML element, just XHTML
  NS_ASSERTION(!mIsHTMLSerializer, "nsHTMLContentSerializer shouldn't call this method !");

  if (kNameSpaceID_XHTML != aNamespaceID) {
    nsXMLContentSerializer::AppendEndOfElementStart(aOriginalElement, aName,
                                                    aNamespaceID, aStr);
    return;
  }

  nsIContent* content = aOriginalElement;

  // for non empty elements, even if they are not a container, we always
  // serialize their content, because the XHTML element could contain non XHTML
  // nodes useful in some context, like in an XSLT stylesheet
  if (HasNoChildren(content)) {

    nsIParserService* parserService = nsContentUtils::GetParserService();
  
    if (parserService) {
      bool isContainer;
      parserService->
        IsContainer(parserService->HTMLCaseSensitiveAtomTagToId(aName),
                    isContainer);
      if (!isContainer) {
        // for backward compatibility with HTML 4 user agents
        // only non-container HTML elements can be closed immediatly,
        // and a space is added before />
        AppendToString(NS_LITERAL_STRING(" />"), aStr);
        return;
      }
    }
  }
  AppendToString(kGreaterThan, aStr);
}

void
nsXHTMLContentSerializer::AfterElementStart(nsIContent * aContent,
                                            nsIContent *aOriginalElement,
                                            nsAString& aStr)
{
  nsIAtom *name = aContent->Tag();
  if (aContent->GetNameSpaceID() == kNameSpaceID_XHTML &&
      mRewriteEncodingDeclaration &&
      name == nsGkAtoms::head) {

    // Check if there already are any content-type meta children.
    // If there are, they will be modified to use the correct charset.
    // If there aren't, we'll insert one here.
    bool hasMeta = false;
    for (nsIContent* child = aContent->GetFirstChild();
         child;
         child = child->GetNextSibling()) {
      if (child->IsHTML(nsGkAtoms::meta) &&
          child->HasAttr(kNameSpaceID_None, nsGkAtoms::content)) {
        nsAutoString header;
        child->GetAttr(kNameSpaceID_None, nsGkAtoms::httpEquiv, header);

        if (header.LowerCaseEqualsLiteral("content-type")) {
          hasMeta = true;
          break;
        }
      }
    }

    if (!hasMeta) {
      AppendNewLineToString(aStr);
      if (mDoFormat) {
        AppendIndentation(aStr);
      }
      AppendToString(NS_LITERAL_STRING("<meta http-equiv=\"content-type\""),
                    aStr);
      AppendToString(NS_LITERAL_STRING(" content=\"text/html; charset="), aStr);
      AppendToString(NS_ConvertASCIItoUTF16(mCharset), aStr);
      if (mIsHTMLSerializer)
        AppendToString(NS_LITERAL_STRING("\">"), aStr);
      else
        AppendToString(NS_LITERAL_STRING("\" />"), aStr);
    }
  }
}

void
nsXHTMLContentSerializer::AfterElementEnd(nsIContent * aContent,
                                          nsAString& aStr)
{
  NS_ASSERTION(!mIsHTMLSerializer, "nsHTMLContentSerializer shouldn't call this method !");

  int32_t namespaceID = aContent->GetNameSpaceID();
  nsIAtom *name = aContent->Tag();

  // this method is not called by nsHTMLContentSerializer
  // so we don't have to check HTML element, just XHTML
  if (kNameSpaceID_XHTML == namespaceID && name == nsGkAtoms::body) {
    --mInBody;
  }
}


NS_IMETHODIMP
nsXHTMLContentSerializer::AppendDocumentStart(nsIDocument *aDocument,
                                              nsAString& aStr)
{
  if (!mBodyOnly)
    return nsXMLContentSerializer::AppendDocumentStart(aDocument, aStr);

  return NS_OK;
}

bool
nsXHTMLContentSerializer::CheckElementStart(nsIContent * aContent,
                                            bool & aForceFormat,
                                            nsAString& aStr)
{
  // The _moz_dirty attribute is emitted by the editor to
  // indicate that this element should be pretty printed
  // even if we're not in pretty printing mode
  aForceFormat = !(mFlags & nsIDocumentEncoder::OutputIgnoreMozDirty) &&
                 aContent->HasAttr(kNameSpaceID_None, nsGkAtoms::mozdirty);

  nsIAtom *name = aContent->Tag();
  int32_t namespaceID = aContent->GetNameSpaceID();

  if (namespaceID == kNameSpaceID_XHTML) {
    if (name == nsGkAtoms::br && mPreLevel > 0 && 
        (mFlags & nsIDocumentEncoder::OutputNoFormattingInPre)) {
      AppendNewLineToString(aStr);
      return false;
    }

    if (name == nsGkAtoms::body) {
      ++mInBody;
    }
  }
  return true;
}

bool
nsXHTMLContentSerializer::CheckElementEnd(nsIContent * aContent,
                                          bool & aForceFormat,
                                          nsAString& aStr)
{
  NS_ASSERTION(!mIsHTMLSerializer, "nsHTMLContentSerializer shouldn't call this method !");

  aForceFormat = !(mFlags & nsIDocumentEncoder::OutputIgnoreMozDirty) &&
                 aContent->HasAttr(kNameSpaceID_None, nsGkAtoms::mozdirty);

  nsIAtom *name = aContent->Tag();
  int32_t namespaceID = aContent->GetNameSpaceID();

  // this method is not called by nsHTMLContentSerializer
  // so we don't have to check HTML element, just XHTML
  if (namespaceID == kNameSpaceID_XHTML) {
    if (mIsCopying && name == nsGkAtoms::ol) {
      NS_ASSERTION((!mOLStateStack.IsEmpty()), "Cannot have an empty OL Stack");
      /* Though at this point we must always have an state to be deleted as all 
      the OL opening tags are supposed to push an olState object to the stack*/
      if (!mOLStateStack.IsEmpty()) {
        mOLStateStack.RemoveElementAt(mOLStateStack.Length() -1);
      }
    }

    if (HasNoChildren(aContent)) {
      nsIParserService* parserService = nsContentUtils::GetParserService();

      if (parserService) {
        bool isContainer;

        parserService->
          IsContainer(parserService->HTMLCaseSensitiveAtomTagToId(name),
                      isContainer);
        if (!isContainer) {
          // non-container HTML elements are already closed,
          // see AppendEndOfElementStart
          return false;
        }
      }
    }
    // for backward compatibility with old HTML user agents,
    // empty elements should have an ending tag, so we mustn't call
    // nsXMLContentSerializer::CheckElementEnd
    return true;
  }

  bool dummyFormat;
  return nsXMLContentSerializer::CheckElementEnd(aContent, dummyFormat, aStr);
}

void
nsXHTMLContentSerializer::AppendAndTranslateEntities(const nsAString& aStr,
                                                     nsAString& aOutputStr)
{
  if (mBodyOnly && !mInBody) {
    return;
  }

  if (mDisableEntityEncoding) {
    aOutputStr.Append(aStr);
    return;
  }
 
  nsXMLContentSerializer::AppendAndTranslateEntities(aStr, aOutputStr);
}

bool
nsXHTMLContentSerializer::IsShorthandAttr(const nsIAtom* aAttrName,
                                          const nsIAtom* aElementName)
{
  // checked
  if ((aAttrName == nsGkAtoms::checked) &&
      (aElementName == nsGkAtoms::input)) {
    return true;
  }

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

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

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

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

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

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

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

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

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

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

#ifdef MOZ_MEDIA
  // autoplay and controls
  if ((aElementName == nsGkAtoms::video || aElementName == nsGkAtoms::audio) &&
    (aAttrName == nsGkAtoms::autoplay || aAttrName == nsGkAtoms::muted ||
     aAttrName == nsGkAtoms::controls)) {
    return true;
  }
#endif

  return false;
}

bool
nsXHTMLContentSerializer::LineBreakBeforeOpen(int32_t aNamespaceID, nsIAtom* aName)
{

  if (aNamespaceID != kNameSpaceID_XHTML) {
    return mAddSpace;
  }

  if (aName == nsGkAtoms::title ||
      aName == nsGkAtoms::meta  ||
      aName == nsGkAtoms::link  ||
      aName == nsGkAtoms::style ||
      aName == nsGkAtoms::select ||
      aName == nsGkAtoms::option ||
      aName == nsGkAtoms::script ||
      aName == nsGkAtoms::html) {
    return true;
  }
  else {
    nsIParserService* parserService = nsContentUtils::GetParserService();

    if (parserService) {
      bool res;
      parserService->
        IsBlock(parserService->HTMLCaseSensitiveAtomTagToId(aName), res);
      return res;
    }
  }

  return mAddSpace;
}

bool 
nsXHTMLContentSerializer::LineBreakAfterOpen(int32_t aNamespaceID, nsIAtom* aName)
{

  if (aNamespaceID != kNameSpaceID_XHTML) {
    return false;
  }

  if ((aName == nsGkAtoms::html) ||
      (aName == nsGkAtoms::head) ||
      (aName == nsGkAtoms::body) ||
      (aName == nsGkAtoms::ul) ||
      (aName == nsGkAtoms::ol) ||
      (aName == nsGkAtoms::dl) ||
      (aName == nsGkAtoms::table) ||
      (aName == nsGkAtoms::tbody) ||
      (aName == nsGkAtoms::tr) ||
      (aName == nsGkAtoms::br) ||
      (aName == nsGkAtoms::meta) ||
      (aName == nsGkAtoms::link) ||
      (aName == nsGkAtoms::script) ||
      (aName == nsGkAtoms::select) ||
      (aName == nsGkAtoms::map) ||
      (aName == nsGkAtoms::area) ||
      (aName == nsGkAtoms::style)) {
    return true;
  }

  return false;
}

bool 
nsXHTMLContentSerializer::LineBreakBeforeClose(int32_t aNamespaceID, nsIAtom* aName)
{

  if (aNamespaceID != kNameSpaceID_XHTML) {
    return false;
  }

  if ((aName == nsGkAtoms::html) ||
      (aName == nsGkAtoms::head) ||
      (aName == nsGkAtoms::body) ||
      (aName == nsGkAtoms::ul) ||
      (aName == nsGkAtoms::ol) ||
      (aName == nsGkAtoms::dl) ||
      (aName == nsGkAtoms::select) ||
      (aName == nsGkAtoms::table) ||
      (aName == nsGkAtoms::tbody)) {
    return true;
  }
  return false;
}

bool 
nsXHTMLContentSerializer::LineBreakAfterClose(int32_t aNamespaceID, nsIAtom* aName)
{

  if (aNamespaceID != kNameSpaceID_XHTML) {
    return false;
  }

  if ((aName == nsGkAtoms::html) ||
      (aName == nsGkAtoms::head) ||
      (aName == nsGkAtoms::body) ||
      (aName == nsGkAtoms::tr) ||
      (aName == nsGkAtoms::th) ||
      (aName == nsGkAtoms::td) ||
      (aName == nsGkAtoms::pre) ||
      (aName == nsGkAtoms::title) ||
      (aName == nsGkAtoms::li) ||
      (aName == nsGkAtoms::dt) ||
      (aName == nsGkAtoms::dd) ||
      (aName == nsGkAtoms::blockquote) ||
      (aName == nsGkAtoms::select) ||
      (aName == nsGkAtoms::option) ||
      (aName == nsGkAtoms::p) ||
      (aName == nsGkAtoms::map) ||
      (aName == nsGkAtoms::div)) {
    return true;
  }
  else {
    nsIParserService* parserService = nsContentUtils::GetParserService();

    if (parserService) {
      bool res;
      parserService->
        IsBlock(parserService->HTMLCaseSensitiveAtomTagToId(aName), res);
      return res;
    }
  }

  return false;
}


void
nsXHTMLContentSerializer::MaybeEnterInPreContent(nsIContent* aNode)
{

  if (aNode->GetNameSpaceID() != kNameSpaceID_XHTML) {
    return;
  }

  nsIAtom *name = aNode->Tag();

  if (name == nsGkAtoms::pre ||
      name == nsGkAtoms::script ||
      name == nsGkAtoms::style ||
      name == nsGkAtoms::noscript ||
      name == nsGkAtoms::noframes
      ) {
    mPreLevel++;
  }
}

void
nsXHTMLContentSerializer::MaybeLeaveFromPreContent(nsIContent* aNode)
{
  if (aNode->GetNameSpaceID() != kNameSpaceID_XHTML) {
    return;
  }

  nsIAtom *name = aNode->Tag();
  if (name == nsGkAtoms::pre ||
      name == nsGkAtoms::script ||
      name == nsGkAtoms::style ||
      name == nsGkAtoms::noscript ||
      name == nsGkAtoms::noframes
    ) {
    --mPreLevel;
  }
}

void 
nsXHTMLContentSerializer::SerializeLIValueAttribute(nsIContent* aElement,
                                                    nsAString& aStr)
{
  // We are copying and we are at the "first" LI node of OL in selected range.
  // It may not be the first LI child of OL but it's first in the selected range.
  // Note that we get into this condition only once per a OL.
  bool found = false;
  nsCOMPtr<nsIDOMNode> currNode = do_QueryInterface(aElement);
  nsAutoString valueStr;

  olState state (0, false);

  if (!mOLStateStack.IsEmpty()) {
    state = mOLStateStack[mOLStateStack.Length()-1];
    // isFirstListItem should be true only before the serialization of the
    // first item in the list.
    state.isFirstListItem = false;
    mOLStateStack[mOLStateStack.Length()-1] = state;
  }

  int32_t startVal = state.startVal;
  int32_t offset = 0;

  // Traverse previous siblings until we find one with "value" attribute.
  // offset keeps track of how many previous siblings we had tocurrNode traverse.
  while (currNode && !found) {
    nsCOMPtr<nsIDOMElement> currElement = do_QueryInterface(currNode);
    // currElement may be null if it were a text node.
    if (currElement) {
      nsAutoString tagName;
      currElement->GetTagName(tagName);
      if (tagName.LowerCaseEqualsLiteral("li")) {
        currElement->GetAttribute(NS_LITERAL_STRING("value"), valueStr);
        if (valueStr.IsEmpty())
          offset++;
        else {
          found = true;
          nsresult rv = NS_OK;
          startVal = valueStr.ToInteger(&rv);
        }
      }
    }
    nsCOMPtr<nsIDOMNode> tmp;
    currNode->GetPreviousSibling(getter_AddRefs(tmp));
    currNode.swap(tmp);
  }
  // If LI was not having "value", Set the "value" attribute for it.
  // Note that We are at the first LI in the selected range of OL.
  if (offset == 0 && found) {
    // offset = 0 => LI itself has the value attribute and we did not need to traverse back.
    // Just serialize value attribute like other tags.
    SerializeAttr(EmptyString(), NS_LITERAL_STRING("value"), valueStr, aStr, false);
  }
  else if (offset == 1 && !found) {
    /*(offset = 1 && !found) means either LI is the first child node of OL
    and LI is not having "value" attribute.
    In that case we would not like to set "value" attribute to reduce the changes.
    */
    //do nothing...
  }
  else if (offset > 0) {
    // Set value attribute.
    nsAutoString valueStr;

    //As serializer needs to use this valueAttr we are creating here, 
    valueStr.AppendInt(startVal + offset);
    SerializeAttr(EmptyString(), NS_LITERAL_STRING("value"), valueStr, aStr, false);
  }
}

bool
nsXHTMLContentSerializer::IsFirstChildOfOL(nsIContent* aElement)
{
  nsCOMPtr<nsIDOMNode> node = do_QueryInterface(aElement);
  nsAutoString parentName;

  nsCOMPtr<nsIDOMNode> parentNode;
  node->GetParentNode(getter_AddRefs(parentNode));
  if (parentNode)
    parentNode->GetNodeName(parentName);
  else
    return false;

  if (parentName.LowerCaseEqualsLiteral("ol")) {

    if (!mOLStateStack.IsEmpty()) {
      olState state = mOLStateStack[mOLStateStack.Length()-1];
      if (state.isFirstListItem)
        return true;
    }

    return false;
  }
  else
    return false;
}

bool
nsXHTMLContentSerializer::HasNoChildren(nsIContent * aContent) {

  for (nsIContent* child = aContent->GetFirstChild();
       child;
       child = child->GetNextSibling()) {
       
    if (!child->IsNodeOfType(nsINode::eTEXT))
      return false;

    if (child->TextLength())
      return false;
  }

  return true;
}