bug 263053: Get rid of the <endnote> tag when parsing fragments with context, since this can fail if the input stream contains another <endnote> tag. Instead, parse the context in different "chunks" from the new buffer, and notify the fragment sink by way of functions. r=peterv sr=bzbarsky

content/html/content/src/nsHTMLAtomList.h

@@ -137,7 +137,6 @@ HTML_ATOM(em, "em")
 HTML_ATOM(embed, "embed")
 HTML_ATOM(encoding, "encoding")
 HTML_ATOM(enctype, "enctype")
-HTML_ATOM(endnote, "endnote")  // contextual fragments
 HTML_ATOM(_event, "event")
 HTML_ATOM(face, "face")
 HTML_ATOM(fieldset, "fieldset")

content/html/document/src/nsHTMLFragmentContentSink.cpp

@@ -123,6 +123,8 @@ public:
   // nsIFragmentContentSink
   NS_IMETHOD GetFragment(nsIDOMDocumentFragment** aFragment);
   NS_IMETHOD SetTargetDocument(nsIDocument* aDocument);
+  NS_IMETHOD WillBuildContent();
+  NS_IMETHOD DidBuildContent();
 
   nsIContent* GetCurrentContent();
   PRInt32 PushContent(nsIContent *aContent);
@@ -140,13 +142,12 @@ public:
 
   nsresult Init();
 
-  PRBool mHitSentinel;
-  PRBool mSeenBody;
+  PRPackedBool mAllContent;
+  PRPackedBool mProcessing;
+  PRPackedBool mSeenBody;
 
-  nsIContent* mRoot;
-  nsIParser* mParser;
-  nsIDOMHTMLFormElement* mCurrentForm;
-  nsGenericHTMLElement* mCurrentMap;
+  nsCOMPtr<nsIContent> mRoot;
+  nsCOMPtr<nsIParser> mParser;
 
   nsVoidArray* mContentStack;
 
@@ -192,22 +193,14 @@ NS_NewHTMLFragmentContentSink(nsIFragmentContentSink** aResult)
 }
 
 nsHTMLFragmentContentSink::nsHTMLFragmentContentSink(PRBool aAllContent)
+  : mAllContent(aAllContent),
+    mProcessing(aAllContent),
+    mSeenBody(!aAllContent),
+    mContentStack(nsnull),
+    mText(nsnull),
+    mTextLength(0),
+    mTextSize(0)
 {
-  if (aAllContent) {
-    mHitSentinel = PR_TRUE;
-    mSeenBody = PR_FALSE;
-  } else {
-    mHitSentinel = PR_FALSE;
-    mSeenBody = PR_TRUE;
-  }
-  mRoot = nsnull;
-  mParser = nsnull;
-  mCurrentForm = nsnull;
-  mCurrentMap = nsnull;
-  mContentStack = nsnull;
-  mText = nsnull;
-  mTextLength = 0;
-  mTextSize = 0;
 }
 
 nsHTMLFragmentContentSink::~nsHTMLFragmentContentSink()
@@ -215,10 +208,6 @@ nsHTMLFragmentContentSink::~nsHTMLFragmentContentSink()
   // Should probably flush the text buffer here, just to make sure:
   //FlushText();
 
-  NS_IF_RELEASE(mRoot);
-  NS_IF_RELEASE(mParser);
-  NS_IF_RELEASE(mCurrentForm);
-  NS_IF_RELEASE(mCurrentMap);
   if (nsnull != mContentStack) {
     // there shouldn't be anything here except in an error condition
     PRInt32 indx = mContentStack->Count();
@@ -256,7 +245,9 @@ nsHTMLFragmentContentSink::WillBuildModel(void)
   nsresult rv = NS_NewDocumentFragment(getter_AddRefs(frag), mTargetDocument);
   NS_ENSURE_SUCCESS(rv, rv);
 
-  return CallQueryInterface(frag, &mRoot);
+  mRoot = do_QueryInterface(frag, &rv);
+  
+  return rv;
 }
 
 NS_IMETHODIMP
@@ -266,7 +257,7 @@ nsHTMLFragmentContentSink::DidBuildModel(void)
 
   // Drop our reference to the parser to get rid of a circular
   // reference.
-  NS_IF_RELEASE(mParser);
+  mParser = nsnull;
 
   return NS_OK;
 }
@@ -286,9 +277,7 @@ nsHTMLFragmentContentSink::WillResume(void)
 NS_IMETHODIMP
 nsHTMLFragmentContentSink::SetParser(nsIParser* aParser)
 {
-  NS_IF_RELEASE(mParser);
   mParser = aParser;
-  NS_IF_ADDREF(mParser);
 
   return NS_OK;
 }
@@ -449,15 +438,9 @@ nsHTMLFragmentContentSink::OpenContainer(const nsIParserNode& aNode)
 {
   NS_ENSURE_TRUE(mNodeInfoManager, NS_ERROR_NOT_INITIALIZED);
 
-  nsAutoString tag;
   nsresult result = NS_OK;
 
-  tag.Assign(aNode.GetText());
-
-  if (nsHTMLAtoms::endnote->Equals(tag)) {
-    mHitSentinel = PR_TRUE;
-  }
-  else if (mHitSentinel) {
+  if (mProcessing) {
     FlushText();
 
     nsHTMLTag nodeType = nsHTMLTag(aNode.GetNodeType());
@@ -522,7 +505,7 @@ nsHTMLFragmentContentSink::OpenContainer(const nsIParserNode& aNode)
 NS_IMETHODIMP
 nsHTMLFragmentContentSink::CloseContainer(const nsHTMLTag aTag)
 {
-  if (mHitSentinel && (nsnull != GetCurrentContent())) {
+  if (mProcessing && (nsnull != GetCurrentContent())) {
     nsIContent* content;
     FlushText();
     content = PopContent();
@@ -597,7 +580,7 @@ nsHTMLFragmentContentSink::AddLeaf(const nsIParserNode& aNode)
           if(nodeType == eHTMLTag_script    ||
              nodeType == eHTMLTag_style     ||
              nodeType == eHTMLTag_textarea  ||
-             nodeType == eHTMLTag_xmp) {
+             nodeType == eHTMLTag_server) {
 
             // Create a text node holding the content
             nsCOMPtr<nsIDTD> dtd;
@@ -711,6 +694,25 @@ nsHTMLFragmentContentSink::SetTargetDocument(nsIDocument* aTargetDocument)
   return NS_OK;
 }
 
+NS_IMETHODIMP
+nsHTMLFragmentContentSink::WillBuildContent()
+{
+  mProcessing = PR_TRUE;
+
+  return NS_OK;
+}
+
+NS_IMETHODIMP
+nsHTMLFragmentContentSink::DidBuildContent()
+{
+  if (!mAllContent) {
+    FlushText();
+    mProcessing = PR_FALSE;
+  }
+
+  return NS_OK;
+}
+
 nsIContent*
 nsHTMLFragmentContentSink::GetCurrentContent()
 {

content/xml/document/src/nsXMLFragmentContentSink.cpp

@@ -20,6 +20,7 @@
  * the Initial Developer. All Rights Reserved.
  *
  * Contributor(s):
+ *   Blake Kaplan <mrbkap@gmail.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"),
@@ -87,6 +88,8 @@ public:
   // nsIFragmentContentSink
   NS_IMETHOD GetFragment(nsIDOMDocumentFragment** aFragment);
   NS_IMETHOD SetTargetDocument(nsIDocument* aDocument);
+  NS_IMETHOD WillBuildContent();
+  NS_IMETHOD DidBuildContent();
 
 protected:
   virtual PRBool SetDocElement(PRInt32 aNameSpaceID, 
@@ -114,7 +117,6 @@ protected:
 
   // if FALSE, take content inside endnote tag
   PRPackedBool          mAllContent;
-  nsCOMPtr<nsIContent>  mEndnote;
 };
 
 static nsresult
@@ -171,29 +173,20 @@ nsXMLFragmentContentSink::WillBuildModel(void)
   NS_ENSURE_SUCCESS(rv, rv);
 
   mRoot = do_QueryInterface(frag);
-  PushContent(mRoot); // preload content stack because we know all content goes in the fragment
+  
+  if (mAllContent) {
+    // Preload content stack because we know all content goes in the fragment
+    PushContent(mRoot);
+  }
+
   return rv;
 }
 
 NS_IMETHODIMP 
 nsXMLFragmentContentSink::DidBuildModel()
 {
-  PopContent();  // remove mRoot pushed above
-  
-  if (!mAllContent && !mParseError) {
-    NS_ASSERTION(mEndnote, "<endnote> missing in fragment string.");
-    if (mEndnote) {
-      NS_ASSERTION(mRoot->GetChildCount() == 1, "contents have too many children!");
-      // move guts
-      for (PRUint32 child = mEndnote->GetChildCount(); child > 0; child--) {
-        nsCOMPtr<nsIContent> firstchild = mEndnote->GetChildAt(0);
-        mEndnote->RemoveChildAt( 0, PR_FALSE );
-        mRoot->AppendChildTo( firstchild, PR_FALSE, PR_FALSE );
-      }
-      // delete outer content
-      mRoot->RemoveChildAt( 0, PR_FALSE );
-    }
-    // else just leave the content in the fragment.  or should we fail?
+  if (mAllContent) {
+    PopContent();  // remove mRoot pushed above
   }
 
   nsCOMPtr<nsIParser> kungFuDeathGrip(mParser);
@@ -237,11 +230,17 @@ nsXMLFragmentContentSink::CreateElement(const PRUnichar** aAtts, PRUint32 aAttsC
   nsresult rv = nsXMLContentSink::CreateElement(aAtts, aAttsCount,
                                 aNodeInfo, aLineNumber,
                                 aResult, aAppendContent);
-  *aAppendContent = PR_TRUE;  // make sure scripts added immediately, not on close.
 
-  if (NS_SUCCEEDED(rv) && aNodeInfo->Equals(nsHTMLAtoms::endnote))
-    mEndnote = *aResult;
-  
+  // Make sure that scripts are added immediately, not on close.
+  *aAppendContent = PR_TRUE;
+
+  // However, when we aren't grabbing all of the content we, never open a doc
+  // element, we run into trouble on the first element, so we don't append,
+  // and simply push this onto the content stack.
+  if (!mAllContent && mContentStack.Count() == 0) {
+    *aAppendContent = PR_FALSE;
+  }
+
   return rv;
 }
 
@@ -387,3 +386,28 @@ nsXMLFragmentContentSink::SetTargetDocument(nsIDocument* aTargetDocument)
   return NS_OK;
 }
 
+NS_IMETHODIMP
+nsXMLFragmentContentSink::WillBuildContent()
+{
+  // If we're taking all of the content, then we've already pushed mRoot
+  // onto the content stack, otherwise, start here.
+  if (!mAllContent) {
+    PushContent(mRoot);
+  }
+
+  return NS_OK;
+}
+
+NS_IMETHODIMP
+nsXMLFragmentContentSink::DidBuildContent()
+{
+  // If we're taking all of the content, then this is handled in DidBuildModel
+  if (!mAllContent) {
+    // Note: we need to FlushText() here because if we don't, we might not get
+    // an end element to do it for us, so make sure.
+    FlushText();
+    PopContent();
+  }
+
+  return NS_OK;
+}

parser/htmlparser/public/nsHTMLTagList.h

@@ -96,7 +96,6 @@ HTML_TAG(dl, SharedList)
 HTML_TAG(dt, Span)
 HTML_TAG(em, Span)
 HTML_TAG(embed, Shared)
-HTML_TAG(endnote, Span)
 HTML_TAG(fieldset, FieldSet)
 HTML_TAG(font, Font)
 HTML_TAG(form, NOTUSED)

parser/htmlparser/public/nsHTMLTags.h

@@ -63,7 +63,7 @@ enum nsHTMLTag {
 #undef HTML_TAG
 #undef HTML_OTHER
 
-// Currently there are 108 HTML tags. eHTMLTag_text = 110.
+// Currently there are 107 HTML tags. eHTMLTag_text = 109.
 #define NS_HTML_TAG_MAX PRInt32(eHTMLTag_text - 1)
 
 class nsHTMLTags {

parser/htmlparser/public/nsIFragmentContentSink.h

@@ -43,8 +43,8 @@ class nsIDOMDocumentFragment;
 class nsIDocument;
 
 #define NS_I_FRAGMENT_CONTENT_SINK_IID \
- { 0x2b23c1fb, 0xb83c, 0x436b, \
- { 0xb7, 0x2a, 0x9c, 0xbe, 0xf1, 0xe9, 0x9b, 0x20 } };
+ { 0xe9ea6afb, 0x92f3, 0x4270, \
+ { 0xb2, 0x67, 0xd2, 0xe3, 0x8d, 0x0e, 0x95, 0x45 } }
 
 class nsIFragmentContentSink : public nsISupports {
 public:
@@ -66,11 +66,25 @@ public:
    * (should not be null)
    */
   NS_IMETHOD SetTargetDocument(nsIDocument* aDocument) = 0;
+
+  /**
+   * This method is used to indicate to the sink that we're done building
+   * the context and should start paying attention to the incoming content
+   */
+  NS_IMETHOD WillBuildContent() = 0;
+
+  /**
+   * This method is used to indicate to the sink that we're done building
+   * The real content. This is useful if you want to parse additional context
+   * (such as an end context).
+   */
+  NS_IMETHOD DidBuildContent() = 0;
 };
 
 /**
- * Base version takes string nested in context, context ends with <endnote>.
- * 2 version just loads whole string.
+ * Base version takes string nested in context, content surrounded by
+ * WillBuildContent()/DidBuildContent() calls. The 2nd version just loads
+ * the whole string.
  */
 
 #define NS_HTMLFRAGMENTSINK_CONTRACTID "@mozilla.org/layout/htmlfragmentsink;1"

parser/htmlparser/public/nsIParser.h

@@ -259,6 +259,19 @@ class nsIParser : public nsISupports {
     
     NS_IMETHOD Terminate(void) = 0;
 
+    /**
+     * This method gets called when you want to parse a fragment of HTML or XML
+     * surrounded by the context |aTagStack|. It requires that the parser have
+     * been given a fragment content sink.
+     *
+     * @param aSourceBuffer The XML or HTML that hasn't been parsed yet.
+     * @param aKey The key used by the parser.
+     * @param aTagStack The context of the source buffer.
+     * @param aXMLMode Whether this is XML or HTML
+     * @param aContentType The content-type of this document.
+     * @param aMode The DTDMode that the parser should parse this fragment in.
+     * @return Success or failure.
+     */
     NS_IMETHOD ParseFragment(const nsAString& aSourceBuffer,
                              void* aKey,
                              nsVoidArray& aTagStack,

parser/htmlparser/src/COtherElements.h

@@ -2251,7 +2251,6 @@ void CElementTable::InitializeElements() {
   
   CPhraseElement::Initialize(       mDfltElements[eHTMLTag_em],         eHTMLTag_em);
   CElement::Initialize(   mDfltElements[eHTMLTag_embed],                eHTMLTag_embed);
-  CBlockElement::Initialize(        mDfltElements[eHTMLTag_endnote],    eHTMLTag_endnote);
 
   CElement::Initialize(             mDfltElements[eHTMLTag_fieldset],   eHTMLTag_fieldset, CBlockElement::GetGroup(),  CFlowElement::GetContainedGroups());
   mDfltElements[eHTMLTag_fieldset].mIncludeKids=kFieldsetKids;

parser/htmlparser/src/nsElementTable.cpp

@@ -545,15 +545,6 @@ void InitializeElementTable(void) {
       /*special props, prop-range*/       kNonContainer,kDefaultPropRange,
       /*special parents,kids,skip*/       0,&gContainsParam,eHTMLTag_unknown);
 
-    Initialize( 
-      /*tag*/                             eHTMLTag_endnote,
-      /*req-parent excl-parent*/          eHTMLTag_unknown,eHTMLTag_unknown,
-	    /*rootnodes,endrootnodes*/          &gRootTags,&gRootTags,	
-      /*autoclose starttags and endtags*/ 0,0,0,0,
-      /*parent,incl,exclgroups*/          kFlowEntity, kFlowEntity, kNone,	
-      /*special props, prop-range*/       kLegalOpen,kDefaultPropRange, // Don't munge the content model around <endnote>. bug 100175
-      /*special parents,kids,skip*/       0,0,eHTMLTag_unknown);
-
     Initialize( 
       /*tag*/                             eHTMLTag_fieldset,
       /*requiredAncestor*/                eHTMLTag_unknown,eHTMLTag_unknown,

parser/htmlparser/src/nsHTMLTags.cpp

@@ -111,8 +111,6 @@ static const PRUnichar sHTMLTagUnicodeName_em[] =
   {'e', 'm', '\0'};
 static const PRUnichar sHTMLTagUnicodeName_embed[] =
   {'e', 'm', 'b', 'e', 'd', '\0'};
-static const PRUnichar sHTMLTagUnicodeName_endnote[] =
-  {'e', 'n', 'd', 'n', 'o', 't', 'e', '\0'};
 static const PRUnichar sHTMLTagUnicodeName_fieldset[] =
   {'f', 'i', 'e', 'l', 'd', 's', 'e', 't', '\0'};
 static const PRUnichar sHTMLTagUnicodeName_font[] =

parser/htmlparser/src/nsParser.cpp

@@ -65,6 +65,7 @@
 #include "nsIServiceManager.h"
 #include "nsICategoryManager.h"
 #include "nsISupportsPrimitives.h"
+#include "nsIFragmentContentSink.h"
 
 #ifdef MOZ_VIEW_SOURCE
 #include "nsViewSourceHTML.h" 
@@ -1804,25 +1805,40 @@ nsParser::ParseFragment(const nsAString& aSourceBuffer,
                         nsDTDMode aMode)
 {
   nsresult result = NS_OK;
-  nsAutoString  theContext, endContext;
+  nsAutoString  theContext;
   PRUint32 theCount = aTagStack.Count();
   PRUint32 theIndex = 0;
-  
+
+  // Disable observers for fragments
+  mFlags &= ~NS_PARSER_FLAG_OBSERVERS_ENABLED;
+
   for (theIndex = 0; theIndex < theCount; theIndex++) {
     theContext.AppendLiteral("<");
     theContext.Append((PRUnichar*)aTagStack.ElementAt(theCount - theIndex - 1));
     theContext.AppendLiteral(">");
   }
-  
-  // Note duplication: nsHTMLAtoms::endnote == an atom in nsHTMLTags
-  theContext.AppendLiteral("<");
-  theContext.Append(nsHTMLTags::GetStringValue(eHTMLTag_endnote));
-  theContext.AppendLiteral(">");
 
-  if (aXMLMode) {
-    endContext.AppendLiteral("</");
-    endContext.Append(nsHTMLTags::GetStringValue(eHTMLTag_endnote));
-    endContext.AppendLiteral(">");
+  // First, parse the context to build up the DTD's tag stack. Note that we
+  // pass PR_FALSE for the aLastCall parameter.
+  result = Parse(theContext, (void*)&theContext, aMimeType, 
+                 PR_FALSE, PR_FALSE, aMode);
+  if (NS_FAILED(result)) {
+    mFlags |= NS_PARSER_FLAG_OBSERVERS_ENABLED;
+    return result;
+  }
+
+  nsCOMPtr<nsIFragmentContentSink> fragSink = do_QueryInterface(mSink);
+  NS_ASSERTION(fragSink, "ParseFragment requires a fragment content sink");
+
+  fragSink->WillBuildContent();
+  // Now, parse the actual content. Note that this is the last call for HTML
+  // content, but for XML, we will want to build and parse the end tags.
+  result = Parse(aSourceBuffer, (void*)&theContext, aMimeType,
+                 PR_FALSE, !aXMLMode, aMode);
+  fragSink->DidBuildContent();
+
+  if (aXMLMode && NS_SUCCEEDED(result)) {
+    nsAutoString endContext;
 
     for (theIndex = 0; theIndex < theCount; theIndex++) {
       endContext.AppendLiteral("</");
@@ -1834,12 +1850,11 @@ nsParser::ParseFragment(const nsAString& aSourceBuffer,
         endContext.Append( Substring(thisTag,0,endOfTag) );
       endContext.AppendLiteral(">");
     }
+
+    result = Parse(endContext, (void*)&theContext, aMimeType,
+                   PR_FALSE, PR_TRUE, aMode);
   }
     
-  //now it's time to try to build the model from this fragment
-
-  mFlags &= ~NS_PARSER_FLAG_OBSERVERS_ENABLED; //disable observers for fragments
-  result = Parse(theContext + aSourceBuffer + endContext,(void*)&theContext,aMimeType,PR_FALSE,PR_TRUE, aMode);
   mFlags |= NS_PARSER_FLAG_OBSERVERS_ENABLED; //now reenable.
 
   return result;