bug 70918: view-source adds > to tags that are missing them. r=bzbarsky sr=dmose

parser/htmlparser/public/nsToken.h

@@ -264,6 +264,16 @@ class CToken {
       mLineNumber = mLineNumber == 0 ? aLineNumber : mLineNumber;
     }
 
+    void SetInError(PRBool aInError)
+    {
+      mInError = aInError;
+    }
+
+    PRBool IsInError()
+    {
+      return mInError;
+    }
+
     void SetAttributeCount(PRInt16 aValue) {  mAttrCount = aValue; }
 
     /**
@@ -285,7 +295,8 @@ protected:
     PRInt32 mTypeID;
     PRInt32 mUseCount;
     PRInt32 mNewlineCount;
-    PRInt32 mLineNumber;
+    PRUint32 mLineNumber : 31;
+    PRUint32 mInError : 1;
     PRInt16 mAttrCount;
 };
 

parser/htmlparser/src/nsHTMLTokenizer.cpp

@@ -688,12 +688,17 @@ nsresult nsHTMLTokenizer::ConsumeAttributes(PRUnichar aChar,
           done = PR_TRUE;
         }
         else if(aChar == kLessThan) {
+          aToken->SetInError(PR_TRUE);
           done = PR_TRUE;
         }
       }//if
     }//if
   }//while
 
+  if (NS_FAILED(result)) {
+    aToken->SetInError(PR_TRUE);
+  }
+
   aToken->SetAttributeCount(theAttrCount);
   return result;
 }
@@ -728,6 +733,9 @@ nsresult nsHTMLTokenizer::ConsumeStartTag(PRUnichar aChar,CToken*& aToken,nsScan
       //Good. Now, let's see if the next char is ">". 
       //If so, we have a complete tag, otherwise, we have attributes.
       result = aScanner.Peek(aChar);
+      if (NS_FAILED(result)) {
+        aToken->SetInError(PR_TRUE);
+      }
       NS_ENSURE_SUCCESS(result, result);
 
       if(kGreaterThan != aChar) { //look for '>' 
@@ -854,6 +862,9 @@ nsresult nsHTMLTokenizer::ConsumeEndTag(PRUnichar aChar,CToken*& aToken,nsScanne
     NS_ENSURE_SUCCESS(result, result);
       
     result = aScanner.Peek(aChar);
+    if (NS_FAILED(result)) {
+      aToken->SetInError(PR_TRUE);
+    }
     NS_ENSURE_SUCCESS(result, result);
 
     if(kGreaterThan != aChar) {

parser/htmlparser/src/nsHTMLTokens.cpp

@@ -746,11 +746,6 @@ nsresult CCDATASectionToken::Consume(PRUnichar aChar, nsScanner& aScanner,PRInt3
         if (NS_OK==result &&
             (!inCDATA || kGreaterThan == aChar)) {
           result=aScanner.GetChar(aChar); //strip off the >
-
-          // XXX take me out when view source isn't stupid anymore
-          if (aFlag & NS_IPARSER_FLAG_VIEW_SOURCE) {
-            mTextValue.Append(aChar);
-          }
           done=PR_TRUE;
         }
       }
@@ -763,8 +758,7 @@ nsresult CCDATASectionToken::Consume(PRUnichar aChar, nsScanner& aScanner,PRInt3
     // In order to not completely lose the entire section, treat everything
     // until the end of the document as part of the CDATA section and let
     // the DTD handle it.
-    // XXX when view source actually displays errors, we'll need to propagate
+    mInError = PR_TRUE;
-    // the EOF down to it (i.e., not do this if we're viewing source).
     result = NS_OK;
   }
 
@@ -896,6 +890,14 @@ nsresult CMarkupDeclToken::Consume(PRUnichar aChar, nsScanner& aScanner,PRInt32
   
   aScanner.BindSubstring(mTextValue, origin, end);
 
+  if (kEOF == result) {
+    mInError = PR_TRUE;
+    if (!aScanner.IsIncremental()) {
+      // Hide this EOF.
+      result = NS_OK;
+    }
+  }
+
   return result;
 }
 
@@ -1732,6 +1734,7 @@ nsresult CAttributeToken::Consume(PRUnichar aChar, nsScanner& aScanner,PRInt32 a
                       mTextValue.Append(aChar);
                     } else if (result == NS_ERROR_HTMLPARSER_UNTERMINATEDSTRINGLITERAL) {
                       result = NS_OK;
+                      mInError = PR_TRUE;
                     }
                     // According to spec. we ( who? ) should ignore linefeeds. But look,
                     // even the carriage return was getting stripped ( wonder why! ) -
@@ -1744,6 +1747,7 @@ nsresult CAttributeToken::Consume(PRUnichar aChar, nsScanner& aScanner,PRInt32 a
                   }
                   else if (kGreaterThan==aChar){      
                     mHasEqualWithoutValue=PR_TRUE;
+                    mInError=PR_TRUE;
                   }
                   else {
                     static const nsReadEndCondition
@@ -1777,6 +1781,7 @@ nsresult CAttributeToken::Consume(PRUnichar aChar, nsScanner& aScanner,PRInt32 a
             //so let's see what it is. If it's a '"', then assume we're reading
             //from the middle of the value. Try stripping the quote and continuing...
             if (kQuote==aChar || kApostrophe==aChar){
+              mInError=PR_TRUE;
 
               if (!(aFlag & NS_IPARSER_FLAG_VIEW_SOURCE)) {
                 result=aScanner.SkipOver(aChar); //strip quote.
@@ -2244,6 +2249,7 @@ nsresult CInstructionToken::Consume(PRUnichar aChar,nsScanner& aScanner,PRInt32
 
   if (kEOF==result && !aScanner.IsIncremental()) {
     //Hide the EOF result because there is no more text coming.
+    mInError=PR_TRUE;
     result=NS_OK;
   }
 
@@ -2307,12 +2313,17 @@ nsresult CDoctypeDeclToken::Consume(PRUnichar aChar, nsScanner& aScanner,PRInt32
       // could belong to another tag.
       aScanner.GetChar(ch);
       end.advance(1); 
+    } else {
+      NS_ASSERTION(kLessThan == ch, 
+                   "Make sure this doctype decl. is really in error.");
+      mInError = PR_TRUE;
     }
   }
   else if (!aScanner.IsIncremental()) {
     // We have reached the document end but haven't
     // found either a '<' or a '>'. Therefore use
     // whatever we have.
+    mInError = PR_TRUE;
     result = NS_OK; 
   }
   

parser/htmlparser/src/nsToken.cpp

@@ -69,6 +69,7 @@ CToken::CToken(PRInt32 aTag) {
   mAttrCount=0;
   mNewlineCount=0;
   mLineNumber = 0;
+  mInError = PR_FALSE;
   mTypeID=aTag;
   // Note that the use count starts with 1 instead of 0. This
   // is because of the assumption that any token created is in

parser/htmlparser/src/nsViewSourceHTML.cpp

@@ -978,7 +978,7 @@ nsresult CViewSourceHTML::WriteAttributes(PRInt32 attrCount) {
  *  @param   
  *  @return  result status
  */
-nsresult CViewSourceHTML::WriteTag(PRInt32 aTagType,const nsAString & aText,PRInt32 attrCount,PRBool aNewlineRequired) {
+nsresult CViewSourceHTML::WriteTag(PRInt32 aTagType,const nsAString & aText,PRInt32 attrCount,PRBool aTagInError) {
   nsresult result=NS_OK;
 
   // adjust line number to what it will be after we finish writing this tag
@@ -997,9 +997,8 @@ nsresult CViewSourceHTML::WriteTag(PRInt32 aTagType,const nsAString & aText,PRIn
     return NS_ERROR_FAILURE;
 
   if (kBeforeText[aTagType][0] != 0) {
-    nsAutoString beforeText;
+    theContext.mITextToken.SetIndirectString(
-    beforeText.AssignWithConversion(kBeforeText[aTagType]);
+        NS_ConvertASCIItoUTF16(kBeforeText[aTagType]));
-    theContext.mITextToken.SetIndirectString(beforeText);
     nsCParserNode theNode(&theContext.mITextToken, 0/*stack token*/);
     mSink->AddLeaf(theNode);
   }
@@ -1052,15 +1051,16 @@ nsresult CViewSourceHTML::WriteTag(PRInt32 aTagType,const nsAString & aText,PRIn
     result=WriteAttributes(attrCount);
   }
 
-  if (kAfterText[aTagType][0] != 0) {
+  // Tokens are set in error if their ending > is not there, so don't output 
-    nsAutoString afterText;
+  // the after-text
-    afterText.AssignWithConversion(kAfterText[aTagType]);
+  if (!aTagInError && kAfterText[aTagType][0] != 0) {
-    theContext.mITextToken.SetIndirectString(afterText);
+    theContext.mITextToken.SetIndirectString(
+      NS_ConvertASCIItoUTF16(kAfterText[aTagType]));
     nsCParserNode theNode(&theContext.mITextToken, 0/*stack token*/);
     mSink->AddLeaf(theNode);
   }
 #ifdef DUMP_TO_FILE
-  if (gDumpFile && kDumpFileAfterText[aTagType][0])
+  if (!aTagInError && gDumpFile && kDumpFileAfterText[aTagType][0])
     fprintf(gDumpFile, kDumpFileAfterText[aTagType]);
 #endif // DUMP_TO_FILE
 
@@ -1097,7 +1097,7 @@ NS_IMETHODIMP CViewSourceHTML::HandleToken(CToken* aToken,nsIParser* aParser) {
         ++mTagCount;
 
         const nsAString& startValue = aToken->GetStringValue();
-        result=WriteTag(mStartTag,startValue,aToken->GetAttributeCount(),PR_TRUE);
+        result=WriteTag(mStartTag,startValue,aToken->GetAttributeCount(),aToken->IsInError());
 
         if((ePlainText!=mDocType) && mParser && (NS_OK==result)) {
           result = mSink->NotifyTagObservers(&theContext.mTokenNode);
@@ -1112,7 +1112,7 @@ NS_IMETHODIMP CViewSourceHTML::HandleToken(CToken* aToken,nsIParser* aParser) {
         }
 
         const nsAString& endValue = aToken->GetStringValue();
-        result=WriteTag(mEndTag,endValue,aToken->GetAttributeCount(),PR_TRUE);
+        result=WriteTag(mEndTag,endValue,aToken->GetAttributeCount(),aToken->IsInError());
       }
       break;
 
@@ -1121,10 +1121,10 @@ NS_IMETHODIMP CViewSourceHTML::HandleToken(CToken* aToken,nsIParser* aParser) {
         nsAutoString theStr;
         theStr.AssignLiteral("<!");
         theStr.Append(aToken->GetStringValue());
-        // Treat CDATA sections specially because they remember their last
+        if (!aToken->IsInError()) {
-        // character and can come back malformed.
+          theStr.AppendLiteral(">");
-//        theStr.AppendLiteral(">");
+        }
-        result=WriteTag(mCDATATag,theStr,0,PR_TRUE);
+        result=WriteTag(mCDATATag,theStr,0,aToken->IsInError());
       }
       break;
 
@@ -1133,8 +1133,10 @@ NS_IMETHODIMP CViewSourceHTML::HandleToken(CToken* aToken,nsIParser* aParser) {
         nsAutoString theStr;
         theStr.AssignLiteral("<!");
         theStr.Append(aToken->GetStringValue());
-        theStr.AppendLiteral(">");
+        if (!aToken->IsInError()) {
-        result=WriteTag(mMarkupDeclaration,theStr,0,PR_TRUE);
+          theStr.AppendLiteral(">");
+        }
+        result=WriteTag(mMarkupDeclaration,theStr,0,aToken->IsInError());
       }
       break;
 
@@ -1142,14 +1144,14 @@ NS_IMETHODIMP CViewSourceHTML::HandleToken(CToken* aToken,nsIParser* aParser) {
       {
         nsAutoString theStr;
         aToken->AppendSourceTo(theStr);
-        result=WriteTag(mCommentTag,theStr,0,PR_TRUE);
+        result=WriteTag(mCommentTag,theStr,0,aToken->IsInError());
       }
       break;
 
     case eToken_doctypeDecl:
       {
         const nsAString& doctypeValue = aToken->GetStringValue();
-        result=WriteTag(mDocTypeTag,doctypeValue,0,PR_TRUE);
+        result=WriteTag(mDocTypeTag,doctypeValue,0,aToken->IsInError());
       }
       break;
 
@@ -1182,7 +1184,7 @@ NS_IMETHODIMP CViewSourceHTML::HandleToken(CToken* aToken,nsIParser* aParser) {
     case eToken_text:
       {
         const nsAString& str = aToken->GetStringValue();         
-        result=WriteTag(mText,str,aToken->GetAttributeCount(),PR_TRUE);
+        result=WriteTag(mText,str,aToken->GetAttributeCount(),aToken->IsInError());
         ++mTokenCount;
         if (NS_VIEWSOURCE_TOKENS_PER_BLOCK > 0 &&
             mTokenCount > NS_VIEWSOURCE_TOKENS_PER_BLOCK && !str.IsEmpty()) {
@@ -1204,12 +1206,12 @@ NS_IMETHODIMP CViewSourceHTML::HandleToken(CToken* aToken,nsIParser* aParser) {
             theStr.Assign(NS_LITERAL_STRING("#") + theStr);
           }
         }
-        result=WriteTag(mEntityTag,theStr,0,PR_FALSE);
+        result=WriteTag(mEntityTag,theStr,0,aToken->IsInError());
       }
       break;
 
     case eToken_instruction:
-      result=WriteTag(mPITag,aToken->GetStringValue(),0,PR_TRUE);
+      result=WriteTag(mPITag,aToken->GetStringValue(),0,aToken->IsInError());
 
     default:
       result=NS_OK;

parser/htmlparser/src/nsViewSourceHTML.h

@@ -85,7 +85,7 @@ private:
     nsresult WriteTag(PRInt32 tagType,
                       const nsAString &aText,
                       PRInt32 attrCount,
-                      PRBool aNewlineRequired);
+                      PRBool aTagInError);
     
     nsresult WriteAttributes(PRInt32 attrCount);
     nsresult GenerateSummary();