Bug 563526 - Ignore U+0000 in element content when the tree builder is not in the "text" mode or the "in foreign" mode. r=jonas.

--HG--
extra : rebase_source : 97a67dc820c9b8ae1265b0c2c7e39c69f1733cee

parser/html/javasrc/Tokenizer.java

@@ -5860,7 +5860,7 @@ public class Tokenizer implements Locator {
     private void emitReplacementCharacter(@NoLength char[] buf, int pos)
             throws SAXException {
         flushChars(buf, pos);
-        tokenHandler.characters(Tokenizer.REPLACEMENT_CHARACTER, 0, 1);
+        tokenHandler.zeroOriginatingReplacementCharacter();
         cstart = pos + 1;
     }
 

parser/html/javasrc/TreeBuilder.java

@@ -60,6 +60,11 @@ import org.xml.sax.SAXParseException;
 public abstract class TreeBuilder<T> implements TokenHandler,
         TreeBuilderState<T> {
     
+    /**
+     * Array version of U+FFFD.
+     */
+    private static final @NoLength char[] REPLACEMENT_CHARACTER = { '\uFFFD' };
+    
     // Start dispatch groups
 
     final static int OTHER = 0;
@@ -839,8 +844,11 @@ public abstract class TreeBuilder<T> implements TokenHandler,
             needToDropLF = false;
         }
 
+        if (inForeign) {
+            accumulateCharacters(buf, start, length);
+            return;
+        }
         // optimize the most common case
-        // XXX should there be an IN FOREIGN check here?
         switch (mode) {
             case IN_BODY:
             case IN_CELL:
@@ -1206,6 +1214,16 @@ public abstract class TreeBuilder<T> implements TokenHandler,
         }
     }
 
+    /**
+     * @see nu.validator.htmlparser.common.TokenHandler#zeroOriginatingReplacementCharacter()
+     */
+    @Override public void zeroOriginatingReplacementCharacter()
+            throws SAXException {
+        if (inForeign || mode == TEXT) {
+            characters(REPLACEMENT_CHARACTER, 0, 1);
+        }
+    }
+
     public final void eof() throws SAXException {
         flushCharacters();
         if (inForeign) {

parser/html/nsHtml5Tokenizer.cpp

@@ -3339,7 +3339,7 @@ void
 nsHtml5Tokenizer::emitReplacementCharacter(PRUnichar* buf, PRInt32 pos)
 {
   flushChars(buf, pos);
-  tokenHandler->characters(nsHtml5Tokenizer::REPLACEMENT_CHARACTER, 0, 1);
+  tokenHandler->zeroOriginatingReplacementCharacter();
   cstart = pos + 1;
 }
 

parser/html/nsHtml5TreeBuilder.cpp

@@ -190,6 +190,10 @@ nsHtml5TreeBuilder::characters(const PRUnichar* buf, PRInt32 start, PRInt32 leng
     }
     needToDropLF = PR_FALSE;
   }
+  if (inForeign) {
+    accumulateCharacters(buf, start, length);
+    return;
+  }
   switch(mode) {
     case NS_HTML5TREE_BUILDER_IN_BODY:
     case NS_HTML5TREE_BUILDER_IN_CELL:
@@ -413,6 +417,14 @@ nsHtml5TreeBuilder::characters(const PRUnichar* buf, PRInt32 start, PRInt32 leng
   }
 }
 
+void 
+nsHtml5TreeBuilder::zeroOriginatingReplacementCharacter()
+{
+  if (inForeign || mode == NS_HTML5TREE_BUILDER_TEXT) {
+    characters(REPLACEMENT_CHARACTER, 0, 1);
+  }
+}
+
 void 
 nsHtml5TreeBuilder::eof()
 {

parser/html/nsHtml5TreeBuilder.h

@@ -70,6 +70,7 @@ class nsHtml5Portability;
 class nsHtml5TreeBuilder : public nsAHtml5TreeBuilderState
 {
   private:
+    static PRUnichar REPLACEMENT_CHARACTER[];
     static jArray<const char*,PRInt32> QUIRKY_PUBLIC_IDS;
     PRInt32 mode;
     PRInt32 originalMode;
@@ -100,6 +101,7 @@ class nsHtml5TreeBuilder : public nsAHtml5TreeBuilderState
     void doctype(nsIAtom* name, nsString* publicIdentifier, nsString* systemIdentifier, PRBool forceQuirks);
     void comment(PRUnichar* buf, PRInt32 start, PRInt32 length);
     void characters(const PRUnichar* buf, PRInt32 start, PRInt32 length);
+    void zeroOriginatingReplacementCharacter();
     void eof();
     void endTokenization();
     void startTag(nsHtml5ElementName* elementName, nsHtml5HtmlAttributes* attributes, PRBool selfClosing);
@@ -239,6 +241,7 @@ class nsHtml5TreeBuilder : public nsAHtml5TreeBuilderState
 };
 
 #ifdef nsHtml5TreeBuilder_cpp__
+PRUnichar nsHtml5TreeBuilder::REPLACEMENT_CHARACTER[] = { 0xfffd };
 jArray<const char*,PRInt32> nsHtml5TreeBuilder::QUIRKY_PUBLIC_IDS = nsnull;
 #endif