Bug 612527 - Implement spec tweaks to U+0000 handling in the HTML5 parser. rs=jonas, a=blocking2.0-betaN.

parser/html/javasrc/Tokenizer.java

@@ -3424,7 +3424,7 @@ public class Tokenizer implements Locator {
                         }
                         switch (c) {
                             case '\u0000':
-                                emitReplacementCharacter(buf, pos);
+                                emitPlaintextReplacementCharacter(buf, pos);
                                 continue;
                             case '\r':
                                 emitCarriageReturn(buf, pos);
@@ -5785,6 +5785,13 @@ public class Tokenizer implements Locator {
         cstart = pos + 1;
     }
 
+    private void emitPlaintextReplacementCharacter(@NoLength char[] buf, int pos)
+            throws SAXException {
+        flushChars(buf, pos);
+        tokenHandler.characters(REPLACEMENT_CHARACTER, 0, 1);
+        cstart = pos + 1;
+    }
+
     private void setAdditionalAndRememberAmpersandLocation(char add) {
         additional = add;
         // [NOCPP[

parser/html/javasrc/TreeBuilder.java

@@ -1214,10 +1214,24 @@ public abstract class TreeBuilder<T> implements TokenHandler,
     /**
      * @see nu.validator.htmlparser.common.TokenHandler#zeroOriginatingReplacementCharacter()
      */
-    public void zeroOriginatingReplacementCharacter()
-            throws SAXException {
-        if (isInForeign() || mode == TEXT) {
-            characters(REPLACEMENT_CHARACTER, 0, 1);
+    public void zeroOriginatingReplacementCharacter() throws SAXException {
+        if (mode == TEXT) {
+            accumulateCharacters(REPLACEMENT_CHARACTER, 0, 1);
+            return;
+        }
+        if (currentPtr >= 0) {
+            StackNode<T> stackNode = stack[currentPtr];
+            if (stackNode.ns == "http://www.w3.org/1999/xhtml") {
+                return;
+            }
+            if (stackNode.isHtmlIntegrationPoint()) {
+                return;
+            }
+            if (stackNode.ns == "http://www.w3.org/1998/Math/MathML"
+                    && stackNode.getGroup() == MI_MO_MN_MS_MTEXT) {
+                return;
+            }
+            accumulateCharacters(REPLACEMENT_CHARACTER, 0, 1);
         }
     }
 

parser/html/nsHtml5Tokenizer.cpp

@@ -1717,7 +1717,7 @@ nsHtml5Tokenizer::stateLoop(PRInt32 state, PRUnichar c, PRInt32 pos, PRUnichar*
           }
           switch(c) {
             case '\0': {
-              emitReplacementCharacter(buf, pos);
+              emitPlaintextReplacementCharacter(buf, pos);
               continue;
             }
             case '\r': {
@@ -3300,6 +3300,14 @@ nsHtml5Tokenizer::emitReplacementCharacter(PRUnichar* buf, PRInt32 pos)
   cstart = pos + 1;
 }
 
+void 
+nsHtml5Tokenizer::emitPlaintextReplacementCharacter(PRUnichar* buf, PRInt32 pos)
+{
+  flushChars(buf, pos);
+  tokenHandler->characters(REPLACEMENT_CHARACTER, 0, 1);
+  cstart = pos + 1;
+}
+
 void 
 nsHtml5Tokenizer::setAdditionalAndRememberAmpersandLocation(PRUnichar add)
 {

parser/html/nsHtml5Tokenizer.h

@@ -259,6 +259,7 @@ class nsHtml5Tokenizer
   private:
     void emitCarriageReturn(PRUnichar* buf, PRInt32 pos);
     void emitReplacementCharacter(PRUnichar* buf, PRInt32 pos);
+    void emitPlaintextReplacementCharacter(PRUnichar* buf, PRInt32 pos);
     void setAdditionalAndRememberAmpersandLocation(PRUnichar add);
     void bogusDoctype();
     void bogusDoctypeWithoutQuirks();

parser/html/nsHtml5TreeBuilder.cpp

@@ -423,8 +423,22 @@ nsHtml5TreeBuilder::characters(const PRUnichar* buf, PRInt32 start, PRInt32 leng
 void 
 nsHtml5TreeBuilder::zeroOriginatingReplacementCharacter()
 {
-  if (isInForeign() || mode == NS_HTML5TREE_BUILDER_TEXT) {
-    characters(REPLACEMENT_CHARACTER, 0, 1);
+  if (mode == NS_HTML5TREE_BUILDER_TEXT) {
+    accumulateCharacters(REPLACEMENT_CHARACTER, 0, 1);
+    return;
+  }
+  if (currentPtr >= 0) {
+    nsHtml5StackNode* stackNode = stack[currentPtr];
+    if (stackNode->ns == kNameSpaceID_XHTML) {
+      return;
+    }
+    if (stackNode->isHtmlIntegrationPoint()) {
+      return;
+    }
+    if (stackNode->ns == kNameSpaceID_MathML && stackNode->getGroup() == NS_HTML5TREE_BUILDER_MI_MO_MN_MS_MTEXT) {
+      return;
+    }
+    accumulateCharacters(REPLACEMENT_CHARACTER, 0, 1);
   }
 }
 

parser/htmlparser/tests/mochitest/html5_tree_construction_exceptions.js

@@ -7,10 +7,6 @@
 var html5Exceptions = {
   "<!doctype html><keygen><frameset>" : true, // Bug 101019
   "<select><keygen>" : true, // Bug 101019
-  "<plaintext>\u0000filler\u0000text\u0000" : true, // Bug 612527
-  "<body><svg><foreignObject>\u0000filler\u0000text" : true, // Bug 612527
-  "<svg>\u0000</svg><frameset>" : true, // Bug 612527
-  "<svg>\u0000 </svg><frameset>" : true, // Bug 612527
   "<option><span><option>" : true, // Bug 612528
   "<!doctype html><div><body><frameset>" : true, // Bug 614241
 }

parser/htmlparser/tests/mochitest/html5lib_tree_construction/tests19.dat

@@ -1204,3 +1204,17 @@
 |     <summary>
 |       <article>
 |     "a"
+
+#data
+<!doctype html><p><a><plaintext>b
+#errors
+#document
+| <!DOCTYPE html>
+| <html>
+|   <head>
+|   <body>
+|     <p>
+|       <a>
+|     <plaintext>
+|       <a>
+|         "b"