Considering HTMLTokenizer state in HTMLSourceTracker

third_party/WebKit/Source/core/html/parser/HTMLSourceTracker.cpp

 /*
  * Copyright (C) 2010 Adam Barth. All Rights Reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
  * 1. Redistributions of source code must retain the above copyright
  *    notice, this list of conditions and the following disclaimer.
  * 2. Redistributions in binary form must reproduce the above copyright
  *    notice, this list of conditions and the following disclaimer in the
@@ skipping 19 matching lines @@
 
 namespace blink {
 
 HTMLSourceTracker::HTMLSourceTracker() : m_isStarted(false) {}
 
 void HTMLSourceTracker::start(SegmentedString& currentInput,
                               HTMLTokenizer* tokenizer,
                               HTMLToken& token) {
   if (token.type() == HTMLToken::Uninitialized && !m_isStarted) {
     m_previousSource.clear();
-    if (tokenizer->numberOfBufferedCharacters())
+    if (needToCheckTokenizerBuffer(tokenizer) &&
+        tokenizer->numberOfBufferedCharacters())
       m_previousSource = tokenizer->bufferedCharacters();
   } else
     m_previousSource.append(m_currentSource);

[Charlie Harrison, 2017/03/07 18:37:48]

while you are here could you add braces to the else statement to fit our style
guide?

[wanchang, 2017/03/08 00:35:35]

Done.

 
   m_isStarted = true;
   m_currentSource = currentInput;
   token.setBaseOffset(m_currentSource.numberOfCharactersConsumed() -
                       m_previousSource.length());
 }
 
 void HTMLSourceTracker::end(SegmentedString& currentInput,
                             HTMLTokenizer* tokenizer,
                             HTMLToken& token) {
   m_isStarted = false;
 
   m_cachedSourceForToken = String();
 
   // FIXME: This work should really be done by the HTMLTokenizer.
+  size_t numberOfBufferedCharacters = 0u;
+  if (needToCheckTokenizerBuffer(tokenizer)) {
+    numberOfBufferedCharacters = tokenizer->numberOfBufferedCharacters();
+  }
   token.end(currentInput.numberOfCharactersConsumed() -
-            tokenizer->numberOfBufferedCharacters());
+            numberOfBufferedCharacters);
 }
 
 String HTMLSourceTracker::sourceForToken(const HTMLToken& token) {
   if (!m_cachedSourceForToken.isEmpty())
     return m_cachedSourceForToken;
 
   size_t length;
   if (token.type() == HTMLToken::EndOfFile) {
     // Consume the remainder of the input, omitting the null character we use to
     // mark the end of the file.
@@ skipping 14 matching lines @@
   for (; i < length; ++i) {
     ASSERT(!m_currentSource.isEmpty());
     source.append(m_currentSource.currentChar());
     m_currentSource.advance();
   }
 
   m_cachedSourceForToken = source.toString();
   return m_cachedSourceForToken;
 }
 
+bool HTMLSourceTracker::needToCheckTokenizerBuffer(HTMLTokenizer* tokenizer) {
+  HTMLTokenizer::State state = tokenizer->getState();
+  // considering tokenizer temporarybuffer only if in the middle of

[Charlie Harrison, 2017/03/07 18:37:47]

Phrasing suggestion:
"Consider checking the tokenizer's temporary buffer if it is in an end tag
buffering state, or the token has finished parsing.

The temporary buffer must not be used unconditionally, because in some states
(e.g. ScriptDataDoubleEscapeStartState), data is appended to both the temporary
buffer and the token itself.
"

[wanchang, 2017/03/08 00:35:35]

Done.

+  // endtagbufferingstate or completed parsing the token
+  return state == HTMLTokenizer::DataState || isEndTagBufferingState(state);
+}
+
 }  // namespace blink