Move ownership of the TextResourceDecoder to DecodedDataDocumentParser

Source/core/dom/Document.cpp

 /*
  * Copyright (C) 1999 Lars Knoll (knoll@kde.org)
  *           (C) 1999 Antti Koivisto (koivisto@kde.org)
  *           (C) 2001 Dirk Mueller (mueller@kde.org)
  *           (C) 2006 Alexey Proskuryakov (ap@webkit.org)
  * Copyright (C) 2004, 2005, 2006, 2007, 2008, 2009, 2011, 2012 Apple Inc. All rights reserved.
  * Copyright (C) 2008, 2009 Torch Mobile Inc. All rights reserved. (http://www.torchmobile.com/)
  * Copyright (C) 2008, 2009, 2011, 2012 Google Inc. All rights reserved.
  * Copyright (C) 2010 Nokia Corporation and/or its subsidiary(-ies)
  * Copyright (C) Research In Motion Limited 2010-2011. All rights reserved.
@@ skipping 500 matching lines @@
     // generally keep at least one reference to an Element which would in turn
     // has a reference to the Document.  If you hit this ASSERT, then that
     // assumption is wrong.  DocumentParser::detach() should ensure that even
     // if the DocumentParser outlives the Document it won't cause badness.
     ASSERT(!m_parser || m_parser->refCount() == 1);
     detachParser();
 
     if (this == topDocument())
         clearAXObjectCache();
 
-    setDecoder(PassRefPtr<TextResourceDecoder>());
-
     if (m_styleSheetList)
         m_styleSheetList->detachFromDocument();
 
     if (m_import) {
         m_import->wasDetachedFromDocument();
         m_import = 0;
     }
 
     m_styleEngine.clear();
 
@@ skipping 560 matching lines @@
 bool Document::isLoadCompleted()
 {
     return m_readyState == Complete;
 }
 
 String Document::encodingName() const
 {
     // TextEncoding::name() returns a char*, no need to allocate a new
     // String for it each time.
     // FIXME: We should fix TextEncoding to speak AtomicString anyway.
-    return AtomicString(m_encoding.name());
+    return AtomicString(encoding().name());
 }
 
 String Document::defaultCharset() const
 {
     if (Settings* settings = this->settings())
         return settings->defaultTextEncodingName();
     return String();
 }
 
 void Document::setCharset(const String& charset)
 {
-    if (!decoder())
+    if (DocumentLoader* documentLoader = loader())
+        documentLoader->setUserChosenEncoding(charset);
+    WTF::TextEncoding encoding(charset);
+    // In case the encoding didn't exist, we keep the old one (helps some sites specifying invalid encodings).
+    if (!encoding.isValid())
         return;
-    decoder()->setEncoding(charset, TextResourceDecoder::UserChosenEncoding);
-    setEncoding(m_decoder->encoding());
+    DocumentEncodingData data;
+    data.encoding = encoding;
+    setEncodingData(data);

[abarth-chromium, 2013/11/14 07:04:33]

It's ok to clobber all the data?  We don't want to keep sawDecodingError?  Is
there a test that fails if we just write the encoding field?  If not, can we
write one to show that we're not changing behavior?

[oystein (OOO til 10th of July), 2013/11/14 19:02:29]

Unsure, maybe jamesr@ knows. I've modified this to preserve the old behavior.

 }
 
 void Document::setContentLanguage(const String& language)
 {
     if (m_contentLanguage == language)
         return;
     m_contentLanguage = language;
 
     // Document's style depends on the content language.
     setNeedsStyleRecalc();
@@ skipping 1972 matching lines @@
 {
     DocumentInit init(url());
     if (isXHTMLDocument())
         return createXHTML(init.withRegistrationContext(registrationContext()));
     return create(init);
 }
 
 void Document::cloneDataFromDocument(const Document& other)
 {
     setCompatibilityMode(other.compatibilityMode());
-    setEncoding(other.encoding());
+    setEncodingData(other.encodingData());

[abarth-chromium, 2013/11/14 07:04:33]

Notice that there's no reason to call encodingData here.  We can just get
m_encodingData from other directly.

[oystein (OOO til 10th of July), 2013/11/14 19:02:29]

Done.

     setContextFeatures(other.contextFeatures());
     setSecurityOrigin(other.securityOrigin()->isolatedCopy());
 }
 
 StyleSheetList* Document::styleSheets()
 {
     if (!m_styleSheetList)
         m_styleSheetList = StyleSheetList::create(this);
     return m_styleSheetList.get();
 }
@@ skipping 799 matching lines @@
     if (!length) {
         es.throwUninformativeAndGenericDOMException(InvalidCharacterError);
         return false;
     }
 
     if (qualifiedName.is8Bit())
         return parseQualifiedNameInternal(qualifiedName, qualifiedName.characters8(), length, prefix, localName, es);
     return parseQualifiedNameInternal(qualifiedName, qualifiedName.characters16(), length, prefix, localName, es);
 }
 
-void Document::setDecoder(PassRefPtr<TextResourceDecoder> decoder)
+void Document::setEncodingData(const DocumentEncodingData& newData)
 {
-    m_decoder = decoder;
-    setEncoding(m_decoder ? m_decoder->encoding() : WTF::TextEncoding());
-}
-
-void Document::setEncoding(const WTF::TextEncoding& encoding)
-{
-    if (m_encoding == encoding)
-        return;
-
     // It's possible for the encoding of the document to change while we're decoding
     // data. That can only occur while we're processing the <head> portion of the
     // document. There isn't much user-visible content in the <head>, but there is
     // the <title> element. This function detects that situation and re-decodes the
     // document's title so that the user doesn't see an incorrectly decoded title
     // in the title bar.
     if (m_titleElement
+        && encoding() != newData.encoding
         && !m_titleElement->firstElementChild()
-        && m_encoding == Latin1Encoding()
+        && encoding() == Latin1Encoding()
         && m_titleElement->textContent().containsOnlyLatin1()) {
 
         CString originalBytes = m_titleElement->textContent().latin1();
-        OwnPtr<TextCodec> codec = newTextCodec(encoding);
+        OwnPtr<TextCodec> codec = newTextCodec(newData.encoding);
         String correctlyDecodedTitle = codec->decode(originalBytes.data(), originalBytes.length(), true);
         m_titleElement->setTextContent(correctlyDecodedTitle, IGNORE_EXCEPTION);
     }
 
-    m_encoding = encoding;
+    m_encodingData = newData;
 }
 
 KURL Document::completeURL(const String& url, const KURL& baseURLOverride) const
 {
     // Always return a null URL when passed a null string.
     // FIXME: Should we change the KURL constructor to have this behavior?
     // See also [CSS]StyleSheet::completeURL(const String&)
     if (url.isNull())
         return KURL();
     const KURL& baseURL = ((baseURLOverride.isEmpty() || baseURLOverride == blankURL()) && parentDocument()) ? parentDocument()->baseURL() : baseURLOverride;
-    if (!m_decoder)
+    if (!encoding().isValid())
         return KURL(baseURL, url);
-    return KURL(baseURL, url, m_decoder->encoding());
+    return KURL(baseURL, url, encoding());
 }
 
 KURL Document::completeURL(const String& url) const
 {
     return completeURL(url, m_baseURL);
 }
 
 // Support for Javascript execCommand, and related methods
 
 static Editor::Command command(Document* document, const String& commandName, bool userInterface = false)
@@ skipping 1227 matching lines @@
 void Document::modifiedStyleSheet(StyleSheet* sheet, RecalcStyleTime when, StyleResolverUpdateMode updateMode)
 {
     if (!isActive())
         return;
 
     styleEngine()->modifiedStyleSheet(sheet);
     styleResolverChanged(when, updateMode);
 }
 
 } // namespace WebCore