Decouple tokenizer/tree builder state from WebVTTParser

Source/core/html/track/WebVTTParser.cpp

 /*
  * Copyright (C) 2011 Google Inc.  All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are
  * met:
  *
  *     * Redistributions of source code must retain the above copyright
  * notice, this list of conditions and the following disclaimer.
  *     * Redistributions in binary form must reproduce the above
@@ skipping 98 matching lines @@
     isValidSetting = isFirstValueValid && isSecondValueValid;
     return FloatPoint(firstCoord, secondCoord);
 }
 
 WebVTTParser::WebVTTParser(WebVTTParserClient* client, Document& document)
     : m_document(&document)
     , m_state(Initial)
     , m_decoder(TextResourceDecoder::create("text/plain", UTF8Encoding()))
     , m_currentStartTime(0)
     , m_currentEndTime(0)
-    , m_tokenizer(WebVTTTokenizer::create())
     , m_client(client)
 {
 }
 
 void WebVTTParser::getNewCues(Vector<RefPtr<TextTrackCue> >& outputCues)
 {
     outputCues = m_cuelist;
     m_cuelist.clear();
 }
 
@@ skipping 163 matching lines @@
     return CueText;
 }
 
 WebVTTParser::ParseState WebVTTParser::ignoreBadCue(const String& line)
 {
     if (!line.isEmpty())
         return BadCue;
     return Id;
 }
 
-PassRefPtr<DocumentFragment>  WebVTTParser::createDocumentFragmentFromCueText(const String& text)
+// A helper class for the construction of a "cue fragment" from the cue text.
+class WebVTTTreeBuilder {
+public:
+    WebVTTTreeBuilder(Document& document)
+        : m_document(document) { }
+
+    PassRefPtr<DocumentFragment> buildFromString(const String& cueText);
+
+private:
+    void constructTreeFromToken(Document&);
+
+    WebVTTToken m_token;
+    RefPtr<ContainerNode> m_currentNode;
+    Vector<AtomicString> m_languageStack;
+    Document& m_document;
+};
+
+PassRefPtr<DocumentFragment> WebVTTTreeBuilder::buildFromString(const String& cueText)
 {
     // Cue text processing based on
     // 4.8.10.13.4 WebVTT cue text parsing rules and
     // 4.8.10.13.5 WebVTT cue text DOM construction rules.
 
-    RefPtr<DocumentFragment> fragment = DocumentFragment::create(*m_document);
+    RefPtr<DocumentFragment> fragment = DocumentFragment::create(m_document);
 
-    if (!text.length()) {
-        fragment->parserAppendChild(Text::create(*m_document, ""));
+    if (cueText.isEmpty()) {
+        fragment->parserAppendChild(Text::create(m_document, ""));
         return fragment;
     }
 
     m_currentNode = fragment;
-    m_tokenizer->reset();
+
+    OwnPtr<WebVTTTokenizer> tokenizer(WebVTTTokenizer::create());
     m_token.clear();
+    m_languageStack.clear();
 
-    m_languageStack.clear();
-    SegmentedString content(text);
-    while (m_tokenizer->nextToken(content, m_token))
-        constructTreeFromToken(*m_document);
+    SegmentedString content(cueText);
+    while (tokenizer->nextToken(content, m_token))
+        constructTreeFromToken(m_document);
 
     return fragment.release();
 }
 
+PassRefPtr<DocumentFragment> WebVTTParser::createDocumentFragmentFromCueText(Document& document, const String& cueText)
+{
+    WebVTTTreeBuilder treeBuilder(document);
+    return treeBuilder.buildFromString(cueText);
+}
+
 void WebVTTParser::createNewCue()
 {
     if (!m_currentContent.length())
         return;
 
     RefPtr<TextTrackCue> cue = TextTrackCue::create(*m_document, m_currentStartTime, m_currentEndTime, m_currentContent.toString());
     cue->setId(m_currentId);
     cue->setCueSettings(m_currentSettings);
 
     m_cuelist.append(cue);
@@ skipping 111 matching lines @@
     case 4:
         if (token.name()[0] == 'r' && token.name()[1] == 'u' && token.name()[2] == 'b' && token.name()[3] == 'y')
             return WebVTTNodeTypeRuby;
         if (token.name()[0] == 'l' && token.name()[1] == 'a' && token.name()[2] == 'n' && token.name()[3] == 'g')
             return WebVTTNodeTypeLanguage;
         break;
     }
     return WebVTTNodeTypeNone;
 }
 
-void WebVTTParser::constructTreeFromToken(Document& document)
+void WebVTTTreeBuilder::constructTreeFromToken(Document& document)
 {
     QualifiedName tagName(nullAtom, AtomicString(m_token.name()), xhtmlNamespaceURI);
 
     // http://dev.w3.org/html5/webvtt/#webvtt-cue-text-dom-construction-rules
 
     switch (m_token.type()) {
     case WebVTTTokenTypes::Character: {
         String content(m_token.characters()); // FIXME: This should be 8bit if possible.
         RefPtr<Text> child = Text::create(document, content);
         m_currentNode->parserAppendChild(child);
@@ skipping 27 matching lines @@
             if (nodeType == WebVTTNodeTypeLanguage && m_currentNode->isWebVTTElement() && toWebVTTElement(m_currentNode.get())->webVTTNodeType() == WebVTTNodeTypeLanguage)
                 m_languageStack.removeLast();
             if (m_currentNode->parentNode())
                 m_currentNode = m_currentNode->parentNode();
         }
         break;
     }
     case WebVTTTokenTypes::TimestampTag: {
         unsigned position = 0;
         String charactersString(StringImpl::create8BitIfPossible(m_token.characters()));
-        double time = collectTimeStamp(charactersString, &position);
+        double time = WebVTTParser::collectTimeStamp(charactersString, &position);
         if (time != malformedTime)
             m_currentNode->parserAppendChild(ProcessingInstruction::create(document, "timestamp", charactersString));
         break;
     }
     default:
         break;
     }
     m_token.clear();
 }
 
@@ skipping 20 matching lines @@
     unsigned oldPosition = *position;
     while (*position < data.length() && data[*position] != '\r' && data[*position] != '\n')
         (*position)++;
     String line = data.substring(oldPosition, *position - oldPosition);
     skipLineTerminator(data, position);
     return line;
 }
 
 }