third_party/WebKit/Source/core/html/parser/HTMLDocumentParser.h

/*
 * Copyright (C) 2010 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:
 * 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
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY
 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL APPLE INC. OR
 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
 * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

#ifndef HTMLDocumentParser_h
#define HTMLDocumentParser_h

#include "core/CoreExport.h"
#include "core/dom/ParserContentPolicy.h"
#include "core/dom/ScriptableDocumentParser.h"
#include "core/html/parser/BackgroundHTMLInputStream.h"
#include "core/html/parser/HTMLInputStream.h"
#include "core/html/parser/HTMLParserOptions.h"
#include "core/html/parser/HTMLParserReentryPermit.h"
#include "core/html/parser/HTMLPreloadScanner.h"
#include "core/html/parser/HTMLScriptRunnerHost.h"
#include "core/html/parser/HTMLSourceTracker.h"
#include "core/html/parser/HTMLToken.h"
#include "core/html/parser/HTMLTokenizer.h"
#include "core/html/parser/HTMLTreeBuilderSimulator.h"
#include "core/html/parser/ParserSynchronizationPolicy.h"
#include "core/html/parser/TextResourceDecoder.h"
#include "core/html/parser/XSSAuditor.h"
#include "core/html/parser/XSSAuditorDelegate.h"
#include "wtf/Deque.h"
#include "wtf/RefPtr.h"
#include "wtf/WeakPtr.h"
#include "wtf/text/TextPosition.h"
#include <memory>

namespace blink {

class BackgroundHTMLParser;
class CompactHTMLToken;
class Document;
class DocumentEncodingData;
class DocumentFragment;
class Element;
class HTMLDocument;
class HTMLParserScheduler;
class HTMLPreloadScanner;
class HTMLResourcePreloader;
class HTMLScriptRunner;
class HTMLTreeBuilder;
class PumpSession;
class SegmentedString;
class TokenizedChunkQueue;
class DocumentWriteEvaluator;

class CORE_EXPORT HTMLDocumentParser : public ScriptableDocumentParser,
                                       private HTMLScriptRunnerHost {
  USING_GARBAGE_COLLECTED_MIXIN(HTMLDocumentParser);
  USING_PRE_FINALIZER(HTMLDocumentParser, dispose);

 public:
  static HTMLDocumentParser* create(
      HTMLDocument& document,
      ParserSynchronizationPolicy backgroundParsingPolicy) {
    return new HTMLDocumentParser(document, backgroundParsingPolicy);
  }
  ~HTMLDocumentParser() override;
  DECLARE_VIRTUAL_TRACE();

  // TODO(alexclarke): Remove when background parser goes away.
  void dispose();

  // Exposed for HTMLParserScheduler
  void resumeParsingAfterYield();

  static void parseDocumentFragment(
      const String&,
      DocumentFragment*,
      Element* contextElement,
      ParserContentPolicy = AllowScriptingContent);

  // Exposed for testing.
  HTMLScriptRunnerHost* asHTMLScriptRunnerHostForTesting() { return this; }

  HTMLTokenizer* tokenizer() const { return m_tokenizer.get(); }

  TextPosition textPosition() const final;
  bool isParsingAtLineNumber() const final;
  OrdinalNumber lineNumber() const final;

  void suspendScheduledTasks() final;
  void resumeScheduledTasks() final;

  HTMLParserReentryPermit* reentryPermit() { return m_reentryPermit.get(); }

  struct TokenizedChunk {
    USING_FAST_MALLOC(TokenizedChunk);

   public:
    std::unique_ptr<CompactHTMLTokenStream> tokens;
    PreloadRequestStream preloads;
    ViewportDescriptionWrapper viewport;
    XSSInfoStream xssInfos;
    HTMLTokenizer::State tokenizerState;
    HTMLTreeBuilderSimulator::State treeBuilderState;
    HTMLInputCheckpoint inputCheckpoint;
    TokenPreloadScannerCheckpoint preloadScannerCheckpoint;
    bool startingScript;
    // Indices into |tokens|.
    Vector<int> likelyDocumentWriteScriptIndices;
    // Index into |tokens| of the last <meta> csp tag in |tokens|. Preloads will
    // be deferred until this token is parsed. Will be noPendingToken if there
    // are no csp tokens.
    int pendingCSPMetaTokenIndex;

    static constexpr int noPendingToken = -1;
  };
  void notifyPendingTokenizedChunks();
  void didReceiveEncodingDataFromBackgroundParser(const DocumentEncodingData&);

  void appendBytes(const char* bytes, size_t length) override;
  void flush() final;
  void setDecoder(std::unique_ptr<TextResourceDecoder>) final;

 protected:
  void insert(const SegmentedString&) final;
  void append(const String&) override;
  void finish() final;

  HTMLDocumentParser(HTMLDocument&, ParserSynchronizationPolicy);
  HTMLDocumentParser(DocumentFragment*,
                     Element* contextElement,
                     ParserContentPolicy);

  HTMLTreeBuilder* treeBuilder() const { return m_treeBuilder.get(); }

  void forcePlaintextForTextDocument();

 private:
  static HTMLDocumentParser* create(DocumentFragment* fragment,
                                    Element* contextElement,
                                    ParserContentPolicy parserContentPolicy) {
    return new HTMLDocumentParser(fragment, contextElement,
                                  parserContentPolicy);
  }
  HTMLDocumentParser(Document&,
                     ParserContentPolicy,
                     ParserSynchronizationPolicy);

  // DocumentParser
  void detach() final;
  bool hasInsertionPoint() final;
  void prepareToStopParsing() final;
  void stopParsing() final;
  bool isWaitingForScripts() const final;
  bool isExecutingScript() const final;
  void executeScriptsWaitingForResources() final;
  void documentElementAvailable() override;

  // HTMLScriptRunnerHost
  void notifyScriptLoaded(Resource*) final;
  HTMLInputStream& inputStream() final { return m_input; }
  bool hasPreloadScanner() const final {
    return m_preloadScanner.get() && !shouldUseThreading();
  }
  void appendCurrentInputStreamToPreloadScannerAndScan() final;

  void startBackgroundParser();
  void stopBackgroundParser();
  void validateSpeculations(std::unique_ptr<TokenizedChunk> lastChunk);
  void discardSpeculationsAndResumeFrom(
      std::unique_ptr<TokenizedChunk> lastChunk,
      std::unique_ptr<HTMLToken>,
      std::unique_ptr<HTMLTokenizer>);
  size_t processTokenizedChunkFromBackgroundParser(
      std::unique_ptr<TokenizedChunk>);
  void pumpPendingSpeculations();

  bool canTakeNextToken();
  void pumpTokenizer();
  void pumpTokenizerIfPossible();
  void constructTreeFromHTMLToken();
  void constructTreeFromCompactHTMLToken(const CompactHTMLToken&);

  void runScriptsForPausedTreeBuilder();
  void resumeParsingAfterScriptExecution();

  void attemptToEnd();
  void endIfDelayed();
  void attemptToRunDeferredScriptsAndEnd();
  void end();

  bool shouldUseThreading() const { return m_shouldUseThreading; }

  bool isParsingFragment() const;
  bool isScheduledForResume() const;
  bool inPumpSession() const { return m_pumpSessionNestingLevel > 0; }
  bool shouldDelayEnd() const {
    return inPumpSession() || isWaitingForScripts() || isScheduledForResume() ||
           isExecutingScript();
  }

  std::unique_ptr<HTMLPreloadScanner> createPreloadScanner();

  void fetchQueuedPreloads();

  void evaluateAndPreloadScriptForDocumentWrite(const String& source);

  // Temporary enum for the ParseHTMLOnMainThread experiment. This is used to
  // annotate whether a given task should post a task or not on the main thread
  // if the lookahead parser is living on the main thread.
  enum LookaheadParserTaskSynchrony {
    Synchronous,
    Asynchronous,
  };

  // Setting |synchronyPolicy| to Synchronous will just call the function with
  // the given parameters. Note, this method is completely temporary as we need
  // to maintain both threading implementations until the ParseHTMLOnMainThread
  // experiment finishes.
  template <typename FunctionType, typename... Ps>
  void postTaskToLookaheadParser(LookaheadParserTaskSynchrony synchronyPolicy,
                                 FunctionType,
                                 Ps&&... parameters);

  HTMLToken& token() { return *m_token; }

  HTMLParserOptions m_options;
  HTMLInputStream m_input;
  RefPtr<HTMLParserReentryPermit> m_reentryPermit;

  std::unique_ptr<HTMLToken> m_token;
  std::unique_ptr<HTMLTokenizer> m_tokenizer;
  Member<HTMLScriptRunner> m_scriptRunner;
  Member<HTMLTreeBuilder> m_treeBuilder;
  std::unique_ptr<HTMLPreloadScanner> m_preloadScanner;
  std::unique_ptr<HTMLPreloadScanner> m_insertionPreloadScanner;
  std::unique_ptr<WebTaskRunner> m_loadingTaskRunner;
  Member<HTMLParserScheduler> m_parserScheduler;
  HTMLSourceTracker m_sourceTracker;
  TextPosition m_textPosition;
  XSSAuditor m_xssAuditor;
  XSSAuditorDelegate m_xssAuditorDelegate;

  // FIXME: m_lastChunkBeforeScript, m_tokenizer, m_token, and m_input should be
  // combined into a single state object so they can be set and cleared together
  // and passed between threads together.
  std::unique_ptr<TokenizedChunk> m_lastChunkBeforeScript;
  Deque<std::unique_ptr<TokenizedChunk>> m_speculations;
  WeakPtrFactory<HTMLDocumentParser> m_weakFactory;
  WeakPtr<BackgroundHTMLParser> m_backgroundParser;
  Member<HTMLResourcePreloader> m_preloader;
  PreloadRequestStream m_queuedPreloads;
  Vector<String> m_queuedDocumentWriteScripts;
  RefPtr<TokenizedChunkQueue> m_tokenizedChunkQueue;
  std::unique_ptr<DocumentWriteEvaluator> m_evaluator;

  // If this is non-null, then there is a meta CSP token somewhere in the
  // speculation buffer. Preloads will be deferred until a token matching this
  // pointer is parsed and the CSP policy is applied. Note that this pointer
  // tracks the *last* meta token in the speculation buffer, so it overestimates
  // how long to defer preloads. This is for simplicity, as the alternative
  // would require keeping track of token positions of preload requests.
  CompactHTMLToken* m_pendingCSPMetaToken;

  bool m_shouldUseThreading;
  bool m_endWasDelayed;
  bool m_haveBackgroundParser;
  bool m_tasksWereSuspended;
  unsigned m_pumpSessionNestingLevel;
  unsigned m_pumpSpeculationsSessionNestingLevel;
  bool m_isParsingAtLineNumber;
  bool m_triedLoadingLinkHeaders;
};

}  // namespace blink

#endif  // HTMLDocumentParser_h