Bug 1807017 - Consider to reuse the memory when nsHtml5Tokenizer::end() is called, r=hsivonen

Differential Revision: https://phabricator.services.mozilla.com/D201982

parser/html/javasrc/Tokenizer.java

@@ -511,6 +511,8 @@ public class Tokenizer implements Locator, Locator2 {
 
     private boolean shouldSuspend;
 
+    private boolean keepBuffer;
+
     protected boolean confident;
 
     private int line;
@@ -570,6 +572,7 @@ public class Tokenizer implements Locator, Locator2 {
         this.systemIdentifier = null;
         this.attributes = null;
         this.shouldSuspend = false;
+        this.keepBuffer = false;
         this.confident = false;
         this.line = 0;
         // CPPONLY: this.attributeLine = 0;
@@ -632,6 +635,7 @@ public class Tokenizer implements Locator, Locator2 {
         // CPPONLY: this.attributes = tokenHandler.HasBuilder() ? new HtmlAttributes(mappingLangToXmlLang) : null;
         // CPPONLY: this.newAttributesEachTime = !tokenHandler.HasBuilder();
         this.shouldSuspend = false;
+        this.keepBuffer = false;
         this.confident = false;
         this.line = 0;
         // CPPONLY: this.attributeLine = 0;
@@ -653,6 +657,18 @@ public class Tokenizer implements Locator, Locator2 {
     // CPPONLY: return viewingXmlSource;
     // CPPONLY: }
 
+    public void setKeepBuffer(boolean keepBuffer) {
+        this.keepBuffer = keepBuffer;
+    }
+
+    public boolean dropBufferIfLongerThan(int length) {
+        if (strBuf.length > length) {
+            strBuf = null;
+            return true;
+        }
+        return false;
+    }
+
     // [NOCPP[
 
     /**
@@ -7225,7 +7241,9 @@ public class Tokenizer implements Locator, Locator2 {
     }
 
     public void end() throws SAXException {
-        strBuf = null;
+        if (!keepBuffer) {
+            strBuf = null;
+        }
         doctypeName = null;
         if (systemIdentifier != null) {
             Portability.releaseString(systemIdentifier);
@@ -7415,7 +7433,9 @@ public class Tokenizer implements Locator, Locator2 {
 
     public void initializeWithoutStarting() throws SAXException {
         confident = false;
-        strBuf = null;
+        if (!keepBuffer) {
+            strBuf = null;
+        }
         line = 1;
         // CPPONLY: attributeLine = 1;
         // [NOCPP[

parser/html/javasrc/TreeBuilder.java

@@ -436,6 +436,8 @@ public abstract class TreeBuilder<T> implements TokenHandler,
 
     private boolean allowDeclarativeShadowRoots = false;
 
+    private boolean keepBuffer = false;
+
     // [NOCPP[
 
     private boolean reportingDoctype = true;
@@ -577,6 +579,18 @@ public abstract class TreeBuilder<T> implements TokenHandler,
 
     // ]NOCPP]
 
+    public void setKeepBuffer(boolean keepBuffer) {
+        this.keepBuffer = keepBuffer;
+    }
+
+    public boolean dropBufferIfLongerThan(int length) {
+        if (charBuffer.length > length) {
+            charBuffer = null;
+            return true;
+        }
+        return false;
+    }
+
     @SuppressWarnings("unchecked") public final void startTokenization(Tokenizer self) throws SAXException {
         tokenizer = self;
         stackNodes = new StackNode[64];
@@ -598,7 +612,9 @@ public abstract class TreeBuilder<T> implements TokenHandler,
         // ]NOCPP]
         start(fragment);
         charBufferLen = 0;
-        charBuffer = null;
+        if (!keepBuffer) {
+            charBuffer = null;
+        }
         framesetOk = true;
         if (fragment) {
             T elt;
@@ -1451,7 +1467,10 @@ public abstract class TreeBuilder<T> implements TokenHandler,
         // [NOCPP[
         idLocations.clear();
         // ]NOCPP]
-        charBuffer = null;
+
+        if (!keepBuffer) {
+            charBuffer = null;
+        }
         end();
     }
 

parser/html/nsHtml5StringParser.cpp

@@ -20,9 +20,11 @@ nsHtml5StringParser::nsHtml5StringParser()
       mTreeBuilder(new nsHtml5TreeBuilder(mBuilder)),
       mTokenizer(new nsHtml5Tokenizer(mTreeBuilder.get(), false)) {
   mTokenizer->setInterner(&mAtomTable);
+  mTokenizer->setKeepBuffer(true);
+  mTreeBuilder->setKeepBuffer(true);
 }
 
-nsHtml5StringParser::~nsHtml5StringParser() {}
+nsHtml5StringParser::~nsHtml5StringParser() { ClearCaches(); }
 
 nsresult nsHtml5StringParser::ParseFragment(
     const nsAString& aSourceBuffer, nsIContent* aTargetNode,
@@ -69,6 +71,31 @@ nsresult nsHtml5StringParser::ParseDocument(
                   aTargetDoc->AllowsDeclarativeShadowRoots());
 }
 
+void nsHtml5StringParser::ClearCaches() {
+  mTokenizer->dropBufferIfLongerThan(0);
+  mTreeBuilder->dropBufferIfLongerThan(0);
+  if (mCacheClearer) {
+    mCacheClearer->Disconnect();
+    mCacheClearer = nullptr;
+  }
+}
+
+void nsHtml5StringParser::TryCache() {
+  const int32_t kMaxBuffer = 1024 * 1024;
+  bool didDrop = mTokenizer->dropBufferIfLongerThan(kMaxBuffer);
+  didDrop |= mTreeBuilder->dropBufferIfLongerThan(kMaxBuffer);
+  if (didDrop) {
+    return;
+  }
+
+  if (!mCacheClearer) {
+    mCacheClearer = new CacheClearer(this);
+    nsCOMPtr<nsIRunnable> runnable = mCacheClearer.get();
+    NS_DispatchToMainThreadQueue(runnable.forget(),
+                                 mozilla::EventQueuePriority::Idle);
+  }
+}
+
 nsresult nsHtml5StringParser::Tokenize(const nsAString& aSourceBuffer,
                                        Document* aDocument,
                                        bool aScriptingEnabledForNoscriptParsing,
@@ -109,8 +136,10 @@ nsresult nsHtml5StringParser::Tokenize(const nsAString& aSourceBuffer,
   if (NS_SUCCEEDED(rv)) {
     mTokenizer->eof();
   }
+
   mTokenizer->end();
   mBuilder->Finish();
   mAtomTable.Clear();
+  TryCache();
   return rv;
 }

parser/html/nsHtml5StringParser.h

@@ -67,6 +67,9 @@ class nsHtml5StringParser : public nsParserBase {
                     bool aScriptingEnabledForNoscriptParsing,
                     bool aDeclarativeShadowRootsAllowed);
 
+  void TryCache();
+  void ClearCaches();
+
   /**
    * The tree operation executor
    */
@@ -86,6 +89,24 @@ class nsHtml5StringParser : public nsParserBase {
    * The scoped atom table
    */
   nsHtml5AtomTable mAtomTable;
+
+  class CacheClearer : public mozilla::Runnable {
+   public:
+    explicit CacheClearer(nsHtml5StringParser* aParser)
+        : Runnable("CacheClearer"), mParser(aParser) {}
+    NS_IMETHOD Run() {
+      if (mParser) {
+        mParser->ClearCaches();
+      }
+      return NS_OK;
+    }
+    void Disconnect() { mParser = nullptr; }
+
+   private:
+    nsHtml5StringParser* mParser;
+  };
+
+  RefPtr<CacheClearer> mCacheClearer;
 };
 
 #endif  // nsHtml5StringParser_h

parser/html/nsHtml5Tokenizer.cpp

@@ -126,6 +126,7 @@ nsHtml5Tokenizer::nsHtml5Tokenizer(nsHtml5TreeBuilder* tokenHandler,
                                             : nullptr),
       newAttributesEachTime(!tokenHandler->HasBuilder()),
       shouldSuspend(false),
+      keepBuffer(false),
       confident(false),
       line(0),
       attributeLine(0),
@@ -146,6 +147,18 @@ void nsHtml5Tokenizer::initLocation(nsHtml5String newPublicId,
 
 bool nsHtml5Tokenizer::isViewingXmlSource() { return viewingXmlSource; }
 
+void nsHtml5Tokenizer::setKeepBuffer(bool keepBuffer) {
+  this->keepBuffer = keepBuffer;
+}
+
+bool nsHtml5Tokenizer::dropBufferIfLongerThan(int32_t length) {
+  if (strBuf.length > length) {
+    strBuf = nullptr;
+    return true;
+  }
+  return false;
+}
+
 void nsHtml5Tokenizer::setState(int32_t specialTokenizerState) {
   this->stateSave = specialTokenizerState;
   this->endTagExpectation = nullptr;
@@ -5022,7 +5035,9 @@ void nsHtml5Tokenizer::emitOrAppendOne(const char16_t* val,
 }
 
 void nsHtml5Tokenizer::end() {
-  strBuf = nullptr;
+  if (!keepBuffer) {
+    strBuf = nullptr;
+  }
   doctypeName = nullptr;
   if (systemIdentifier) {
     systemIdentifier.Release();
@@ -5148,7 +5163,9 @@ void nsHtml5Tokenizer::loadState(nsHtml5Tokenizer* other) {
 
 void nsHtml5Tokenizer::initializeWithoutStarting() {
   confident = false;
-  strBuf = nullptr;
+  if (!keepBuffer) {
+    strBuf = nullptr;
+  }
   line = 1;
   attributeLine = 1;
   resetToDataState();

parser/html/nsHtml5Tokenizer.h

@@ -308,6 +308,7 @@ class nsHtml5Tokenizer {
   nsHtml5HtmlAttributes* attributes;
   bool newAttributesEachTime;
   bool shouldSuspend;
+  bool keepBuffer;
 
  protected:
   bool confident;
@@ -323,6 +324,8 @@ class nsHtml5Tokenizer {
   void setInterner(nsHtml5AtomTable* interner);
   void initLocation(nsHtml5String newPublicId, nsHtml5String newSystemId);
   bool isViewingXmlSource();
+  void setKeepBuffer(bool keepBuffer);
+  bool dropBufferIfLongerThan(int32_t length);
   void setState(int32_t specialTokenizerState);
   void setStateAndEndTagExpectation(int32_t specialTokenizerState,
                                     nsHtml5ElementName* endTagExpectation);

parser/html/nsHtml5TreeBuilder.cpp

@@ -100,6 +100,18 @@ static const char* const QUIRKY_PUBLIC_IDS_DATA[] = {
     "-//webtechs//dtd mozilla html//"};
 staticJArray<const char*, int32_t> nsHtml5TreeBuilder::QUIRKY_PUBLIC_IDS = {
     QUIRKY_PUBLIC_IDS_DATA, MOZ_ARRAY_LENGTH(QUIRKY_PUBLIC_IDS_DATA)};
+void nsHtml5TreeBuilder::setKeepBuffer(bool keepBuffer) {
+  this->keepBuffer = keepBuffer;
+}
+
+bool nsHtml5TreeBuilder::dropBufferIfLongerThan(int32_t length) {
+  if (charBuffer.length > length) {
+    charBuffer = nullptr;
+    return true;
+  }
+  return false;
+}
+
 void nsHtml5TreeBuilder::startTokenization(nsHtml5Tokenizer* self) {
   tokenizer = self;
   stackNodes = jArray<nsHtml5StackNode*, int32_t>::newJArray(64);
@@ -118,7 +130,9 @@ void nsHtml5TreeBuilder::startTokenization(nsHtml5Tokenizer* self) {
   headPointer = nullptr;
   start(fragment);
   charBufferLen = 0;
-  charBuffer = nullptr;
+  if (!keepBuffer) {
+    charBuffer = nullptr;
+  }
   framesetOk = true;
   if (fragment) {
     nsIContentHandle* elt;
@@ -652,7 +666,9 @@ void nsHtml5TreeBuilder::endTokenization() {
     stackNodesIdx = 0;
     stackNodes = nullptr;
   }
-  charBuffer = nullptr;
+  if (!keepBuffer) {
+    charBuffer = nullptr;
+  }
   end();
 }
 

parser/html/nsHtml5TreeBuilder.h

@@ -315,6 +315,7 @@ class nsHtml5TreeBuilder : public nsAHtml5TreeBuilderState {
   bool quirks;
   bool forceNoQuirks;
   bool allowDeclarativeShadowRoots;
+  bool keepBuffer;
   inline nsHtml5ContentCreatorFunction htmlCreator(
       mozilla::dom::HTMLContentCreatorFunction htmlCreator) {
     nsHtml5ContentCreatorFunction creator;
@@ -330,6 +331,8 @@ class nsHtml5TreeBuilder : public nsAHtml5TreeBuilderState {
   }
 
  public:
+  void setKeepBuffer(bool keepBuffer);
+  bool dropBufferIfLongerThan(int32_t length);
   void startTokenization(nsHtml5Tokenizer* self);
   void doctype(nsAtom* name, nsHtml5String publicIdentifier,
                nsHtml5String systemIdentifier, bool forceQuirks);

parser/html/nsHtml5TreeBuilderCppSupplement.h

@@ -42,6 +42,7 @@ nsHtml5TreeBuilder::nsHtml5TreeBuilder(nsHtml5OplessBuilder* aBuilder)
       quirks(false),
       forceNoQuirks(false),
       allowDeclarativeShadowRoots(false),
+      keepBuffer(false),
       mBuilder(aBuilder),
       mViewSource(nullptr),
       mOpSink(nullptr),
@@ -85,6 +86,7 @@ nsHtml5TreeBuilder::nsHtml5TreeBuilder(nsAHtml5TreeOpSink* aOpSink,
       quirks(false),
       forceNoQuirks(false),
       allowDeclarativeShadowRoots(false),
+      keepBuffer(false),
       mBuilder(nullptr),
       mViewSource(nullptr),
       mOpSink(aOpSink),