Bug 404386 - nsScanner construction is 5% of setting innerHTML. r=mrbkap,smontagu. sr=jst. a=blocking1.9.

intl/uconv/src/nsCharsetAliasImp.cpp

@@ -63,6 +63,16 @@ nsCharsetAlias2::~nsCharsetAlias2()
   if(mDelegate)
      delete mDelegate;
 }
+
+// 
+static const char* kAliases[][3] = {
+  // Triple with { lower-case test string, out string, length of out string }
+  { "iso-8859-1", "ISO-8859-1", (const char*)10 },
+  { "utf-8",      "UTF-8",      (const char*)5  },
+  { "x-sjis",     "Shift_JIS",  (const char*)9  },
+  { "shift_jis",  "Shift_JIS",  (const char*)9  }
+};
+
 //--------------------------------------------------------------
 NS_IMETHODIMP nsCharsetAlias2::GetPreferred(const nsACString& aAlias,
                                             nsACString& oResult)
@@ -70,35 +80,26 @@ NS_IMETHODIMP nsCharsetAlias2::GetPreferred(const nsACString& aAlias,
    if (aAlias.IsEmpty()) return NS_ERROR_NULL_POINTER;
    NS_TIMELINE_START_TIMER("nsCharsetAlias2:GetPreferred");
 
-   nsCAutoString aKey(aAlias);
-   ToLowerCase(aKey);
-   oResult.Truncate();
 
    // Delay loading charsetalias.properties by hardcoding the most
    // frequent aliases.  Note that it's possible to recur in to this
    // function *while loading* charsetalias.properties (see bug 190951),
    // so we might have an |mDelegate| already that isn't valid yet, but
    // the load is guaranteed to be "UTF-8" so things will be OK.
-   if(aKey.EqualsLiteral("utf-8")) {
-     oResult.AssignLiteral("UTF-8");
-     NS_TIMELINE_STOP_TIMER("nsCharsetAlias2:GetPreferred");
-     return NS_OK;
-   } 
-   if(aKey.EqualsLiteral("iso-8859-1")) {
-     oResult.AssignLiteral("ISO-8859-1");
-     NS_TIMELINE_STOP_TIMER("nsCharsetAlias2:GetPreferred");
-     return NS_OK;
-   } 
-   if(aKey.EqualsLiteral("x-sjis") ||
-      aKey.EqualsLiteral("shift_jis")) {
-     oResult.AssignLiteral("Shift_JIS");
-     NS_TIMELINE_STOP_TIMER("nsCharsetAlias2:GetPreferred");
-     return NS_OK;
-   } 
+   for (PRUint32 index = 0; index < NS_ARRAY_LENGTH(kAliases); index++) {
+     if (aAlias.LowerCaseEqualsASCII(kAliases[index][0])) {
+       oResult.Assign(nsDependentCString(kAliases[index][1],
+                                         (PRUint32)kAliases[index][2]));
+       NS_TIMELINE_STOP_TIMER("nsCharsetAlias2:GetPreferred");
+       return NS_OK;
+     }
+   }
+
+   oResult.Truncate();
 
    if(!mDelegate) {
      //load charsetalias.properties string bundle with all remaining aliases
-     // we may need to protect the following section with a lock so we won't call the 
+     // we may need to protect the following section with a lock so we won't call the
      // 'new nsGREResProperties' from two different threads
      mDelegate = new nsGREResProperties( NS_LITERAL_CSTRING("charsetalias.properties") );
      NS_ASSERTION(mDelegate, "cannot create nsGREResProperties");
@@ -109,10 +110,13 @@ NS_IMETHODIMP nsCharsetAlias2::GetPreferred(const nsACString& aAlias,
    NS_TIMELINE_STOP_TIMER("nsCharsetAlias2:GetPreferred");
    NS_TIMELINE_MARK_TIMER("nsCharsetAlias2:GetPreferred");
 
+   nsCAutoString key(aAlias);
+   ToLowerCase(key);
+
    // hack for now, have to fix nsGREResProperties, but we can't until
    // string bundles use UTF8 keys
    nsAutoString result;
-   nsresult rv = mDelegate->Get(NS_ConvertASCIItoUTF16(aKey), result);
+   nsresult rv = mDelegate->Get(NS_ConvertASCIItoUTF16(key), result);
    LossyAppendUTF16toASCII(result, oResult);
    return rv;
 }

intl/uconv/src/nsCharsetConverterManager.cpp

@@ -86,6 +86,12 @@ nsCharsetConverterManager::~nsCharsetConverterManager()
   NS_IF_RELEASE(mTitleBundle);
 }
 
+nsresult nsCharsetConverterManager::Init()
+{
+  if (!mDecoderHash.Init())
+    return NS_ERROR_OUT_OF_MEMORY;
+  return NS_OK;
+}
 
 nsresult nsCharsetConverterManager::RegisterConverterManagerData()
 {
@@ -254,30 +260,28 @@ nsCharsetConverterManager::GetUnicodeDecoderRaw(const char * aSrc,
 #endif
   nsresult rv = NS_OK;
 
-  NS_NAMED_LITERAL_CSTRING(kUnicodeDecoderContractIDBase,
-                           NS_UNICODEDECODER_CONTRACTID_BASE);
-
-  nsCAutoString contractid(kUnicodeDecoderContractIDBase +
-                           nsDependentCString(aSrc));
-
-  if (!strncmp(aSrc,
-               NS_1BYTE_CODER_PATTERN,
-               NS_1BYTE_CODER_PATTERN_LEN))
+  NS_NAMED_LITERAL_CSTRING(contractbase, NS_UNICODEDECODER_CONTRACTID_BASE);
+  nsDependentCString src(aSrc);
+  
+  if (!strncmp(aSrc, NS_1BYTE_CODER_PATTERN, NS_1BYTE_CODER_PATTERN_LEN))
   {
-    // Single byte decoders dont hold state. Optimize by using a service.
-    decoder = do_GetService(contractid.get(), &rv);
+    // Single byte decoders don't hold state. Optimize by using a service, and
+    // cache it in our hash to avoid repeated trips through the service manager.
+    if (!mDecoderHash.Get(aSrc, getter_AddRefs(decoder))) {
+      decoder = do_GetService(PromiseFlatCString(contractbase + src).get(),
+                              &rv);
+      if (NS_SUCCEEDED(rv))
+        mDecoderHash.Put(aSrc, decoder);
+    }
   }
   else
   {
-    decoder = do_CreateInstance(contractid.get(), &rv);
-  }
-  if(NS_FAILED(rv))
-    rv = NS_ERROR_UCONV_NOCONV;
-  else
-  {
-    *aResult = decoder.get();
-    NS_ADDREF(*aResult);
+    decoder = do_CreateInstance(PromiseFlatCString(contractbase + src).get(),
+                                &rv);
   }
+  NS_ENSURE_SUCCESS(rv, NS_ERROR_UCONV_NOCONV);
+
+  decoder.forget(aResult);
   return rv;
 }
 

intl/uconv/src/nsCharsetConverterManager.h

@@ -40,6 +40,7 @@
 #include "nsISupports.h"
 #include "nsICharsetConverterManager.h"
 #include "nsIStringBundle.h"
+#include "nsInterfaceHashtable.h"
 
 #ifdef MOZ_USE_NATIVE_UCONV
 #include "nsINativeUConvService.h"
@@ -55,6 +56,8 @@ public:
   nsCharsetConverterManager();
   virtual ~nsCharsetConverterManager();
 
+  nsresult Init();
+
 private:
 
   nsIStringBundle * mDataBundle;
@@ -64,6 +67,8 @@ private:
   nsCOMPtr<nsINativeUConvService> mNativeUC;
 #endif
 
+  nsInterfaceHashtable<nsCharPtrHashKey, nsIUnicodeDecoder> mDecoderHash;
+
   nsresult LoadExtensibleBundle(const char * aRegistryKey, 
       nsIStringBundle ** aResult);
 

intl/uconv/src/nsUConvModule.cpp

@@ -610,7 +610,7 @@ nsConverterManagerDataRegister(nsIComponentManager* aCompMgr,
   return nsCharsetConverterManager::RegisterConverterManagerData();
 }
 
-NS_GENERIC_FACTORY_CONSTRUCTOR(nsCharsetConverterManager)
+NS_GENERIC_FACTORY_CONSTRUCTOR_INIT(nsCharsetConverterManager, Init)
 NS_GENERIC_FACTORY_CONSTRUCTOR(nsTextToSubURI)
 NS_GENERIC_FACTORY_CONSTRUCTOR(nsUTF8ConverterService)
 NS_GENERIC_FACTORY_CONSTRUCTOR(nsCharsetAlias2)

parser/htmlparser/src/nsParser.cpp

@@ -48,6 +48,7 @@
 #include "nsICachingChannel.h"
 #include "nsICacheEntryDescriptor.h"
 #include "nsICharsetAlias.h"
+#include "nsICharsetConverterManager.h"
 #include "nsIInputStream.h"
 #include "CNavDTD.h"
 #include "prenv.h"
@@ -154,6 +155,9 @@ public:
 
 //-------------- End ParseContinue Event Definition ------------------------
 
+nsICharsetAlias* nsParser::sCharsetAliasService = nsnull;
+nsICharsetConverterManager* nsParser::sCharsetConverterManager = nsnull;
+
 /**
  *  This gets called when the htmlparser module is initialized.
  */
@@ -204,6 +208,17 @@ nsParser::Init()
     }
   }
 
+  nsCOMPtr<nsICharsetAlias> charsetAlias =
+    do_GetService(NS_CHARSETALIAS_CONTRACTID, &rv);
+  NS_ENSURE_SUCCESS(rv, rv);
+  
+  nsCOMPtr<nsICharsetConverterManager> charsetConverter =
+    do_GetService(NS_CHARSETCONVERTERMANAGER_CONTRACTID, &rv);
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  charsetAlias.swap(sCharsetAliasService);
+  charsetConverter.swap(sCharsetConverterManager);
+
   return NS_OK;
 }
 
@@ -216,9 +231,11 @@ void nsParser::Shutdown()
 {
   delete sParserDataListeners;
   sParserDataListeners = nsnull;
+  
+  NS_IF_RELEASE(sCharsetAliasService);
+  NS_IF_RELEASE(sCharsetConverterManager);
 }
 
-
 #ifdef DEBUG
 static PRBool gDumpContent=PR_FALSE;
 #endif

parser/htmlparser/src/nsParser.h

@@ -90,6 +90,8 @@
 #include "nsIUnicharStreamListener.h"
 #include "nsCycleCollectionParticipant.h"
 
+class nsICharsetConverterManager;
+class nsICharsetAlias;
 class nsIDTD;
 class nsScanner;
 class nsIProgressEventSink;
@@ -124,7 +126,6 @@ class nsParser : public nsIParser,
      */
     nsParser();
 
-
     /**
      * Destructor
      * @update	gess5/11/98
@@ -372,6 +373,14 @@ class nsParser : public nsIParser,
 
     static nsCOMArray<nsIUnicharStreamListener> *sParserDataListeners;
 
+    static nsICharsetAlias* GetCharsetAliasService() {
+      return sCharsetAliasService;
+    }
+
+    static nsICharsetConverterManager* GetCharsetConverterManager() {
+      return sCharsetConverterManager;
+    }
+
 protected:
 
     /**
@@ -455,7 +464,8 @@ protected:
     nsCString           mCharset;
     nsCString           mCommandStr;
 
-    
+    static nsICharsetAlias*            sCharsetAliasService;
+    static nsICharsetConverterManager* sCharsetConverterManager;
    
 public:  
    

parser/htmlparser/src/nsScanner.cpp

@@ -103,7 +103,6 @@ nsScanner::nsScanner(const nsAString& anHTMLString, const nsACString& aCharset,
   mIncremental = PR_FALSE;
   mUnicodeDecoder = 0;
   mCharsetSource = kCharsetUninitialized;
-  SetDocumentCharset(aCharset, aSource);
 }
 
 /**
@@ -149,42 +148,48 @@ nsresult nsScanner::SetDocumentCharset(const nsACString& aCharset , PRInt32 aSou
   if( aSource < mCharsetSource) // priority is lower the the current one , just
     return res;
 
-  nsCOMPtr<nsICharsetAlias> calias(do_GetService(NS_CHARSETALIAS_CONTRACTID, &res));
-  NS_ASSERTION( nsnull != calias, "cannot find charset alias");
-  if( NS_SUCCEEDED(res) && (nsnull != calias))
+  nsICharsetAlias* calias = nsParser::GetCharsetAliasService();
+  NS_ASSERTION(calias, "Must have the charset alias service!");
+
+  if (!mCharset.IsEmpty())
   {
-    PRBool same = PR_FALSE;
+    PRBool same;
     res = calias->Equals(aCharset, mCharset, &same);
     if(NS_SUCCEEDED(res) && same)
     {
       return NS_OK; // no difference, don't change it
     }
-    // different, need to change it
-    nsCAutoString charsetName;
-    res = calias->GetPreferred(aCharset, charsetName);
-
-    if(NS_FAILED(res) && (kCharsetUninitialized == mCharsetSource) )
-    {
-       // failed - unknown alias , fallback to ISO-8859-1
-      charsetName.AssignLiteral("ISO-8859-1");
-    }
-    mCharset = charsetName;
-    mCharsetSource = aSource;
-
-    nsCOMPtr<nsICharsetConverterManager> ccm = 
-             do_GetService(NS_CHARSETCONVERTERMANAGER_CONTRACTID, &res);
-    if(NS_SUCCEEDED(res) && (nsnull != ccm))
-    {
-      nsIUnicodeDecoder * decoder = nsnull;
-      res = ccm->GetUnicodeDecoderRaw(mCharset.get(), &decoder);
-      if(NS_SUCCEEDED(res) && (nsnull != decoder))
-      {
-         NS_IF_RELEASE(mUnicodeDecoder);
-
-         mUnicodeDecoder = decoder;
-      }    
-    }
   }
+
+  // different, need to change it
+  nsCString charsetName;
+  res = calias->GetPreferred(aCharset, charsetName);
+
+  if(NS_FAILED(res) && (mCharsetSource == kCharsetUninitialized))
+  {
+     // failed - unknown alias , fallback to ISO-8859-1
+    mCharset.AssignLiteral("ISO-8859-1");
+  }
+  else
+  {
+    mCharset.Assign(charsetName);
+  }
+
+  mCharsetSource = aSource;
+
+  NS_ASSERTION(nsParser::GetCharsetConverterManager(),
+               "Must have the charset converter manager!");
+
+  nsIUnicodeDecoder * decoder = nsnull;
+  res = nsParser::GetCharsetConverterManager()->
+    GetUnicodeDecoderRaw(mCharset.get(), &decoder);
+  if(NS_SUCCEEDED(res) && (nsnull != decoder))
+  {
+     NS_IF_RELEASE(mUnicodeDecoder);
+
+     mUnicodeDecoder = decoder;
+  }
+
   return res;
 }