[Libreoffice-commits] core.git: Branch 'feature/perfwork' - comphelper/source include/comphelper sax/source

[Matúš Kukan]

 comphelper/source/streaming/seqstream.cxx |    2 ++
 include/comphelper/seqstream.hxx          |   11 +++++++----
 sax/source/tools/fastserializer.cxx       |   19 +++++++++++++++----
 sax/source/tools/fastserializer.hxx       |    2 ++
 4 files changed, 26 insertions(+), 8 deletions(-)

New commits:
commit 8524d5aba59148eef94859289f00ab6477e0e821
Author: Matúš Kukan <matus.kukan at collabora.com>
Date:   Thu Sep 25 14:45:55 2014 +0200

    FastSerializer: Do not keep whole sequence in memory; write in chunks
    
    Change-Id: Ibca65a2b8397358ac4013c8e5b53389ac6949e4d

diff --git a/comphelper/source/streaming/seqstream.cxx b/comphelper/source/streaming/seqstream.cxx
index d66ce3b..aec4519 100644
--- a/comphelper/source/streaming/seqstream.cxx
+++ b/comphelper/source/streaming/seqstream.cxx
@@ -214,6 +214,8 @@ void SAL_CALL OSequenceOutputStream::flush(  ) throw(NotConnectedException, Buff
 
     // cut the sequence to the real size
     m_rSequence.realloc(m_nSize);
+    // and next time write to the beginning
+    m_nSize = 0;
 }
 
 
diff --git a/include/comphelper/seqstream.hxx b/include/comphelper/seqstream.hxx
index 13dd4dc..75b1934 100644
--- a/include/comphelper/seqstream.hxx
+++ b/include/comphelper/seqstream.hxx
@@ -115,13 +115,16 @@ public:
 
     virtual ~OSequenceOutputStream() { if (m_bConnected) closeOutput(); }
 
+    sal_Int32 getSize() const { return m_nSize; }
+
     /// same as XOutputStream::writeBytes (as expected :)
     virtual void SAL_CALL writeBytes( const ::com::sun::star::uno::Sequence< sal_Int8 >& aData ) throw(::com::sun::star::io::NotConnectedException, ::com::sun::star::io::BufferSizeExceededException, ::com::sun::star::io::IOException, ::com::sun::star::uno::RuntimeException, std::exception) SAL_OVERRIDE;
-    /// this is a dummy in this implementation, no buffering is used
+    /** Resizes the sequence used for writing to the really used size.
+     *  Next time, writeBytes will write to the beginning of the sequence.
+    */
     virtual void SAL_CALL flush(  ) throw(::com::sun::star::io::NotConnectedException, ::com::sun::star::io::BufferSizeExceededException, ::com::sun::star::io::IOException, ::com::sun::star::uno::RuntimeException, std::exception) SAL_OVERRIDE;
-    /** closes the output stream. In the case of this class, this means that the sequence used for writing is
-        resized to the really used size and not used any further, every subsequent call to one of the XOutputStream
-        methods will throw a <code>NotConnectedException</code>.
+    /** Calls flush() and closes the output stream to prevent further manipulation with the sequence.
+        Every subsequent call to one of the XOutputStream methods will throw a <code>NotConnectedException</code>.
     */
     virtual void SAL_CALL closeOutput(  ) throw(::com::sun::star::io::NotConnectedException, ::com::sun::star::io::BufferSizeExceededException, ::com::sun::star::io::IOException, ::com::sun::star::uno::RuntimeException, std::exception) SAL_OVERRIDE;
 };
diff --git a/sax/source/tools/fastserializer.cxx b/sax/source/tools/fastserializer.cxx
index dfa37c7..d85bfef 100644
--- a/sax/source/tools/fastserializer.cxx
+++ b/sax/source/tools/fastserializer.cxx
@@ -53,8 +53,8 @@ using ::com::sun::star::io::BufferSizeExceededException;
 
 namespace sax_fastparser {
     FastSaxSerializer::FastSaxSerializer( )
-        : maOutputData()
-        , maOutputStream(maOutputData)
+        : maOutputData(0x4000)
+        , maOutputStream(maOutputData, 1.3, 0x1000, 0x4000)
         , mxOutputStream()
         , mxFastTokenHandler()
         , maMarkStack()
@@ -298,7 +298,7 @@ namespace sax_fastparser {
 
         if ( maMarkStack.size() == 1  && eMergeType != MERGE_MARKS_IGNORE)
         {
-            maOutputStream.writeBytes( maMarkStack.top()->getData() );
+            writeOutput( maMarkStack.top()->getData() );
             maMarkStack.pop();
             return;
         }
@@ -319,11 +319,22 @@ namespace sax_fastparser {
     void FastSaxSerializer::writeBytes( const Sequence< ::sal_Int8 >& aData ) throw ( NotConnectedException, BufferSizeExceededException, IOException, RuntimeException )
     {
         if ( maMarkStack.empty() )
-            maOutputStream.writeBytes( aData );
+            writeOutput( aData );
         else
             maMarkStack.top()->append( aData );
     }
 
+    void FastSaxSerializer::writeOutput( const Sequence< ::sal_Int8 >& aData ) throw ( NotConnectedException, BufferSizeExceededException, IOException, RuntimeException )
+    {
+        maOutputStream.writeBytes( aData );
+        // Write when the sequence gets big enough
+        if (maOutputStream.getSize() > 0x10000)
+        {
+            maOutputStream.flush();
+            mxOutputStream->writeBytes(maOutputData);
+        }
+    }
+
     FastSaxSerializer::Int8Sequence& FastSaxSerializer::ForMerge::getData()
     {
         merge( maData, maPostponed, true );
diff --git a/sax/source/tools/fastserializer.hxx b/sax/source/tools/fastserializer.hxx
index 147d767..45535f2 100644
--- a/sax/source/tools/fastserializer.hxx
+++ b/sax/source/tools/fastserializer.hxx
@@ -217,6 +217,8 @@ private:
 
     void writeFastAttributeList( const ::com::sun::star::uno::Reference< ::com::sun::star::xml::sax::XFastAttributeList >& Attribs );
     void write( const OUString& s );
+    void writeOutput( const css::uno::Sequence< ::sal_Int8 >& aData )
+        throw (css::io::NotConnectedException, css::io::BufferSizeExceededException, css::io::IOException, css::uno::RuntimeException);
 
 protected:
     rtl::ByteSequence maClosingBracket;