fastparser: Avoid copying all tokens into a sequence.

3 files changed, 40 insertions, 3 deletions

diff --git a/include/oox/core/fasttokenhandler.hxx b/include/oox/core/fasttokenhandler.hxx
index b79472c7e466..fb4e0e16fac6 100644
--- a/include/oox/core/fasttokenhandler.hxx
+++ b/include/oox/core/fasttokenhandler.hxx
@@ -23,6 +23,7 @@
 #include <com/sun/star/lang/XServiceInfo.hpp>
 #include <com/sun/star/xml/sax/XFastTokenHandler.hpp>
 #include <cppuhelper/implbase2.hxx>
+#include <sax/fastattribs.hxx>
 
 namespace oox { class TokenMap; }
 
@@ -36,7 +37,8 @@ typedef ::cppu::WeakImplHelper2< ::com::sun::star::lang::XServiceInfo, ::com::su
 /** Wrapper implementing the com.sun.star.xml.sax.XFastTokenHandler API interface
     that provides access to the tokens generated from the internal token name list.
  */
-class FastTokenHandler : public FastTokenHandler_BASE
+class FastTokenHandler : public FastTokenHandler_BASE,
+                         public sax_fastparser::FastTokenHandlerBase
 {
 public:
     explicit            FastTokenHandler();
@@ -53,6 +55,9 @@ public:
     virtual ::com::sun::star::uno::Sequence< sal_Int8 > SAL_CALL getUTF8Identifier( sal_Int32 nToken ) throw (::com::sun::star::uno::RuntimeException);
     virtual sal_Int32 SAL_CALL getTokenFromUTF8( const ::com::sun::star::uno::Sequence< sal_Int8 >& Identifier ) throw (::com::sun::star::uno::RuntimeException);
 
+    // Much faster direct C++ shortcut to the method that matters
+    virtual sal_Int32 getTokenDirect( const char *pToken, sal_Int32 nLength ) const;
+
 private:
     const TokenMap&     mrTokenMap;     ///< Reference to global token map singleton.
 };
diff --git a/include/oox/token/tokenmap.hxx b/include/oox/token/tokenmap.hxx
index 495fa2d6eb3e..bddc0ce6a34d 100644
--- a/include/oox/token/tokenmap.hxx
+++ b/include/oox/token/tokenmap.hxx
@@ -47,9 +47,29 @@ public:
 
     /** Returns the token identifier for the passed UTF8 token name. */
     sal_Int32           getTokenFromUtf8(
-                            const ::com::sun::star::uno::Sequence< sal_Int8 >& rUtf8Name ) const;
+                            const ::com::sun::star::uno::Sequence< sal_Int8 >& rUtf8Name ) const
+    {
+        return getTokenFromUTF8( reinterpret_cast< const char * >(
+                                rUtf8Name.getConstArray() ),
+                             rUtf8Name.getLength() );
+    }
+
+    /** Returns the token identifier for a UTF8 string passed in pToken */
+    sal_Int32 getTokenFromUTF8( const char *pToken, sal_Int32 nLength ) const
+    {
+        // 50% of OOXML tokens are primarily 1 lower-case character, a-z
+        if( nLength == 1)
+        {
+            sal_Char c = pToken[0];
+            if (c >= 'a' && c <= 'z')
+                return mnAlphaTokens[ c - 'a' ];
+        }
+        return getTokenPerfectHash( pToken, nLength );
+    }
 
 private:
+    sal_Int32 getTokenPerfectHash( const char *pToken, sal_Int32 nLength ) const;
+
     struct TokenName
     {
         OUString maUniName;
diff --git a/include/sax/fastattribs.hxx b/include/sax/fastattribs.hxx
index 42b285c9e847..f1f64dd4b9ed 100644
--- a/include/sax/fastattribs.hxx
+++ b/include/sax/fastattribs.hxx
@@ -49,6 +49,14 @@ struct UnknownAttribute
 
 typedef std::vector< UnknownAttribute > UnknownAttributeList;
 
+/// A native C++ interface to tokenisation
+class SAX_DLLPUBLIC FastTokenHandlerBase
+{
+    public:
+        virtual ~FastTokenHandlerBase() {}
+        virtual sal_Int32 getTokenDirect( const char *pToken, sal_Int32 nLength ) const = 0;
+};
+
 /// avoid constantly allocating and freeing sequences.
 class SAX_DLLPUBLIC FastTokenLookup
 {
@@ -58,13 +66,15 @@ public:
     FastTokenLookup();
     sal_Int32 getTokenFromChars(
         const ::css::uno::Reference< ::css::xml::sax::XFastTokenHandler > &mxTokenHandler,
+        FastTokenHandlerBase *pTokenHandler,
         const char *pStr, size_t nLength = 0 );
 };
 
 class SAX_DLLPUBLIC FastAttributeList : public ::cppu::WeakImplHelper1< ::com::sun::star::xml::sax::XFastAttributeList >
 {
 public:
-    FastAttributeList( const ::com::sun::star::uno::Reference< ::com::sun::star::xml::sax::XFastTokenHandler >& xTokenHandler );
+    FastAttributeList( const ::com::sun::star::uno::Reference< ::com::sun::star::xml::sax::XFastTokenHandler >& xTokenHandler,
+                       FastTokenHandlerBase *pOptHandlerBase = NULL );
     virtual ~FastAttributeList();
 
     void clear();
@@ -100,6 +110,8 @@ private:
     std::vector< sal_Int32 > maAttributeTokens;
     UnknownAttributeList maUnknownAttributes;
     ::com::sun::star::uno::Reference< ::com::sun::star::xml::sax::XFastTokenHandler > mxTokenHandler;
+    FastTokenHandlerBase *mpTokenHandler;
+
     FastTokenLookup maTokenLookup;
 };