tdf#79878 perf loading docx file, sax improvements

these are the smaller improvements, they make about 5% worth of
difference

- use std::vector instead of std::deque
- use std::move on pendingCharacters instead of copying
- in FastAttributeList::add, when reallocate the buffer, allocate twice
the existing size, instead of increasing to only what we need
- in FastAttributeList, create getAttributeIndex and friends, so
we can avoid iterating the attribute list more often than necessary

Change-Id: I3e3380ea50b77c6845b66e83404e245778ec06eb
Reviewed-on: https://gerrit.libreoffice.org/57021
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>

2 files changed, 24 insertions, 7 deletions

diff --git a/sax/source/fastparser/fastparser.cxx b/sax/source/fastparser/fastparser.cxx
index 2765584a08fc..9477559c12e8 100644
--- a/sax/source/fastparser/fastparser.cxx
+++ b/sax/source/fastparser/fastparser.cxx
@@ -178,13 +178,13 @@ struct Entity : public ParserData
     void throwException( const ::rtl::Reference< FastLocatorImpl > &xDocumentLocator,
                          bool mbDuringParse );
 
-    std::stack< NameWithToken >           maNamespaceStack;
+    std::stack< NameWithToken, std::vector<NameWithToken> > maNamespaceStack;
     /* Context for main thread consuming events.
      * startElement() stores the data, which characters() and endElement() uses
      */
-    std::stack< SaxContext>               maContextStack;
+    std::stack< SaxContext, std::vector<SaxContext> >  maContextStack;
     // Determines which elements of maNamespaceDefines are valid in current context
-    std::stack< sal_uInt32 >              maNamespaceCount;
+    std::stack< sal_uInt32, std::vector<sal_uInt32> >  maNamespaceCount;
     std::vector< std::shared_ptr< NamespaceDefine > >
                                           maNamespaceDefines;
 
@@ -270,7 +270,7 @@ private:
     ParserData maData;                      /// Cached parser configuration for next call of parseStream().
 
     Entity *mpTop;                          /// std::stack::top() is amazingly slow => cache this.
-    std::stack< Entity > maEntities;      /// Entity stack for each call of parseStream().
+    std::stack< Entity > maEntities;        /// Entity stack for each call of parseStream().
     OUString pendingCharacters;             /// Data from characters() callback that needs to be sent.
 };
 
@@ -674,11 +674,12 @@ sal_Int32 FastSaxParserImpl::GetTokenWithPrefix( const xmlChar* pPrefix, int nPr
     sal_uInt32 nNamespace = rEntity.maNamespaceCount.top();
     while( nNamespace-- )
     {
-        const OString& rPrefix( rEntity.maNamespaceDefines[nNamespace]->maPrefix );
+        const auto & rNamespaceDefine = rEntity.maNamespaceDefines[nNamespace];
+        const OString& rPrefix( rNamespaceDefine->maPrefix );
         if( (rPrefix.getLength() == nPrefixLen) &&
             (strncmp( rPrefix.getStr(), XML_CAST( pPrefix ), nPrefixLen ) == 0 ) )
         {
-            nNamespaceToken = rEntity.maNamespaceDefines[nNamespace]->mnToken;
+            nNamespaceToken = rNamespaceDefine->mnToken;
             break;
         }
 
diff --git a/sax/source/tools/fastattribs.cxx b/sax/source/tools/fastattribs.cxx
index a9f0baf7a2c0..30a764368947 100644
--- a/sax/source/tools/fastattribs.cxx
+++ b/sax/source/tools/fastattribs.cxx
@@ -86,7 +86,7 @@ void FastAttributeList::add( sal_Int32 nToken, const sal_Char* pValue, size_t nV
     maAttributeValues.push_back( maAttributeValues.back() + nValueLength + 1 );
     if (maAttributeValues.back() > mnChunkLength)
     {
-        mnChunkLength = maAttributeValues.back();
+        mnChunkLength = std::max(mnChunkLength * 2, maAttributeValues.back());
         mpChunk = static_cast<sal_Char *>(realloc( mpChunk, mnChunkLength ));
     }
     strncpy(mpChunk + nWritePosition, pValue, nValueLength);
@@ -166,6 +166,11 @@ bool FastAttributeList::getAsInteger( sal_Int32 nToken, sal_Int32 &rInt) const
     return false;
 }
 
+sal_Int32 FastAttributeList::getAsIntegerByIndex( sal_Int32 nTokenIndex ) const
+{
+    return rtl_str_toInt32( getFastAttributeValue(nTokenIndex), 10 );
+}
+
 bool FastAttributeList::getAsDouble( sal_Int32 nToken, double &rDouble) const
 {
     rDouble = 0.0;
@@ -193,6 +198,12 @@ bool FastAttributeList::getAsChar( sal_Int32 nToken, const char*& rPos ) const
     return false;
 }
 
+const char* FastAttributeList::getAsCharByIndex( sal_Int32 nTokenIndex ) const
+{
+    sal_Int32 nOffset = maAttributeValues[nTokenIndex];
+    return mpChunk + nOffset;
+}
+
 OUString FastAttributeList::getValue( ::sal_Int32 Token )
 {
     for (size_t i = 0; i < maAttributeTokens.size(); ++i)
@@ -202,6 +213,11 @@ OUString FastAttributeList::getValue( ::sal_Int32 Token )
     throw SAXException();
 }
 
+OUString FastAttributeList::getValueByIndex( ::sal_Int32 nTokenIndex ) const
+{
+    return OUString( getFastAttributeValue(nTokenIndex), AttributeValueLength(nTokenIndex), RTL_TEXTENCODING_UTF8 );
+}
+
 OUString FastAttributeList::getOptionalValue( ::sal_Int32 Token )
 {
     for (size_t i = 0; i < maAttributeTokens.size(); ++i)