Modifying the impl. of startUnknownElement of FastParser:

Modifying it to emit the namespace URI instead of
prefix and qualified name instead of local name.
This will be useful for handling arbitrary elements
in the fast contexts.

Change-Id: I0f150b862574612e97491f6c335f3f4c9966da0a
Reviewed-on: https://gerrit.libreoffice.org/47055
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>

2 files changed, 43 insertions, 34 deletions

diff --git a/sax/source/fastparser/fastparser.cxx b/sax/source/fastparser/fastparser.cxx
index 55b6545c5114..f56d368580ac 100644
--- a/sax/source/fastparser/fastparser.cxx
+++ b/sax/source/fastparser/fastparser.cxx
@@ -249,6 +249,7 @@ private:
     bool consume(EventList&);
     void deleteUsedEvents();
     void sendPendingCharacters();
+    void addUnknownElementWithPrefix(const xmlChar **attributes, int i, rtl::Reference< FastAttributeList >& xAttributes);
 
     sal_Int32 GetToken( const xmlChar* pName, sal_Int32 nameLen );
     /// @throws css::xml::sax::SAXException
@@ -1148,8 +1149,7 @@ void FastSaxParserImpl::callbackStartElement(const xmlChar *localName , const xm
                     if( nAttributeToken != FastToken::DONTKNOW )
                         rEvent.mxAttributes->add( nAttributeToken, XML_CAST( attributes[ i + 3 ] ), attributes[ i + 4 ] - attributes[ i + 3 ] );
                     else
-                        rEvent.mxAttributes->addUnknown( OUString( XML_CAST( attributes[ i + 1 ] ), strlen( XML_CAST( attributes[ i + 1 ] )), RTL_TEXTENCODING_UTF8 ),
-                                OString( XML_CAST( attributes[ i ] )), OString( XML_CAST( attributes[ i + 3 ] ), attributes[ i + 4 ] - attributes[ i + 3 ] ));
+                        addUnknownElementWithPrefix(attributes, i, rEvent.mxAttributes);
                 }
                 else
                 {
@@ -1174,8 +1174,7 @@ void FastSaxParserImpl::callbackStartElement(const xmlChar *localName , const xm
             for (int i = 0; i < numAttributes * 5; i += 5)
             {
                 if( attributes[ i + 1 ] != nullptr )
-                    rEvent.mxAttributes->addUnknown( OUString( XML_CAST( attributes[ i + 1 ] ), strlen( XML_CAST( attributes[ i + 1 ] )), RTL_TEXTENCODING_UTF8 ),
-                            OString( XML_CAST( attributes[ i ] )), OString( XML_CAST( attributes[ i + 3 ] ), attributes[ i + 4 ] - attributes[ i + 3 ] ));
+                    addUnknownElementWithPrefix(attributes, i, rEvent.mxAttributes);
                 else
                     rEvent.mxAttributes->addUnknown( XML_CAST( attributes[ i ] ),
                             OString( XML_CAST( attributes[ i + 3 ] ), attributes[ i + 4 ] - attributes[ i + 3 ] ));
@@ -1186,16 +1185,19 @@ void FastSaxParserImpl::callbackStartElement(const xmlChar *localName , const xm
 
         if( rEvent.mnElementToken == FastToken::DONTKNOW )
         {
+            OUString aElementPrefix;
             if( prefix != nullptr )
             {
                 if ( !m_bIgnoreMissingNSDecl || URI != nullptr )
                     sNamespace = OUString( XML_CAST( URI ), strlen( XML_CAST( URI )), RTL_TEXTENCODING_UTF8 );
+                else
+                    sNamespace.clear();
                 nNamespaceToken = GetNamespaceToken( sNamespace );
-                rEvent.msNamespace = OUString( XML_CAST( prefix ), strlen( XML_CAST( prefix )), RTL_TEXTENCODING_UTF8 );
+                aElementPrefix = OUString( XML_CAST( prefix ), strlen( XML_CAST( prefix )), RTL_TEXTENCODING_UTF8 );
             }
-            else
-                rEvent.msNamespace.clear();
-            rEvent.msElementName = OUString( XML_CAST( localName ), strlen( XML_CAST( localName )), RTL_TEXTENCODING_UTF8 );
+            const OUString& rElementLocalName = OUString( XML_CAST( localName ), strlen( XML_CAST( localName )), RTL_TEXTENCODING_UTF8 );
+            rEvent.msNamespace = sNamespace;
+            rEvent.msElementName = (aElementPrefix.isEmpty())? rElementLocalName : aElementPrefix + ":" + rElementLocalName;
         }
         else // token is always preferred.
             rEvent.msElementName.clear();
@@ -1215,6 +1217,18 @@ void FastSaxParserImpl::callbackStartElement(const xmlChar *localName , const xm
     }
 }
 
+void FastSaxParserImpl::addUnknownElementWithPrefix(const xmlChar **attributes, int i, rtl::Reference< FastAttributeList >& xAttributes)
+{
+    OUString aNamespaceURI;
+    if ( !m_bIgnoreMissingNSDecl || attributes[i + 2] != nullptr )
+        aNamespaceURI = OUString( XML_CAST( attributes[ i + 2 ] ), strlen( XML_CAST( attributes[ i + 2 ] )), RTL_TEXTENCODING_UTF8 );
+    const OString& rPrefix = OString( XML_CAST( attributes[ i + 1 ] ));
+    const OString& rLocalName = OString( XML_CAST( attributes[ i ] ));
+    OString aQualifiedName = (rPrefix.isEmpty())? rLocalName : rPrefix + ":" + rLocalName;
+    xAttributes->addUnknown( aNamespaceURI, aQualifiedName,
+        OString( XML_CAST( attributes[ i + 3 ] ), attributes[ i + 4 ] - attributes[ i + 3 ] ));
+}
+
 void FastSaxParserImpl::callbackEndElement()
 {
     if (!pendingCharacters.isEmpty())
diff --git a/sax/source/fastparser/legacyfastparser.cxx b/sax/source/fastparser/legacyfastparser.cxx
index 458e91b2af89..97c5ab277f89 100644
--- a/sax/source/fastparser/legacyfastparser.cxx
+++ b/sax/source/fastparser/legacyfastparser.cxx
@@ -134,6 +134,9 @@ private:
     const OUString getNamespacePrefixFromToken( sal_Int32 nToken );
     const OUString getNameFromToken( sal_Int32 nToken );
 
+    static const OUString aDefaultNamespace;
+    static const OUString aNamespaceSeparator;
+
 public:
     CallbackDocumentHandler( Reference< XDocumentHandler > const & xDocumentHandler,
                              rtl::Reference< NamespaceHandler > const & rNamespaceHandler,
@@ -156,6 +159,9 @@ public:
 
 };
 
+const OUString CallbackDocumentHandler::aDefaultNamespace = OUString("");
+const OUString CallbackDocumentHandler::aNamespaceSeparator = OUString(":");
+
 const OUString CallbackDocumentHandler::getNamespacePrefixFromToken( sal_Int32 nToken )
 {
     if ( ( nToken & 0xffff0000 ) != 0 )
@@ -210,32 +216,28 @@ void SAL_CALL CallbackDocumentHandler::setDocumentLocator( const Reference< XLoc
 
 void SAL_CALL CallbackDocumentHandler::startFastElement( sal_Int32 nElement , const Reference< XFastAttributeList >& Attribs  )
 {
-    startUnknownElement( CallbackDocumentHandler::getNamespacePrefixFromToken( nElement ),
-                         CallbackDocumentHandler::getNameFromToken( nElement ), Attribs );
+    const OUString& rPrefix = CallbackDocumentHandler::getNamespacePrefixFromToken( nElement );
+    const OUString& rLocalName = CallbackDocumentHandler::getNameFromToken( nElement );
+    startUnknownElement( aDefaultNamespace, (rPrefix.isEmpty())? rLocalName : rPrefix + aNamespaceSeparator + rLocalName, Attribs );
 }
 
-void SAL_CALL CallbackDocumentHandler::startUnknownElement( const OUString& Namespace, const OUString& Name, const Reference< XFastAttributeList >& Attribs  )
+void SAL_CALL CallbackDocumentHandler::startUnknownElement( const OUString& /*Namespace*/, const OUString& Name, const Reference< XFastAttributeList >& Attribs  )
 {
     if ( m_xDocumentHandler.is() )
     {
-        OUString elementName;
         rtl::Reference < comphelper::AttributeList > rAttrList = new comphelper::AttributeList;
         m_aNamespaceHandler->addNSDeclAttributes( rAttrList );
-        if ( !Namespace.isEmpty() )
-            elementName =  Namespace + ":" + Name;
-        else
-            elementName = Name;
 
         Sequence< xml::FastAttribute > fastAttribs = Attribs->getFastAttributes();
         sal_uInt16 len = fastAttribs.getLength();
         for (sal_uInt16 i = 0; i < len; i++)
         {
-            OUString& rAttrValue = fastAttribs[i].Value;
+            const OUString& rAttrValue = fastAttribs[i].Value;
             sal_Int32 nToken = fastAttribs[i].Token;
             const OUString& rAttrNamespacePrefix = CallbackDocumentHandler::getNamespacePrefixFromToken( nToken );
             OUString sAttrName = CallbackDocumentHandler::getNameFromToken( nToken );
             if ( !rAttrNamespacePrefix.isEmpty() )
-                sAttrName = rAttrNamespacePrefix + ":" + sAttrName;
+                sAttrName = rAttrNamespacePrefix + aNamespaceSeparator + sAttrName;
 
             rAttrList->AddAttribute( sAttrName, "CDATA", rAttrValue );
         }
@@ -244,35 +246,28 @@ void SAL_CALL CallbackDocumentHandler::startUnknownElement( const OUString& Name
         len = unknownAttribs.getLength();
         for (sal_uInt16 i = 0; i < len; i++)
         {
-            OUString& rAttrValue = unknownAttribs[i].Value;
-            OUString sAttrName = unknownAttribs[i].Name;
-            OUString& rAttrNamespacePrefix = unknownAttribs[i].NamespaceURL;
-            if ( !rAttrNamespacePrefix.isEmpty() )
-                sAttrName = rAttrNamespacePrefix + ":" + sAttrName;
+            const OUString& rAttrValue = unknownAttribs[i].Value;
+            const OUString& rAttrName = unknownAttribs[i].Name;
 
-            rAttrList->AddAttribute( sAttrName, "CDATA", rAttrValue );
+            rAttrList->AddAttribute( rAttrName, "CDATA", rAttrValue );
         }
-        m_xDocumentHandler->startElement( elementName, rAttrList.get() );
+        m_xDocumentHandler->startElement( Name, rAttrList.get() );
     }
 }
 
 void SAL_CALL CallbackDocumentHandler::endFastElement( sal_Int32 nElement )
 {
-    endUnknownElement( CallbackDocumentHandler::getNamespacePrefixFromToken( nElement ),
-                       CallbackDocumentHandler::getNameFromToken( nElement ) );
+    const OUString& rPrefix = CallbackDocumentHandler::getNamespacePrefixFromToken( nElement );
+    const OUString& rLocalName = CallbackDocumentHandler::getNameFromToken( nElement );
+    endUnknownElement( aDefaultNamespace, (rPrefix.isEmpty())? rLocalName : rPrefix + aNamespaceSeparator + rLocalName );
 }
 
 
-void SAL_CALL CallbackDocumentHandler::endUnknownElement( const OUString& Namespace, const OUString& Name )
+void SAL_CALL CallbackDocumentHandler::endUnknownElement( const OUString& /*Namespace*/, const OUString& Name )
 {
     if ( m_xDocumentHandler.is() )
     {
-        OUString elementName;
-        if ( !Namespace.isEmpty() )
-            elementName = Namespace + ":" + Name;
-        else
-            elementName = Name;
-        m_xDocumentHandler->endElement( elementName );
+        m_xDocumentHandler->endElement( Name );
     }
 }