tdf#115429 xmloff: ODF import: fix handling of unknown attributes

... in SvXMLLegacyToFastDocHandler::startElement(), so that it does not
create invalid "-1" tokens that end up stored as empty-string
attribute names in SvXMLAttrCollection and ultimately exported
as invalid XML.

One surprising aspect is that the maNamespaceURL in struct
UnknownAttribute stores the namespace prefix in the libreoffice-5-4
branch, as it lacks commit bb59a80ee6000d3922fa95262f67e291fd9d8ee2.

The attributes are read and converted again in
SvXMLImportContext::startUnknownElement().

Change-Id: Id081c677286a77ec50d9884cdbd9135cf4f6e5b6
Reviewed-on: https://gerrit.libreoffice.org/50583
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>

2 files changed, 21 insertions, 3 deletions

diff --git a/xmloff/source/core/xmlictxt.cxx b/xmloff/source/core/xmlictxt.cxx
index 63025083d78f..158ba094b60c 100644
--- a/xmloff/source/core/xmlictxt.cxx
+++ b/xmloff/source/core/xmlictxt.cxx
@@ -111,6 +111,8 @@ void SAL_CALL SvXMLImportContext::startUnknownElement(const OUString & rPrefix,
         {
             const OUString& rAttrValue = unknownAttribs[i].Value;
             OUString sAttrName = unknownAttribs[i].Name;
+            // NOTE: in *this* release branch, the NamespaceURL is *not*
+            // the URL, but the *prefix*!
             const OUString& rAttrNamespacePrefix = unknownAttribs[i].NamespaceURL;
             if ( !rAttrNamespacePrefix.isEmpty() )
                 sAttrName = rAttrNamespacePrefix + ":" + sAttrName;
diff --git a/xmloff/source/core/xmlimp.cxx b/xmloff/source/core/xmlimp.cxx
index eba9d2ad21b6..334419ade958 100644
--- a/xmloff/source/core/xmlimp.cxx
+++ b/xmloff/source/core/xmlimp.cxx
@@ -2134,15 +2134,31 @@ void SAL_CALL SvXMLLegacyToFastDocHandler::startElement( const OUString& rName,
     for( sal_Int16 i=0; i < nAttrCount; i++ )
     {
         OUString aLocalAttrName;
+        OUString aPrefix;
+        OUString aNamespace;
         const OUString& rAttrName = xAttrList->getNameByIndex( i );
         const OUString& rAttrValue = xAttrList->getValueByIndex( i );
-        sal_uInt16 nAttrPrefix = mrImport->mpNamespaceMap->GetKeyByAttrName( rAttrName, &aLocalAttrName );
+        sal_uInt16 const nAttrPrefix(mrImport->mpNamespaceMap->GetKeyByAttrName(
+                rAttrName, &aPrefix, &aLocalAttrName, &aNamespace));
         if( XML_NAMESPACE_XMLNS != nAttrPrefix )
         {
             Sequence< sal_Int8 > aAttrSeq( reinterpret_cast<sal_Int8 const *>(
                                     OUStringToOString( aLocalAttrName, RTL_TEXTENCODING_UTF8 ).getStr()), aLocalAttrName.getLength() );
-            sal_Int32 nAttr = NAMESPACE_TOKEN( nAttrPrefix ) | mrImport->mxTokenHandler->getTokenFromUTF8( aAttrSeq ) ;
-            mxFastAttributes->add( nAttr, OUStringToOString( rAttrValue, RTL_TEXTENCODING_UTF8 ).getStr() );
+            auto const nToken(mrImport->mxTokenHandler->getTokenFromUTF8(aAttrSeq));
+            if (nToken == xmloff::XML_TOKEN_INVALID)
+            {
+                // NOTE: in *this* release branch, the NamespaceURL is *not*
+                // the URL, but the *prefix*! see
+                // sax_fastparser::FastSaxParserImpl::callbackStartElement 1181
+                mxFastAttributes->addUnknown(aPrefix,
+                    OUStringToOString(aLocalAttrName, RTL_TEXTENCODING_UTF8),
+                    OUStringToOString(rAttrValue, RTL_TEXTENCODING_UTF8));
+            }
+            else
+            {
+                sal_Int32 const nAttr = NAMESPACE_TOKEN(nAttrPrefix) | nToken;
+                mxFastAttributes->add(nAttr, OUStringToOString(rAttrValue, RTL_TEXTENCODING_UTF8).getStr());
+            }
         }
     }
     mrImport->startFastElement( mnElement, mxFastAttributes.get() );