AutoText: loading dotx documents

For testing purposes patch introduces simple
import. New AutoText entries are imported
if document contains single paragraph AutoTexts.

Change-Id: I3f0e17c63e109eac6514ae0cb8cc168e8282b55b
Reviewed-on: https://gerrit.libreoffice.org/36634
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Szymon Kłos <szymon.klos@collabora.com>

2 files changed, 109 insertions, 3 deletions

diff --git a/sw/source/filter/docx/swdocxreader.cxx b/sw/source/filter/docx/swdocxreader.cxx
index 5975f1b862bd..b8ea965d3504 100644
--- a/sw/source/filter/docx/swdocxreader.cxx
+++ b/sw/source/filter/docx/swdocxreader.cxx
@@ -19,7 +19,22 @@
 
 #include "swdocxreader.hxx"
 
+#include <com/sun/star/document/XFilter.hpp>
+#include <com/sun/star/document/XImporter.hpp>
+#include <com/sun/star/xml/dom/XDocument.hpp>
+#include <com/sun/star/xml/dom/XElement.hpp>
+#include <com/sun/star/xml/dom/XNode.hpp>
+#include <com/sun/star/xml/dom/XNodeList.hpp>
+#include <comphelper/processfactory.hxx>
+#include <comphelper/propertyvalue.hxx>
+#include <comphelper/sequenceashashmap.hxx>
+#include <docsh.hxx>
 #include <swerror.h>
+#include <tools/ref.hxx>
+#include <unotxdoc.hxx>
+#include <unotools/streamwrap.hxx>
+
+using namespace css;
 
 extern "C" SAL_DLLPUBLIC_EXPORT Reader* SAL_CALL ImportDOCX()
 {
@@ -42,10 +57,95 @@ bool SwDOCXReader::HasGlossaries() const
     return true;
 }
 
-bool SwDOCXReader::ReadGlossaries( SwTextBlocks& /* rBlocks */, bool /* bSaveRelFiles */ ) const
+bool SwDOCXReader::ReadGlossaries( SwTextBlocks& rBlocks, bool /* bSaveRelFiles */ ) const
 {
-    // TODO
-    return false;
+    bool bRet = false;
+
+    uno::Reference<xml::dom::XDocument> xDoc = OpenDocument();
+
+    if( xDoc.is() )
+    {
+        uno::Reference<xml::dom::XNodeList> xList = xDoc->getElementsByTagName( "docPartBody" );
+        for( int i = 0; i < xList->getLength(); i++ )
+        {
+            uno::Reference<xml::dom::XNode> xBody = xList->item( i );
+            uno::Reference<xml::dom::XNode> xP = xBody->getFirstChild();
+            uno::Reference<xml::dom::XNode> xR = xP->getFirstChild();
+            uno::Reference<xml::dom::XNode> xT = xR->getFirstChild();
+            uno::Reference<xml::dom::XNode> xText = xT->getFirstChild();
+            OUString aText = xText->getNodeValue();
+            if( !aText.isEmpty() )
+            {
+                rBlocks.PutText( aText, aText, aText );
+                bRet = true;
+            }
+        }
+    }
+
+    return bRet;
+}
+
+uno::Reference<xml::dom::XDocument> SwDOCXReader::OpenDocument() const
+{
+    uno::Reference<lang::XMultiServiceFactory> xMultiServiceFactory(
+                comphelper::getProcessServiceFactory() );
+
+    uno::Reference<uno::XInterface> xInterface(
+                xMultiServiceFactory->createInstance( "com.sun.star.comp.Writer.WriterFilter" ),
+                uno::UNO_QUERY_THROW );
+
+    uno::Reference<document::XFilter> xFilter( xInterface, uno::UNO_QUERY_THROW );
+    uno::Reference<document::XImporter> xImporter( xFilter, uno::UNO_QUERY_THROW );
+
+    SfxObjectShellLock xDocSh( new SwDocShell( SfxObjectCreateMode::INTERNAL ) );
+    xDocSh->DoInitNew();
+
+    uno::Reference<lang::XComponent> xDstDoc( xDocSh->GetModel(), uno::UNO_QUERY_THROW );
+    xImporter->setTargetDocument( xDstDoc );
+
+    uno::Sequence<beans::PropertyValue> aDescriptor( 1 );
+    aDescriptor[0].Name = "InputStream";
+    uno::Reference<io::XStream> xStream( new utl::OStreamWrapper( *pMedium->GetInStream() ) );
+    aDescriptor[0].Value <<= xStream;
+
+    uno::Reference<xml::dom::XDocument> xDoc;
+
+    try
+    {
+        xFilter->filter( aDescriptor );
+
+        comphelper::SequenceAsHashMap aGrabBag = GetGrabBag( xDstDoc );
+        aGrabBag["OOXGlossary"] >>= xDoc;
+    }
+    catch (uno::Exception const& e)
+    {
+        SAL_WARN("sw.docx", "SwDOCXReader::OpenDocument(): exception: " << e.Message);
+    }
+
+    return xDoc;
+}
+
+comphelper::SequenceAsHashMap SwDOCXReader::GetGrabBag( const uno::Reference<lang::XComponent>& xDocument )
+{
+    if( xDocument.is() )
+    {
+        // get glossar document from the GrabBag
+        uno::Reference<beans::XPropertySet> xDocProps( xDocument, uno::UNO_QUERY );
+        if( xDocProps.is() )
+        {
+            uno::Reference<beans::XPropertySetInfo> xPropsInfo = xDocProps->getPropertySetInfo();
+
+            const OUString aGrabBagPropName = "InteropGrabBag";
+            if( xPropsInfo.is() && xPropsInfo->hasPropertyByName( aGrabBagPropName ) )
+            {
+                // get existing grab bag
+                comphelper::SequenceAsHashMap aGrabBag( xDocProps->getPropertyValue( aGrabBagPropName ) );
+                return aGrabBag;
+            }
+        }
+    }
+
+    return comphelper::SequenceAsHashMap();
 }
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/sw/source/filter/docx/swdocxreader.hxx b/sw/source/filter/docx/swdocxreader.hxx
index 35fdee86a643..2646a963523f 100644
--- a/sw/source/filter/docx/swdocxreader.hxx
+++ b/sw/source/filter/docx/swdocxreader.hxx
@@ -21,6 +21,9 @@
 #define INCLUDED_SW_SOURCE_FILTER_DOCX_SWDOCXREADER_HXX
 
 #include <shellio.hxx>
+#include <comphelper/sequenceashashmap.hxx>
+#include <com/sun/star/xml/dom/XDocument.hpp>
+#include <tools/ref.hxx>
 
 /// Wrapper for the UNO DOCX import filter (in writerfilter) for autotext purposes.
 class SwDOCXReader : public StgReader
@@ -33,6 +36,9 @@ public:
 
 private:
     virtual sal_uLong Read( SwDoc&, const OUString&, SwPaM&, const OUString& ) override;
+
+    uno::Reference<css::xml::dom::XDocument> OpenDocument() const;
+    static comphelper::SequenceAsHashMap GetGrabBag( const uno::Reference<lang::XComponent>& xDocument );
 };
 
 #endif