SkipImages option for document import

This option allows images and drawings to be skipped while importing
DOC and DOCX, for text-only conversion or indexing.

Change-Id: Id4bfbbe48d8a8e970b1cb5922187768bf94dc918
Reviewed-on: https://gerrit.libreoffice.org/14733
Reviewed-by: Németh László <nemeth@numbertext.org>
Tested-by: Németh László <nemeth@numbertext.org>

5 files changed, 18 insertions, 9 deletions

diff --git a/writerfilter/inc/ooxml/OOXMLDocument.hxx b/writerfilter/inc/ooxml/OOXMLDocument.hxx
index c533e359eb86..d9b76e29c634 100644
--- a/writerfilter/inc/ooxml/OOXMLDocument.hxx
+++ b/writerfilter/inc/ooxml/OOXMLDocument.hxx
@@ -260,7 +260,9 @@ public:
     createStream(OOXMLStream::Pointer_t pStream, const OUString & rId);
 
     static OOXMLDocument *
-    createDocument(OOXMLStream::Pointer_t pStream, const css::uno::Reference<css::task::XStatusIndicator>& xStatusIndicator);
+    createDocument(OOXMLStream::Pointer_t pStream,
+                 const css::uno::Reference<css::task::XStatusIndicator>& xStatusIndicator,
+                 bool bSkipImage);
 
 };
 
diff --git a/writerfilter/source/filter/WriterFilter.cxx b/writerfilter/source/filter/WriterFilter.cxx
index 1a968d98fc43..327477b57d20 100644
--- a/writerfilter/source/filter/WriterFilter.cxx
+++ b/writerfilter/source/filter/WriterFilter.cxx
@@ -154,6 +154,7 @@ sal_Bool WriterFilter::filter(const uno::Sequence< beans::PropertyValue >& aDesc
     {
         utl::MediaDescriptor aMediaDesc(aDescriptor);
         bool bRepairStorage = aMediaDesc.getUnpackedValueOrDefault("RepairPackage", false);
+        bool bSkipImages = aMediaDesc.getUnpackedValueOrDefault("FilterOptions", OUString("")) == "SkipImages";
 
         uno::Reference< io::XInputStream > xInputStream;
         try
@@ -175,7 +176,7 @@ sal_Bool WriterFilter::filter(const uno::Sequence< beans::PropertyValue >& aDesc
         //create the tokenizer and domain mapper
         writerfilter::ooxml::OOXMLStream::Pointer_t pDocStream = writerfilter::ooxml::OOXMLDocumentFactory::createStream(m_xContext, xInputStream, bRepairStorage);
         uno::Reference<task::XStatusIndicator> xStatusIndicator = aMediaDesc.getUnpackedValueOrDefault(utl::MediaDescriptor::PROP_STATUSINDICATOR(), uno::Reference<task::XStatusIndicator>());
-        writerfilter::ooxml::OOXMLDocument::Pointer_t pDocument(writerfilter::ooxml::OOXMLDocumentFactory::createDocument(pDocStream, xStatusIndicator));
+        writerfilter::ooxml::OOXMLDocument::Pointer_t pDocument(writerfilter::ooxml::OOXMLDocumentFactory::createDocument(pDocStream, xStatusIndicator, bSkipImages));
 
         uno::Reference<frame::XModel> xModel(m_xDstDoc, uno::UNO_QUERY_THROW);
         pDocument->setModel(xModel);
diff --git a/writerfilter/source/ooxml/OOXMLDocumentImpl.cxx b/writerfilter/source/ooxml/OOXMLDocumentImpl.cxx
index 097bdfeab3dd..d15d59e81352 100644
--- a/writerfilter/source/ooxml/OOXMLDocumentImpl.cxx
+++ b/writerfilter/source/ooxml/OOXMLDocumentImpl.cxx
@@ -48,13 +48,14 @@ namespace writerfilter {
 namespace ooxml
 {
 
-OOXMLDocumentImpl::OOXMLDocumentImpl(OOXMLStream::Pointer_t pStream, const uno::Reference<task::XStatusIndicator>& xStatusIndicator)
+OOXMLDocumentImpl::OOXMLDocumentImpl(OOXMLStream::Pointer_t pStream, const uno::Reference<task::XStatusIndicator>& xStatusIndicator, bool bSkipImages)
     : mpStream(pStream)
     , mxStatusIndicator(xStatusIndicator)
     , mnXNoteId(0)
     , mXNoteType(0)
     , mxThemeDom(nullptr)
     , mbIsSubstream(false)
+    , mbSkipImages(bSkipImages)
     , mnPercentSize(0)
     , mnProgressLastPos(0)
     , mnProgressCurrentPos(0)
@@ -268,7 +269,7 @@ OOXMLDocumentImpl::getSubStream(const OUString & rId)
 
     OOXMLDocumentImpl * pTemp;
     // Do not pass status indicator to sub-streams: they are typically marginal in size, so we just track the main document for now.
-    writerfilter::Reference<Stream>::Pointer_t pRet( pTemp = new OOXMLDocumentImpl(pStream, uno::Reference<task::XStatusIndicator>()) );
+    writerfilter::Reference<Stream>::Pointer_t pRet( pTemp = new OOXMLDocumentImpl(pStream, uno::Reference<task::XStatusIndicator>(), mbSkipImages ));
     pTemp->setModel(mxModel);
     pTemp->setDrawPage(mxDrawPage);
     pTemp->setIsSubstream( true );
@@ -282,7 +283,7 @@ OOXMLDocumentImpl::getXNoteStream(OOXMLStream::StreamType_t nType, const Id & rT
     OOXMLStream::Pointer_t pStream =
         (OOXMLDocumentFactory::createStream(mpStream, nType));
     // See above, no status indicator for the note stream, either.
-    OOXMLDocumentImpl * pDocument = new OOXMLDocumentImpl(pStream, uno::Reference<task::XStatusIndicator>());
+    OOXMLDocumentImpl * pDocument = new OOXMLDocumentImpl(pStream, uno::Reference<task::XStatusIndicator>(), mbSkipImages);
     pDocument->setXNoteId(nId);
     pDocument->setXNoteType(rType);
 
@@ -950,9 +951,9 @@ uno::Sequence<beans::PropertyValue > OOXMLDocumentImpl::getEmbeddingsList( )
 
 OOXMLDocument *
 OOXMLDocumentFactory::createDocument
-(OOXMLStream::Pointer_t pStream, const uno::Reference<task::XStatusIndicator>& xStatusIndicator)
+(OOXMLStream::Pointer_t pStream, const uno::Reference<task::XStatusIndicator>& xStatusIndicator, bool mbSkipImages)
 {
-    return new OOXMLDocumentImpl(pStream, xStatusIndicator);
+    return new OOXMLDocumentImpl(pStream, xStatusIndicator, mbSkipImages);
 }
 
 }}
diff --git a/writerfilter/source/ooxml/OOXMLDocumentImpl.hxx b/writerfilter/source/ooxml/OOXMLDocumentImpl.hxx
index 8f9cb2ad8490..507a3ad23b85 100644
--- a/writerfilter/source/ooxml/OOXMLDocumentImpl.hxx
+++ b/writerfilter/source/ooxml/OOXMLDocumentImpl.hxx
@@ -55,6 +55,7 @@ class OOXMLDocumentImpl : public OOXMLDocument
     css::uno::Sequence < css::beans::PropertyValue > mxEmbeddingsList;
     std::vector<css::beans::PropertyValue> mxEmbeddingsListTemp;
     bool mbIsSubstream;
+    bool mbSkipImages;
     /// How many paragraphs equal to 1 percent?
     sal_Int32 mnPercentSize;
     /// Position progress when it was last updated, possibly not after every paragraph in case of large documents.
@@ -90,7 +91,7 @@ protected:
     void resolveGlossaryStream(Stream & rStream);
     void resolveEmbeddingsStream(OOXMLStream::Pointer_t pStream);
 public:
-    OOXMLDocumentImpl(OOXMLStream::Pointer_t pStream, const css::uno::Reference<css::task::XStatusIndicator>& xStatusIndicator);
+    OOXMLDocumentImpl(OOXMLStream::Pointer_t pStream, const css::uno::Reference<css::task::XStatusIndicator>& xStatusIndicator, bool bSkipImages);
     virtual ~OOXMLDocumentImpl();
 
     virtual void resolve(Stream & rStream) SAL_OVERRIDE;
@@ -140,6 +141,7 @@ public:
     virtual css::uno::Sequence<css::beans::PropertyValue >  getEmbeddingsList() SAL_OVERRIDE;
 
     void incrementProgress();
+    bool IsSkipImages() { return mbSkipImages; };
 };
 }}
 #endif // OOXML_DOCUMENT_IMPL_HXX
diff --git a/writerfilter/source/ooxml/OOXMLFastContextHandler.cxx b/writerfilter/source/ooxml/OOXMLFastContextHandler.cxx
index 3a47a8a7c572..9a9c841a6b6f 100644
--- a/writerfilter/source/ooxml/OOXMLFastContextHandler.cxx
+++ b/writerfilter/source/ooxml/OOXMLFastContextHandler.cxx
@@ -1826,9 +1826,12 @@ OOXMLFastContextHandlerWrapper::lcl_createFastChildContext
     // filter out a single token. Just hardwire the wrap token here till we
     // need a more generic solution.
     bool bIsWrap = Element == static_cast<sal_Int32>(NMSP_vmlWord | XML_wrap);
+    bool bSkipImages = getDocument()->IsSkipImages() && oox::getNamespace(Element) == static_cast<sal_Int32>(NMSP_dml) &&
+        !((oox::getBaseToken(Element) == XML_linkedTxbx) || (oox::getBaseToken(Element) == XML_txbx));
+
     if ( bInNamespaces && (!bIsWrap || static_cast<OOXMLFastContextHandlerShape*>(mpParent)->isShapeSent()) )
         xResult.set(OOXMLFactory::getInstance()->createFastChildContextFromStart(this, Element));
-    else if (mxContext.is())
+    else if (mxContext.is()  && !bSkipImages)
     {
         OOXMLFastContextHandlerWrapper * pWrapper =
             new OOXMLFastContextHandlerWrapper