SkipImages option for PDF import

This option allows images and drawings to be skipped while importing
a PDF, the main reason was performance improvement where images were
not needed.

Change-Id: Ibca342ff6e7005bedf7b16869473832ce2576fb2
Reviewed-on: https://gerrit.libreoffice.org/14470
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

7 files changed, 86 insertions, 17 deletions

diff --git a/sdext/source/pdfimport/inc/wrapper.hxx b/sdext/source/pdfimport/inc/wrapper.hxx
index 8f0fc10f5a97..6bafce4939d7 100644
--- a/sdext/source/pdfimport/inc/wrapper.hxx
+++ b/sdext/source/pdfimport/inc/wrapper.hxx
@@ -40,7 +40,8 @@ namespace pdfi
                                     css::task::XInteractionHandler >& xIHdl,
                               const OUString&                               rPwd,
                               const css::uno::Reference<
-                                    css::uno::XComponentContext >&    xContext );
+                                    css::uno::XComponentContext >&    xContext,
+                              const OUString&        rFilterOptions = OUString());
     bool xpdf_ImportFromStream( const css::uno::Reference<
                                       css::io::XInputStream >&          xInput,
                                 const ContentSinkSharedPtr&                        rSink,
@@ -48,7 +49,8 @@ namespace pdfi
                                       css::task::XInteractionHandler >& xIHdl,
                                 const OUString&                               rPwd,
                                 const css::uno::Reference<
-                                      css::uno::XComponentContext >&    xContext );
+                                      css::uno::XComponentContext >&    xContext,
+                                const OUString&        rFilterOptions = OUString() );
 }
 
 #endif // INCLUDED_SDEXT_SOURCE_PDFIMPORT_INC_WRAPPER_HXX
diff --git a/sdext/source/pdfimport/pdfiadaptor.cxx b/sdext/source/pdfimport/pdfiadaptor.cxx
index a6cb0983d57d..6162d27b049d 100644
--- a/sdext/source/pdfimport/pdfiadaptor.cxx
+++ b/sdext/source/pdfimport/pdfiadaptor.cxx
@@ -221,7 +221,8 @@ bool PDFIRawAdaptor::parse( const uno::Reference<io::XInputStream>&       xInput
                             const OUString&                          rPwd,
                             const uno::Reference<task::XStatusIndicator>& xStatus,
                             const XmlEmitterSharedPtr&                    rEmitter,
-                            const OUString&                          rURL )
+                            const OUString&                          rURL,
+                            const OUString&                          rFilterOptions )
 {
     // container for metaformat
     boost::shared_ptr<PDFIProcessor> pSink(
@@ -234,9 +235,11 @@ bool PDFIRawAdaptor::parse( const uno::Reference<io::XInputStream>&       xInput
     bool bSuccess=false;
 
     if( xInput.is() )
-        bSuccess = xpdf_ImportFromStream( xInput, pSink, xIHdl, rPwd, m_xContext );
+        bSuccess = xpdf_ImportFromStream( xInput, pSink, xIHdl,
+                                          rPwd, m_xContext, rFilterOptions );
     else
-        bSuccess = xpdf_ImportFromFile( rURL, pSink, xIHdl, rPwd, m_xContext );
+        bSuccess = xpdf_ImportFromFile( rURL, pSink, xIHdl,
+                                        rPwd, m_xContext, rFilterOptions );
 
     if( bSuccess )
         pSink->emit(*rEmitter,*m_pVisitorFactory);
@@ -271,6 +274,7 @@ sal_Bool SAL_CALL PDFIRawAdaptor::importer( const uno::Sequence< beans::Property
     uno::Reference< task::XInteractionHandler > xInteractionHandler;
     OUString aURL;
     OUString aPwd;
+    OUString aFilterOptions;
     const beans::PropertyValue* pAttribs = rSourceData.getConstArray();
     sal_Int32 nAttribs = rSourceData.getLength();
     for( sal_Int32 i = 0; i < nAttribs; i++, pAttribs++ )
@@ -286,12 +290,15 @@ sal_Bool SAL_CALL PDFIRawAdaptor::importer( const uno::Sequence< beans::Property
             pAttribs->Value >>= xInteractionHandler;
         else if ( pAttribs->Name == "Password" )
             pAttribs->Value >>= aPwd;
+        else if ( pAttribs->Name == "FilterOptions" )
+            pAttribs->Value >>= aFilterOptions;
     }
     if( !xInput.is() )
         return sal_False;
 
     XmlEmitterSharedPtr pEmitter = createSaxEmitter(rHdl);
-    const bool bSuccess = parse(xInput,xInteractionHandler, aPwd, xStatus,pEmitter,aURL);
+    const bool bSuccess = parse(xInput, xInteractionHandler,
+                                aPwd, xStatus, pEmitter, aURL, aFilterOptions);
 
     // tell input stream that it is no longer needed
     xInput->closeInput();
diff --git a/sdext/source/pdfimport/pdfiadaptor.hxx b/sdext/source/pdfimport/pdfiadaptor.hxx
index a88f3df44f47..0b7c749b08c5 100644
--- a/sdext/source/pdfimport/pdfiadaptor.hxx
+++ b/sdext/source/pdfimport/pdfiadaptor.hxx
@@ -88,7 +88,8 @@ namespace pdfi
                     const OUString&                                                          rPwd,
                     const css::uno::Reference<css::task::XStatusIndicator>& xStatus,
                     const XmlEmitterSharedPtr&                                                    rEmitter,
-                    const OUString&                                                          rURL );
+                    const OUString&                                                          rURL,
+                    const OUString&                                         rFilterOptions = OUString());
 
     public:
         explicit PDFIRawAdaptor( const css::uno::Reference<
diff --git a/sdext/source/pdfimport/wrapper/wrapper.cxx b/sdext/source/pdfimport/wrapper/wrapper.cxx
index 331b7900dff1..8b0abf34281b 100644
--- a/sdext/source/pdfimport/wrapper/wrapper.cxx
+++ b/sdext/source/pdfimport/wrapper/wrapper.cxx
@@ -1011,7 +1011,8 @@ bool xpdf_ImportFromFile( const OUString&                             rURL,
                           const ContentSinkSharedPtr&                        rSink,
                           const uno::Reference< task::XInteractionHandler >& xIHdl,
                           const OUString&                               rPwd,
-                          const uno::Reference< uno::XComponentContext >&    xContext )
+                          const uno::Reference< uno::XComponentContext >&    xContext,
+                          const OUString&                                    rFilterOptions )
 {
     OSL_ASSERT(rSink);
 
@@ -1054,8 +1055,10 @@ bool xpdf_ImportFromFile( const OUString&                             rURL,
 
     // spawn separate process to keep LGPL/GPL code apart.
 
-    rtl_uString*  args[] = { aSysUPath.pData, errPathname.pData };
-    sal_Int32 nArgs = 2;
+    OUString aOptFlag("-o");
+    rtl_uString*  args[] = { aSysUPath.pData, errPathname.pData,
+                             aOptFlag.pData, rFilterOptions.pData };
+    sal_Int32 nArgs = rFilterOptions.isEmpty() ? 2 : 4;
 
     oslProcess    aProcess;
     oslFileHandle pIn  = NULL;
@@ -1160,7 +1163,8 @@ bool xpdf_ImportFromStream( const uno::Reference< io::XInputStream >&         xI
                             const ContentSinkSharedPtr&                       rSink,
                             const uno::Reference<task::XInteractionHandler >& xIHdl,
                             const OUString&                              rPwd,
-                            const uno::Reference< uno::XComponentContext >&   xContext )
+                            const uno::Reference< uno::XComponentContext >&   xContext,
+                            const OUString&                                   rFilterOptions )
 {
     OSL_ASSERT(xInput.is());
     OSL_ASSERT(rSink);
@@ -1203,7 +1207,7 @@ bool xpdf_ImportFromStream( const uno::Reference< io::XInputStream >&         xI
     osl_closeFile( aFile );
 
     if ( bSuccess )
-        bSuccess = xpdf_ImportFromFile( aURL, rSink, xIHdl, rPwd, xContext );
+        bSuccess = xpdf_ImportFromFile( aURL, rSink, xIHdl, rPwd, xContext, rFilterOptions );
     osl_removeFile( aURL.pData );
 
     return bSuccess;
diff --git a/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.cxx b/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.cxx
index d3e39887e32e..cf19b30834a7 100644
--- a/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.cxx
+++ b/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.cxx
@@ -532,7 +532,8 @@ void PDFOutDev::printPath( GfxPath* pPath ) const
 PDFOutDev::PDFOutDev( PDFDoc* pDoc ) :
     m_pDoc( pDoc ),
     m_aFontMap(),
-    m_pUtf8Map( new UnicodeMap((char*)"UTF-8", gTrue, &mapUTF8) )
+    m_pUtf8Map( new UnicodeMap((char*)"UTF-8", gTrue, &mapUTF8) ),
+    m_bSkipImages(false)
 {
 }
 PDFOutDev::~PDFOutDev()
@@ -632,6 +633,8 @@ void PDFOutDev::updateCTM(GfxState* state,
 
 void PDFOutDev::updateLineDash(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     double* dashArray; int arrayLen; double startOffset;
@@ -649,36 +652,48 @@ void PDFOutDev::updateLineDash(GfxState *state)
 
 void PDFOutDev::updateFlatness(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
     printf( "updateFlatness %d\n", state->getFlatness() );
 }
 
 void PDFOutDev::updateLineJoin(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
     printf( "updateLineJoin %d\n", state->getLineJoin() );
 }
 
 void PDFOutDev::updateLineCap(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
     printf( "updateLineCap %d\n", state->getLineCap() );
 }
 
 void PDFOutDev::updateMiterLimit(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
     printf( "updateMiterLimit %f\n", normalize(state->getMiterLimit()) );
 }
 
 void PDFOutDev::updateLineWidth(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
     printf( "updateLineWidth %f\n", normalize(state->getLineWidth()) );
 }
 
 void PDFOutDev::updateFillColor(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     GfxRGB aRGB;
@@ -693,6 +708,8 @@ void PDFOutDev::updateFillColor(GfxState *state)
 
 void PDFOutDev::updateStrokeColor(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     GfxRGB aRGB;
@@ -707,11 +724,15 @@ void PDFOutDev::updateStrokeColor(GfxState *state)
 
 void PDFOutDev::updateFillOpacity(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     updateFillColor(state);
 }
 
 void PDFOutDev::updateStrokeOpacity(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     updateStrokeColor(state);
 }
 
@@ -774,6 +795,8 @@ void PDFOutDev::updateRender(GfxState *state)
 
 void PDFOutDev::stroke(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     printf( "strokePath" );
@@ -783,6 +806,8 @@ void PDFOutDev::stroke(GfxState *state)
 
 void PDFOutDev::fill(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     printf( "fillPath" );
@@ -792,6 +817,8 @@ void PDFOutDev::fill(GfxState *state)
 
 void PDFOutDev::eoFill(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     printf( "eoFillPath" );
@@ -801,6 +828,8 @@ void PDFOutDev::eoFill(GfxState *state)
 
 void PDFOutDev::clip(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     printf( "clipPath" );
@@ -810,6 +839,8 @@ void PDFOutDev::clip(GfxState *state)
 
 void PDFOutDev::eoClip(GfxState *state)
 {
+    if (m_bSkipImages)
+        return;
     assert(state);
 
     printf( "eoClipPath" );
@@ -915,6 +946,8 @@ void PDFOutDev::drawImageMask(GfxState* pState, Object*, Stream* str,
 #endif
                               GBool /*inlineImg*/ )
 {
+    if (m_bSkipImages)
+        return;
     OutputBuffer aBuf; initBuf(aBuf);
 
     printf( "drawMask %d %d %d", width, height, invert );
@@ -944,6 +977,8 @@ void PDFOutDev::drawImage(GfxState*, Object*, Stream* str,
 #endif
                           int* maskColors, GBool /*inlineImg*/ )
 {
+    if (m_bSkipImages)
+        return;
     OutputBuffer aBuf; initBuf(aBuf);
     OutputBuffer aMaskBuf;
 
@@ -999,6 +1034,8 @@ void PDFOutDev::drawMaskedImage(GfxState*, Object*, Stream* str,
 #endif
                                )
 {
+    if (m_bSkipImages)
+        return;
     OutputBuffer aBuf;     initBuf(aBuf);
     printf( "drawImage %d %d 0", width, height );
     writePng_( aBuf, str, width, height, colorMap, maskStr, maskWidth, maskHeight, maskInvert, true );
@@ -1019,6 +1056,8 @@ void PDFOutDev::drawSoftMaskedImage(GfxState*, Object*, Stream* str,
 #endif
                                    )
 {
+    if (m_bSkipImages)
+        return;
     OutputBuffer aBuf;     initBuf(aBuf);
     printf( "drawImage %d %d 0", width, height );
     writePng_( aBuf, str, width, height, colorMap, maskStr, maskWidth, maskHeight, maskColorMap, true );
@@ -1031,6 +1070,11 @@ void PDFOutDev::setPageNum( int nNumPages )
     printf("setPageNum %d\n", nNumPages);
 }
 
+void PDFOutDev::setSkipImages( bool bSkipImages )
+{
+    m_bSkipImages = bSkipImages;
+}
+
 }
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.hxx b/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.hxx
index daba4454a4d8..cc64db43a9a4 100644
--- a/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.hxx
+++ b/sdext/source/pdfimport/xpdfwrapper/pdfioutdev_gpl.hxx
@@ -138,6 +138,7 @@ namespace pdfi
         mutable std::unordered_map< long long,
                                FontAttributes > m_aFontMap;
         UnicodeMap*                             m_pUtf8Map;
+        bool                                    m_bSkipImages;
 
         int  parseFont( long long nNewId, GfxFont* pFont, GfxState* state ) const;
         void writeFontFile( GfxFont* gfxFont ) const;
@@ -267,6 +268,7 @@ namespace pdfi
                                         ) SAL_OVERRIDE;
 
         void setPageNum( int nNumPages );
+        void setSkipImages ( bool bSkipImages );
     };
 }
 
diff --git a/sdext/source/pdfimport/xpdfwrapper/wrapper_gpl.cxx b/sdext/source/pdfimport/xpdfwrapper/wrapper_gpl.cxx
index 20c45c49a1dd..c72ba95d5f60 100644
--- a/sdext/source/pdfimport/xpdfwrapper/wrapper_gpl.cxx
+++ b/sdext/source/pdfimport/xpdfwrapper/wrapper_gpl.cxx
@@ -27,7 +27,8 @@ FILE* g_binary_out=stderr;
 
 static const char *ownerPassword = "\001";
 static const char *userPassword  = "\001";
-static const char *outputFile   = "\001";
+static const char *outputFile    = "\001";
+static const char *options       = "\001";
 
 int main(int argc, char **argv)
 {
@@ -41,6 +42,14 @@ int main(int argc, char **argv)
             for (int j = k; j < argc; ++j)
                 argv[j] = argv[j+2];
         }
+        else if (!strcmp(argv[k], "-o"))
+        {
+            options = argv[k+1];
+            argc -= 2;
+            for (int j = k; j < argc; ++j)
+                argv[j] = argv[j+2];
+        }
+
         else if (!strcmp(argv[k], "-opw"))
         {
             ownerPassword = argv[k+1];
@@ -58,9 +67,6 @@ int main(int argc, char **argv)
     ++k;
     }
 
-    if( argc != 3 )
-        return 1;
-
     // read config file
     globalParams = new GlobalParams();
     globalParams->setErrQuiet(gTrue);
@@ -119,6 +125,9 @@ int main(int argc, char **argv)
     PDFDoc &rDoc = aDoc.isOk()? aDoc: aErrDoc;
 
     pdfi::PDFOutDev aOutDev(&rDoc);
+    if (!strcmp(options, "SkipImages")) {
+            aOutDev.setSkipImages(true);
+    }
 
     // tell the receiver early - needed for proper progress calculation
     const int nPages = rDoc.isOk()? rDoc.getNumPages(): 0;