sw XHTML import: handle non-image, non-RTF objects as clickable images

A pair of XHTML <object> elements is meant to be interpreted as "native
data" (outer) and "fallback/preview image" (inner). In practice we map
non-PNG + PNG pairs to images and OLE2 data + PNG pairs to OLE2 embedded
objects.

This setup works for OLE2 data and images, but XHTML producers can also
put other raw data to the outer <object> like PDF, DOCX, XLSX, etc.
These were mapped to ODummyEmbeddedObject, which preserves data, but you
can't interact with the data.

Fix the lack of interaction by changing how non-OLE2, non-image data is
handled: map them to images where the image has a URL to the native
data. This way the OS running Writer can decide how to handle that data.

This required changing some existing tests where the intention was to
have simple test data for OLE2 data, but that data was not OLE2, so the
resulting doc model had images, but embedded objects were expected. Such
tests now have OLE2 data instead.

(cherry picked from commit 56769d3982e6afb075cb6d833662f066437fab6a)

Conflicts:
	sw/qa/extras/htmlimport/htmlimport.cxx

Change-Id: I0287ce2d9a02904e28cef619ff9f6e1f354d6147

9 files changed, 62 insertions, 12 deletions

diff --git a/sw/qa/extras/htmlexport/data/reqif-ole-data.xhtml b/sw/qa/extras/htmlexport/data/reqif-ole-data.xhtml
index 9e0cfaa378b4..5f9c8940008e 100644
--- a/sw/qa/extras/htmlexport/data/reqif-ole-data.xhtml
+++ b/sw/qa/extras/htmlexport/data/reqif-ole-data.xhtml
@@ -1,3 +1,3 @@
 <reqif-xhtml:div>
-    <reqif-xhtml:object data="reqif-ole-data.ole" type="text/rtf"/>
+    <reqif-xhtml:object data="ole2.ole" type="text/rtf"/>
 </reqif-xhtml:div>
diff --git a/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml b/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml
index 6217412ae597..df3f4a048ec6 100644
--- a/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml
+++ b/sw/qa/extras/htmlexport/data/reqif-ole-img.xhtml
@@ -1,5 +1,5 @@
 <reqif-xhtml:div><reqif-xhtml:br/>
-    <reqif-xhtml:object data="reqif-ole-data.ole" type="text/rtf">
+    <reqif-xhtml:object data="ole2.ole" type="text/rtf">
         <reqif-xhtml:object data="reqif-ole-img.png?test=true" type="image/png">OLE Object</reqif-xhtml:object>
     </reqif-xhtml:object>
 </reqif-xhtml:div>
diff --git a/sw/qa/extras/htmlexport/htmlexport.cxx b/sw/qa/extras/htmlexport/htmlexport.cxx
index 8e78b288566f..42c13ccac3f6 100644
--- a/sw/qa/extras/htmlexport/htmlexport.cxx
+++ b/sw/qa/extras/htmlexport/htmlexport.cxx
@@ -660,13 +660,12 @@ DECLARE_HTMLEXPORT_ROUNDTRIP_TEST(testReqIfOleImg, "reqif-ole-img.xhtml")
     // Check mime/media types.
     CPPUNIT_ASSERT_EQUAL(OUString("image/png"), getProperty<OUString>(xGraphic, "MimeType"));
 
-    uno::Reference<document::XStorageBasedDocument> xStorageProvider(mxComponent, uno::UNO_QUERY);
-    uno::Reference<embed::XStorage> xStorage = xStorageProvider->getDocumentStorage();
-    auto aStreamName = getProperty<OUString>(xObject, "StreamName");
-    uno::Reference<io::XStream> xStream
-        = xStorage->openStreamElement(aStreamName, embed::ElementModes::READ);
+    uno::Reference<beans::XPropertySet> xObjectProps(xObject, uno::UNO_QUERY);
+    uno::Reference<io::XActiveDataStreamer> xStreamProvider(
+        xObjectProps->getPropertyValue("EmbeddedObject"), uno::UNO_QUERY);
+    uno::Reference<io::XSeekable> xStream(xStreamProvider->getStream(), uno::UNO_QUERY);
     // This was empty when either import or export handling was missing.
-    CPPUNIT_ASSERT_EQUAL(OUString("text/rtf"), getProperty<OUString>(xStream, "MediaType"));
+    CPPUNIT_ASSERT_EQUAL(static_cast<sal_Int64>(37888), xStream->getLength());
 
     // Check alternate text (it was empty, for export the 'alt' attribute was used).
     CPPUNIT_ASSERT_EQUAL(OUString("OLE Object"), getProperty<OUString>(xObject, "Title").trim());
diff --git a/sw/qa/extras/htmlimport/data/data.ole b/sw/qa/extras/htmlimport/data/data.ole
new file mode 100644
index 000000000000..d3dc23d793e2
--- /dev/null
+++ b/sw/qa/extras/htmlimport/data/data.ole
@@ -0,0 +1 @@
+{\pict}
diff --git a/sw/qa/extras/htmlimport/data/ole-data.xhtml b/sw/qa/extras/htmlimport/data/ole-data.xhtml
new file mode 100644
index 000000000000..e8f1910ad4c9
--- /dev/null
+++ b/sw/qa/extras/htmlimport/data/ole-data.xhtml
@@ -0,0 +1,7 @@
+<reqif-xhtml:div>
+<reqif-xhtml:p>
+<reqif-xhtml:object data="data.ole" type="application/octet-stream">
+    <reqif-xhtml:object data="ole2.png" type="image/png"></reqif-xhtml:object>
+</reqif-xhtml:object>
+</reqif-xhtml:p>
+</reqif-xhtml:div>
diff --git a/sw/qa/extras/htmlimport/htmlimport.cxx b/sw/qa/extras/htmlimport/htmlimport.cxx
index 1241bfec3b04..326af44fb298 100644
--- a/sw/qa/extras/htmlimport/htmlimport.cxx
+++ b/sw/qa/extras/htmlimport/htmlimport.cxx
@@ -489,6 +489,33 @@ CPPUNIT_TEST_FIXTURE(SwModelTestBase, testOleImgSvg)
     CPPUNIT_ASSERT_EQUAL(static_cast<sal_Int32>(1), xObjects->getCount());
 }
 
+CPPUNIT_TEST_FIXTURE(SwModelTestBase, testOleData)
+{
+    // Given an XHTML with an <object> (containing non-image, non-OLE2 data) and an inner <object>
+    // (containing PNG):
+    uno::Sequence<beans::PropertyValue> aLoadProperties = {
+        comphelper::makePropertyValue("FilterName", OUString("HTML (StarWriter)")),
+        comphelper::makePropertyValue("FilterOptions", OUString("xhtmlns=reqif-xhtml")),
+    };
+    OUString aURL = m_directories.getURLFromSrc(DATA_DIRECTORY) + "ole-data.xhtml";
+
+    // When loading the document:
+    mxComponent = loadFromDesktop(aURL, "com.sun.star.text.TextDocument", aLoadProperties);
+
+    // Then make sure the result is a single clickable Writer image:
+    uno::Reference<text::XTextGraphicObjectsSupplier> xSupplier(mxComponent, uno::UNO_QUERY);
+    uno::Reference<container::XIndexAccess> xObjects(xSupplier->getGraphicObjects(),
+                                                     uno::UNO_QUERY);
+    // Without the accompanying fix in place, this test would have failed with:
+    // - Expected: 0
+    // - Actual  : 1
+    // i.e. the image was not imported as a Writer image (but as an OLE object).
+    CPPUNIT_ASSERT_EQUAL(static_cast<sal_Int32>(1), xObjects->getCount());
+    uno::Reference<css::drawing::XShape> xShape = getShape(1);
+    // And then the image was not clickable: this was empty.
+    CPPUNIT_ASSERT(getProperty<OUString>(xShape, "HyperLinkURL").endsWith("/data.ole"));
+}
+
 CPPUNIT_PLUGIN_IMPLEMENT();
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/sw/source/filter/html/htmlgrin.cxx b/sw/source/filter/html/htmlgrin.cxx
index 628ef2704630..2a8e8c46dc13 100644
--- a/sw/source/filter/html/htmlgrin.cxx
+++ b/sw/source/filter/html/htmlgrin.cxx
@@ -865,6 +865,14 @@ IMAGE_SETEVENT:
         }
 
     }
+    else if (!m_aEmbedURL.isEmpty())
+    {
+        // This is an inner <object> image and the outer <object> has a URL for us. Set that on the
+        // image.
+        SwFormatURL aURL(pFlyFormat->GetURL());
+        aURL.SetURL(m_aEmbedURL, bIsMap);
+        pFlyFormat->SetFormatAttr(aURL);
+    }
 
     if( !aMacroItem.GetMacroTable().empty() )
     {
diff --git a/sw/source/filter/html/htmlplug.cxx b/sw/source/filter/html/htmlplug.cxx
index 7310d1284765..39a18a4487af 100644
--- a/sw/source/filter/html/htmlplug.cxx
+++ b/sw/source/filter/html/htmlplug.cxx
@@ -562,8 +562,8 @@ bool SwHTMLParser::InsertEmbed()
         aCnt.SwitchPersistence(xStorage);
         aObjName = aCnt.CreateUniqueObjectName();
         {
-            SvFileStream aFileStream(aURLObj.GetMainURL(INetURLObject::DecodeMechanism::NONE),
-                                     StreamMode::READ);
+            OUString aEmbedURL = aURLObj.GetMainURL(INetURLObject::DecodeMechanism::NONE);
+            SvFileStream aFileStream(aEmbedURL, StreamMode::READ);
             uno::Reference<io::XInputStream> xInStream;
             SvMemoryStream aMemoryStream;
 
@@ -601,8 +601,13 @@ bool SwHTMLParser::InsertEmbed()
             }
 
             if (!xInStream.is())
-                // Non-RTF case.
-                xInStream.set(new utl::OStreamWrapper(aFileStream));
+            {
+                // Object data is neither OLE2 in RTF, nor an image. Then map this to an URL that
+                // will be set on the inner image.
+                m_aEmbedURL = aEmbedURL;
+                // Signal success, so the outer object won't fall back to the image handler.
+                return true;
+            }
 
             if (!xObj.is())
             {
diff --git a/sw/source/filter/html/swhtml.hxx b/sw/source/filter/html/swhtml.hxx
index 59b7db28c467..c3cebcf8984d 100644
--- a/sw/source/filter/html/swhtml.hxx
+++ b/sw/source/filter/html/swhtml.hxx
@@ -478,6 +478,9 @@ class SwHTMLParser : public SfxHTMLParser, public SvtListener
 
     std::set<OUString> m_aAllowedRTFOLEMimeTypes;
 
+    /// This is the URL of the outer <object> data if it's not OLE2 or an image.
+    OUString m_aEmbedURL;
+
     void DeleteFormImpl();
 
     void DocumentDetected();