tdf#105461 vcl: add text highlight textcase

Fails with commit ee32c7d8083ae1449d6b379034be92995c142da9 (tdf#105461
PDF export: handle text fill color, 2017-02-01) reverted.

Change-Id: I3628a16d0810e3be3fb352340d06cdba472dcd3f
Reviewed-on: https://gerrit.libreoffice.org/34621
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>

4 files changed, 80 insertions, 13 deletions

diff --git a/include/xmlsecurity/pdfio/pdfdocument.hxx b/include/xmlsecurity/pdfio/pdfdocument.hxx
index 0b27014c7418..5a8ea6c2c71d 100644
--- a/include/xmlsecurity/pdfio/pdfdocument.hxx
+++ b/include/xmlsecurity/pdfio/pdfdocument.hxx
@@ -80,6 +80,8 @@ public:
     void SetDictionary(PDFDictionaryElement* pDictionaryElement);
     void SetArray(PDFArrayElement* pArrayElement);
     void SetStream(PDFStreamElement* pStreamElement);
+    /// Access to the stream of the object, if it has any.
+    PDFStreamElement* GetStream() const;
     PDFArrayElement* GetArray() const;
     /// Parse objects stored in this object stream.
     void ParseStoredObjects();
@@ -88,6 +90,21 @@ public:
     void SetStreamBuffer(std::unique_ptr<SvMemoryStream>& pStreamBuffer);
 };
 
+/// Stream object: a byte array with a known length.
+class XMLSECURITY_DLLPUBLIC PDFStreamElement : public PDFElement
+{
+    size_t m_nLength;
+    sal_uInt64 m_nOffset;
+    /// The byte array itself.
+    SvMemoryStream m_aMemory;
+
+public:
+    explicit PDFStreamElement(size_t nLength);
+    bool Read(SvStream& rStream) override;
+    sal_uInt64 GetOffset() const;
+    SvMemoryStream& GetMemory();
+};
+
 /// Name object: a key string.
 class XMLSECURITY_DLLPUBLIC PDFNameElement : public PDFElement
 {
diff --git a/vcl/qa/cppunit/pdfexport/data/tdf105461.odp b/vcl/qa/cppunit/pdfexport/data/tdf105461.odp
new file mode 100644
index 000000000000..9c86a3bd7901
--- /dev/null
+++ b/vcl/qa/cppunit/pdfexport/data/tdf105461.odp
diff --git a/vcl/qa/cppunit/pdfexport/pdfexport.cxx b/vcl/qa/cppunit/pdfexport/pdfexport.cxx
index 949ef6109043..1c665b8d2174 100644
--- a/vcl/qa/cppunit/pdfexport/pdfexport.cxx
+++ b/vcl/qa/cppunit/pdfexport/pdfexport.cxx
@@ -19,6 +19,7 @@
 #include <unotools/mediadescriptor.hxx>
 #include <unotools/tempfile.hxx>
 #include <xmlsecurity/pdfio/pdfdocument.hxx>
+#include <tools/zcodec.hxx>
 
 using namespace ::com::sun::star;
 
@@ -39,11 +40,14 @@ public:
 #if HAVE_FEATURE_PDFIUM
     /// Tests that a pdf image is roundtripped back to PDF as a vector format.
     void testTdf106059();
+    /// Tests that text highlight from Impress is not lost.
+    void testTdf105461();
 #endif
 
     CPPUNIT_TEST_SUITE(PdfExportTest);
 #if HAVE_FEATURE_PDFIUM
     CPPUNIT_TEST(testTdf106059);
+    CPPUNIT_TEST(testTdf105461);
 #endif
     CPPUNIT_TEST_SUITE_END();
 };
@@ -100,6 +104,52 @@ void PdfExportTest::testTdf106059()
     // This dictionary key was missing, so the XObject wasn't a reference one.
     CPPUNIT_ASSERT(pReferenceXObject->Lookup("Ref"));
 }
+
+void PdfExportTest::testTdf105461()
+{
+    // Import the bugdoc and export as PDF.
+    OUString aURL = m_directories.getURLFromSrc(DATA_DIRECTORY) + "tdf105461.odp";
+    mxComponent = loadFromDesktop(aURL);
+    CPPUNIT_ASSERT(mxComponent.is());
+
+    uno::Reference<frame::XStorable> xStorable(mxComponent, uno::UNO_QUERY);
+    utl::TempFile aTempFile;
+    aTempFile.EnableKillingFile();
+    utl::MediaDescriptor aMediaDescriptor;
+    aMediaDescriptor["FilterName"] <<= OUString("impress_pdf_Export");
+    xStorable->storeToURL(aTempFile.GetURL(), aMediaDescriptor.getAsConstPropertyValueList());
+
+    // Parse the export result.
+    xmlsecurity::pdfio::PDFDocument aDocument;
+    SvFileStream aStream(aTempFile.GetURL(), StreamMode::READ);
+    CPPUNIT_ASSERT(aDocument.Read(aStream));
+
+    // The document has one page.
+    std::vector<xmlsecurity::pdfio::PDFObjectElement*> aPages = aDocument.GetPages();
+    CPPUNIT_ASSERT_EQUAL(static_cast<size_t>(1), aPages.size());
+
+    // The page has a stream.
+    xmlsecurity::pdfio::PDFObjectElement* pContents = aPages[0]->LookupObject("Contents");
+    CPPUNIT_ASSERT(pContents);
+    xmlsecurity::pdfio::PDFStreamElement* pStream = pContents->GetStream();
+    CPPUNIT_ASSERT(pStream);
+    SvMemoryStream& rObjectStream = pStream->GetMemory();
+    // Uncompress it.
+    SvMemoryStream aUncompressed;
+    ZCodec aZCodec;
+    aZCodec.BeginCompression();
+    rObjectStream.Seek(0);
+    aZCodec.Decompress(rObjectStream, aUncompressed);
+    CPPUNIT_ASSERT(aZCodec.EndCompression());
+
+    // Make sure there is a filled rectangle inside.
+    OString aFilledRectangle("re f*");
+    auto pStart = static_cast<const char*>(aUncompressed.GetData());
+    const char* pEnd = pStart + aUncompressed.GetSize();
+    auto it = std::search(pStart, pEnd, aFilledRectangle.getStr(), aFilledRectangle.getStr() + aFilledRectangle.getLength());
+    // This failed, stream contained no filled rectangle.
+    CPPUNIT_ASSERT(it != pEnd);
+}
 #endif
 
 CPPUNIT_TEST_SUITE_REGISTRATION(PdfExportTest);
diff --git a/xmlsecurity/source/pdfio/pdfdocument.cxx b/xmlsecurity/source/pdfio/pdfdocument.cxx
index 50e00d79c067..50174483a832 100644
--- a/xmlsecurity/source/pdfio/pdfdocument.cxx
+++ b/xmlsecurity/source/pdfio/pdfdocument.cxx
@@ -118,18 +118,6 @@ public:
     int GetGenerationValue() const;
 };
 
-/// Stream object: a byte array with a known length.
-class PDFStreamElement : public PDFElement
-{
-    size_t m_nLength;
-    sal_uInt64 m_nOffset;
-
-public:
-    explicit PDFStreamElement(size_t nLength);
-    bool Read(SvStream& rStream) override;
-    sal_uInt64 GetOffset() const;
-};
-
 /// End of a stream: 'endstream' keyword.
 class PDFEndStreamElement : public PDFElement
 {
@@ -3282,6 +3270,11 @@ void PDFObjectElement::SetStream(PDFStreamElement* pStreamElement)
     m_pStreamElement = pStreamElement;
 }
 
+PDFStreamElement* PDFObjectElement::GetStream() const
+{
+    return m_pStreamElement;
+}
+
 PDFArrayElement* PDFObjectElement::GetArray() const
 {
     return m_pArrayElement;
@@ -3668,11 +3661,18 @@ bool PDFStreamElement::Read(SvStream& rStream)
 {
     SAL_INFO("xmlsecurity.pdfio", "PDFStreamElement::Read: length is " << m_nLength);
     m_nOffset = rStream.Tell();
-    rStream.SeekRel(m_nLength);
+    std::vector<unsigned char> aBytes(m_nLength);
+    rStream.ReadBytes(aBytes.data(), aBytes.size());
+    m_aMemory.WriteBytes(aBytes.data(), aBytes.size());
 
     return rStream.good();
 }
 
+SvMemoryStream& PDFStreamElement::GetMemory()
+{
+    return m_aMemory;
+}
+
 sal_uInt64 PDFStreamElement::GetOffset() const
 {
     return m_nOffset;