vcl,sw: PDF/UA export: tag headers and footers as required

ISO 14289-1:2014 has one requirement for specific tagging of artifacts:

7.8 Page headers and footers
Running headers and footers shall be identified as Pagination
artifacts and shall be classified as either Header or Footer
subtypes as per ISO 32000-1:2008, 14.8.2.2.2, Table 330.

It was not immediately obvious how to implement this but the functions
used for tunnelling structure element attributes through MetaFile can
be used for this purpose as well with a few tweaks.

Change-Id: I19a3192b1b56b82ed11972c4bbe8d20ab13567be
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/148387
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@allotropia.de>

4 files changed, 92 insertions, 3 deletions

diff --git a/include/vcl/pdfwriter.hxx b/include/vcl/pdfwriter.hxx
index 8e3c139e8943..1e21fae1c5bd 100644
--- a/include/vcl/pdfwriter.hxx
+++ b/include/vcl/pdfwriter.hxx
@@ -135,6 +135,9 @@ public:
 
     enum StructAttribute
     {
+        // Artifacts
+        Type, Subtype,
+
         Placement, WritingMode, SpaceBefore, SpaceAfter, StartIndent, EndIndent,
         TextIndent, TextAlign, Width, Height, BlockAlign, InlineAlign,
         LineHeight, BaselineShift, TextDecorationType, ListNumbering,
@@ -158,6 +161,9 @@ public:
     {
         Invalid,
         NONE,
+        // Artifacts
+        Pagination, Layout, Page, Background,
+        Header, Footer, Watermark,
         // Placement
         Block, Inline, Before, After, Start, End,
         // WritingMode
diff --git a/sw/source/core/text/EnhancedPDFExportHelper.cxx b/sw/source/core/text/EnhancedPDFExportHelper.cxx
index 1dde77106624..e9f8d310d3fe 100644
--- a/sw/source/core/text/EnhancedPDFExportHelper.cxx
+++ b/sw/source/core/text/EnhancedPDFExportHelper.cxx
@@ -602,6 +602,18 @@ void SwTaggedPDFHelper::SetAttributes( vcl::PDFWriter::StructElement eType )
                 }
                 break;
 
+            case vcl::PDFWriter::NonStructElement:
+                if (pFrame->IsHeaderFrame() || pFrame->IsFooterFrame())
+                {
+                    // ISO 14289-1:2014, Clause: 7.8
+                    mpPDFExtOutDevData->SetStructureAttribute(vcl::PDFWriter::Type, vcl::PDFWriter::Pagination);
+                    mpPDFExtOutDevData->SetStructureAttribute(vcl::PDFWriter::Subtype,
+                        pFrame->IsHeaderFrame()
+                           ? vcl::PDFWriter::Header
+                           : vcl::PDFWriter::Footer);
+                }
+            break;
+
             default :
                 break;
         }
diff --git a/vcl/qa/cppunit/pdfexport/pdfexport.cxx b/vcl/qa/cppunit/pdfexport/pdfexport.cxx
index 4cf31e708220..30d9b5513c97 100644
--- a/vcl/qa/cppunit/pdfexport/pdfexport.cxx
+++ b/vcl/qa/cppunit/pdfexport/pdfexport.cxx
@@ -3050,7 +3050,8 @@ CPPUNIT_TEST_FIXTURE(PdfExportTest, testTdf139736)
 
     // Enable PDF/UA
     uno::Sequence<beans::PropertyValue> aFilterData(
-        comphelper::InitPropertySequence({ { "PDFUACompliance", uno::Any(true) } }));
+        comphelper::InitPropertySequence({ { "PDFUACompliance", uno::Any(true) },
+                                           { "SelectPdfVersion", uno::Any(sal_Int32(17)) } }));
     aMediaDescriptor["FilterData"] <<= aFilterData;
     saveAsPDF(u"tdf139736-1.odt");
 
@@ -3081,6 +3082,8 @@ CPPUNIT_TEST_FIXTURE(PdfExportTest, testTdf139736)
     {
         Default,
         Artifact,
+        ArtifactProps1,
+        ArtifactProps2,
         Tagged
     } state
         = Default;
@@ -3107,6 +3110,23 @@ CPPUNIT_TEST_FIXTURE(PdfExportTest, testTdf139736)
                 state = Artifact;
                 ++nArtifacts;
             }
+            else if (o3tl::starts_with(line, "/Artifact <<"))
+            {
+                CPPUNIT_ASSERT_EQUAL_MESSAGE("unexpected nesting", Default, state);
+                // check header/footer properties
+                CPPUNIT_ASSERT_EQUAL(std::string_view("/Type/Pagination"), line.substr(12));
+                state = ArtifactProps1;
+                ++nArtifacts;
+            }
+            else if (state == ArtifactProps1)
+            {
+                CPPUNIT_ASSERT_EQUAL(std::string_view("/Subtype/Header"), line);
+                state = ArtifactProps2;
+            }
+            else if (state == ArtifactProps2 && line == ">> BDC")
+            {
+                state = Artifact;
+            }
             else if (line == "/Standard<</MCID 0>>BDC")
             {
                 CPPUNIT_ASSERT_EQUAL_MESSAGE("unexpected nesting", Default, state);
diff --git a/vcl/source/gdi/pdfwriter_impl.cxx b/vcl/source/gdi/pdfwriter_impl.cxx
index bbb904458627..a841842a6942 100644
--- a/vcl/source/gdi/pdfwriter_impl.cxx
+++ b/vcl/source/gdi/pdfwriter_impl.cxx
@@ -1830,6 +1830,8 @@ const char* PDFWriterImpl::getAttributeTag( PDFWriter::StructAttribute eAttr )
         aAttributeStrings[ PDFWriter::RowSpan ]             = "RowSpan";
         aAttributeStrings[ PDFWriter::ColSpan ]             = "ColSpan";
         aAttributeStrings[ PDFWriter::Scope ]               = "Scope";
+        aAttributeStrings[ PDFWriter::Type ]                = "Type";
+        aAttributeStrings[ PDFWriter::Subtype ]             = "Subtype";
         aAttributeStrings[ PDFWriter::LinkAnnotation ]      = "LinkAnnotation";
     }
 
@@ -1869,6 +1871,13 @@ const char* PDFWriterImpl::getAttributeValueTag( PDFWriter::StructAttributeValue
         aValueStrings[ PDFWriter::Row ]                     = "Row";
         aValueStrings[ PDFWriter::Column ]                  = "Column";
         aValueStrings[ PDFWriter::Both ]                    = "Both";
+        aValueStrings[ PDFWriter::Pagination ]              = "Pagination";
+        aValueStrings[ PDFWriter::Layout ]                  = "Layout";
+        aValueStrings[ PDFWriter::Page ]                    = "Page";
+        aValueStrings[ PDFWriter::Background ]              = "Background";
+        aValueStrings[ PDFWriter::Header ]                  = "Header";
+        aValueStrings[ PDFWriter::Footer ]                  = "Footer";
+        aValueStrings[ PDFWriter::Watermark ]               = "Watermark";
         aValueStrings[ PDFWriter::Disc ]                    = "Disc";
         aValueStrings[ PDFWriter::Circle ]                  = "Circle";
         aValueStrings[ PDFWriter::Square ]                  = "Square";
@@ -10541,8 +10550,24 @@ void PDFWriterImpl::beginStructureElementMCSeq()
              ! m_aStructure[ m_nCurrentStructElement ].m_bOpenMCSeq // already opened sequence
              )
     {
-        OString aLine = "/Artifact BMC\n";
+        OString aLine = "/Artifact ";
         writeBuffer( aLine.getStr(), aLine.getLength() );
+        // emit property list if requested
+        OStringBuffer buf;
+        for (auto const& rAttr : m_aStructure[m_nCurrentStructElement].m_aAttributes)
+        {
+            appendStructureAttributeLine(rAttr.first, rAttr.second, buf, false);
+        }
+        if (buf.isEmpty())
+        {
+            writeBuffer("BMC\n", 4);
+        }
+        else
+        {
+            writeBuffer("<<", 2);
+            writeBuffer(buf.getStr(), buf.getLength());
+            writeBuffer(">> BDC\n", 7);
+        }
         // mark element MC sequence as open
         m_aStructure[ m_nCurrentStructElement ].m_bOpenMCSeq = true;
     }
@@ -10841,7 +10866,12 @@ bool PDFWriterImpl::setStructureAttribute( enum PDFWriter::StructAttribute eAttr
         return false;
 
     bool bInsert = false;
-    if( m_nCurrentStructElement > 0 && m_bEmitStructure )
+    if (m_nCurrentStructElement > 0
+        && (m_bEmitStructure
+            // allow it for topmost non-structured element
+            || (m_aContext.Tagged
+                && 0 < m_aStructure[m_nCurrentStructElement].m_nParentElement
+                && m_aStructure[m_aStructure[m_nCurrentStructElement].m_nParentElement].m_eType != PDFWriter::NonStructElement)))
     {
         PDFWriter::StructElement eType = m_aStructure[ m_nCurrentStructElement ].m_eType;
         switch( eAttr )
@@ -11008,6 +11038,27 @@ bool PDFWriterImpl::setStructureAttribute( enum PDFWriter::StructAttribute eAttr
                     }
                 }
                 break;
+            case PDFWriter::Type:
+                if (eVal == PDFWriter::Pagination || eVal == PDFWriter::Layout || eVal == PDFWriter::Page)
+                    // + Background for PDF >= 1.7
+                {
+                    if (eType == PDFWriter::NonStructElement)
+                    {
+                        bInsert = true;
+                    }
+                }
+                break;
+            case PDFWriter::Subtype:
+                if (eVal == PDFWriter::Header || eVal == PDFWriter::Footer || eVal == PDFWriter::Watermark)
+                {
+                    if (eType == PDFWriter::NonStructElement
+                        && m_aContext.Version != PDFWriter::PDFVersion::PDF_A_1
+                        && PDFWriter::PDFVersion::PDF_1_7 <= m_aContext.Version)
+                    {
+                        bInsert = true;
+                    }
+                }
+                break;
             case PDFWriter::ListNumbering:
                 if( eVal == PDFWriter::NONE         ||
                     eVal == PDFWriter::Disc         ||