fdo#67370: Hyphens are not visible in tagged PDF

One requirement of tagged PDF is to represent automatically inserted
hyphens using the soft hyphen (U+00AD) character, so we were doing this
by simply passing that character to text layout code when exporting a
tagged PDF (which is the literal suggestion of old PDF specification).

This is wrong, though, since the soft hyphen is a control character and
should not have a visible output by itself (and fonts might not even
have a visible glyph there), but this happened to work because non of
the layout engines we are using treated soft hyphen specially and was
just showing whatever glyph the font had there. This broke with the
switch to HarfBuzz since it will not show any visible glyph for Unicode
control characters (by default), which is the right thing to do.

Latest versions of PDF spec suggest using either ToUnicode mapping or an
ActualText text entry to encode the soft hyphen instead, I found it
easier to use ActualText since we already have code that handles
non-standard hyphenation using it already.

Change-Id: I88deadf3a806f69775b2e0ccff2f9b2f61a0f2e2

2 files changed, 11 insertions, 12 deletions

diff --git a/sw/source/core/text/EnhancedPDFExportHelper.cxx b/sw/source/core/text/EnhancedPDFExportHelper.cxx
index fe8b0bd7c9a2..462576aa49b4 100644
--- a/sw/source/core/text/EnhancedPDFExportHelper.cxx
+++ b/sw/source/core/text/EnhancedPDFExportHelper.cxx
@@ -760,7 +760,8 @@ void SwTaggedPDFHelper::SetAttributes( vcl::PDFWriter::StructElement eType )
             case vcl::PDFWriter::Span :
             case vcl::PDFWriter::Quote :
             case vcl::PDFWriter::Code :
-                if( POR_HYPHSTR == pPor->GetWhichPor() || POR_SOFTHYPHSTR == pPor->GetWhichPor() )
+                if( POR_HYPHSTR == pPor->GetWhichPor() || POR_SOFTHYPHSTR == pPor->GetWhichPor() ||
+                    POR_HYPH == pPor->GetWhichPor() || POR_SOFTHYPH == pPor->GetWhichPor() )
                     bActualText = true;
                 else
                 {
@@ -783,7 +784,11 @@ void SwTaggedPDFHelper::SetAttributes( vcl::PDFWriter::StructElement eType )
 
         if ( bActualText )
         {
-            const OUString aActualTxt = rInf.GetTxt().copy( rInf.GetIdx(), pPor->GetLen() );
+            OUString aActualTxt;
+            if (pPor->GetWhichPor() == POR_SOFTHYPH || pPor->GetWhichPor() == POR_HYPH)
+                aActualTxt = OUString(0xad); // soft hyphen
+            else
+                aActualTxt = rInf.GetTxt().copy(rInf.GetIdx(), pPor->GetLen());
             mpPDFExtOutDevData->SetActualText( aActualTxt );
         }
 
@@ -1364,6 +1369,8 @@ void SwTaggedPDFHelper::BeginInlineStructureElements()
 
     switch ( pPor->GetWhichPor() )
     {
+        case POR_HYPH :
+        case POR_SOFTHYPH :
         // Check for alternative spelling:
         case POR_HYPHSTR :
         case POR_SOFTHYPHSTR :
diff --git a/sw/source/core/text/txthyph.cxx b/sw/source/core/text/txthyph.cxx
index 4aa235e01ea7..accee9b4a976 100644
--- a/sw/source/core/text/txthyph.cxx
+++ b/sw/source/core/text/txthyph.cxx
@@ -20,7 +20,6 @@
 #include <hintids.hxx>
 #include <editeng/unolingu.hxx>
 #include <com/sun/star/i18n/WordType.hpp>
-#include <EnhancedPDFExportHelper.hxx>
 #include <viewopt.hxx>
 #include <viewsh.hxx>
 #include <SwPortionHandler.hxx>
@@ -370,16 +369,9 @@ sal_Bool SwTxtPortion::CreateHyphen( SwTxtFormatInfo &rInf, SwTxtGuess &rGuess )
  *              virtual SwHyphPortion::GetExpTxt()
  *************************************************************************/
 
-sal_Bool SwHyphPortion::GetExpTxt( const SwTxtSizeInfo &rInf, OUString &rTxt ) const
+sal_Bool SwHyphPortion::GetExpTxt( const SwTxtSizeInfo &/*rInf*/, OUString &rTxt ) const
 {
-    // #i16816# tagged pdf support
-    const sal_Unicode cChar = rInf.GetVsh() &&
-                              rInf.GetVsh()->GetViewOptions()->IsPDFExport() &&
-                              SwTaggedPDFHelper::IsExportTaggedPDF( *rInf.GetOut() ) ?
-                              0xad :
-                              '-';
-
-    rTxt = OUString(cChar);
+    rTxt = "-";
     return sal_True;
 }