Merge pull request #2432 from hftf/hyphens

Docx Reader: Parse soft and non-breaking hyphen elements

src/Text/Pandoc/Readers/Docx.hs

@@ -206,11 +206,15 @@ runElemToInlines :: RunElem -> Inlines
 runElemToInlines (TextRun s) = text s
 runElemToInlines (LnBrk) = linebreak
 runElemToInlines (Tab) = space
+runElemToInlines (SoftHyphen) = text "\xad"
+runElemToInlines (NoBreakHyphen) = text "\x2011"
 
 runElemToString :: RunElem -> String
 runElemToString (TextRun s) = s
 runElemToString (LnBrk) = ['\n']
 runElemToString (Tab) = ['\t']
+runElemToString (SoftHyphen) = ['\xad']
+runElemToString (NoBreakHyphen) = ['\x2011']
 
 runToString :: Run -> String
 runToString (Run _ runElems) = concatMap runElemToString runElems

src/Text/Pandoc/Readers/Docx/Parse.hs

@@ -208,7 +208,7 @@ data Run = Run RunStyle [RunElem]
          | InlineDrawing FilePath B.ByteString
            deriving Show
 
-data RunElem = TextRun String | LnBrk | Tab
+data RunElem = TextRun String | LnBrk | Tab | SoftHyphen | NoBreakHyphen
              deriving Show
 
 data VertAlign = BaseLn | SupScrpt | SubScrpt
@@ -877,6 +877,8 @@ elemToRunElem ns element
                   map (\x -> fromMaybe x . getUnicode f . lowerFromPrivate $ x) str
   | isElem ns "w" "br" element = return LnBrk
   | isElem ns "w" "tab" element = return Tab
+  | isElem ns "w" "softHyphen" element = return SoftHyphen
+  | isElem ns "w" "noBreakHyphen" element = return NoBreakHyphen
   | isElem ns "w" "sym" element = return (getSymChar ns element)
   | otherwise = throwError WrongElem
   where

tests/Tests/Readers/Docx.hs

@@ -130,6 +130,10 @@ tests = [ testGroup "inlines"
             "literal tabs"
             "docx/tabs.docx"
             "docx/tabs.native"
+          , testCompare
+            "special punctuation"
+            "docx/special_punctuation.docx"
+            "docx/special_punctuation.native"
           , testCompare
             "normalizing inlines"
             "docx/normalize.docx"

tests/docx/special_punctuation.native

@@ -0,0 +1,2 @@
+[Para [Str "Soft",Space,Str "hyphen:",Space,Str "[\173]"]
+,Para [Str "Non-breaking",Space,Str "hyphen:",Space,Str "[\8209]"]]