Merge pull request #1354 from jkr/literalTab

Parse literal tabs in docx

6 files changed, 29 insertions, 4 deletions

diff --git a/src/Text/Pandoc/Readers/Docx.hs b/src/Text/Pandoc/Readers/Docx.hs
index 4035cde99..9c1d0c5e6 100644
--- a/src/Text/Pandoc/Readers/Docx.hs
+++ b/src/Text/Pandoc/Readers/Docx.hs
@@ -134,10 +134,12 @@ codeDivs = ["SourceCode"]
 runElemToInlines :: RunElem -> [Inline]
 runElemToInlines (TextRun s) = strToInlines s
 runElemToInlines (LnBrk) = [LineBreak]
+runElemToInlines (Tab) = [Space]
 
 runElemToString :: RunElem -> String
 runElemToString (TextRun s) = s
 runElemToString (LnBrk) = ['\n']
+runElemToString (Tab) = ['\t']
 
 runElemsToString :: [RunElem] -> String
 runElemsToString = concatMap runElemToString
@@ -148,6 +150,19 @@ strNormalize (Str "" : ils) = strNormalize ils
 strNormalize ((Str s) : (Str s') : l) = strNormalize ((Str (s++s')) : l)
 strNormalize (il:ils) = il : (strNormalize ils)
 
+blockNormalize :: Block -> Block
+blockNormalize (Plain (Space : ils)) = blockNormalize (Plain ils)
+blockNormalize (Plain ils) = Plain $ strNormalize ils
+blockNormalize (Para (Space : ils)) = blockNormalize (Para ils)
+blockNormalize (Para ils) = Para $ strNormalize ils
+blockNormalize (Header n attr (Space : ils)) =
+  blockNormalize $ Header n attr ils
+blockNormalize (Table (Space : ils) align width hdr cells) =
+  blockNormalize $ Table ils align width hdr cells
+blockNormalize (Table ils align width hdr cells) =
+  Table (strNormalize ils) align width hdr cells
+blockNormalize blk = blk
+
 runToInlines :: ReaderOptions -> Docx -> Run -> [Inline]
 runToInlines _ _ (Run rs runElems)
   | isJust (rStyle rs) && (fromJust (rStyle rs)) `elem` codeSpans =
@@ -294,7 +309,7 @@ makeImagesSelfContained _ inline = inline
 bodyToBlocks :: ReaderOptions -> Docx -> Body -> [Block]
 bodyToBlocks opts docx (Body bps) =
   bottomUp removeEmptyPars $
-  bottomUp strNormalize $
+  bottomUp blockNormalize $
   bottomUp spanRemove $
   bottomUp divRemove $
   map (makeHeaderAnchors) $
diff --git a/src/Text/Pandoc/Readers/Docx/Parse.hs b/src/Text/Pandoc/Readers/Docx/Parse.hs
index 22e9dd909..18200bcf9 100644
--- a/src/Text/Pandoc/Readers/Docx/Parse.hs
+++ b/src/Text/Pandoc/Readers/Docx/Parse.hs
@@ -466,7 +466,7 @@ data Run = Run RunStyle [RunElem]
          | Endnote String
            deriving Show
 
-data RunElem = TextRun String | LnBrk
+data RunElem = TextRun String | LnBrk | Tab
              deriving Show
 
 data RunStyle = RunStyle { isBold :: Bool
@@ -545,6 +545,9 @@ elemToRunElem ns element
   | qName (elName element) == "br" &&
     qURI (elName element) == (lookup "w" ns) =
       Just $ LnBrk
+  | qName (elName element) == "tab" &&
+    qURI (elName element) == (lookup "w" ns) =
+      Just $ Tab
   | otherwise = Nothing
 
 
diff --git a/tests/Tests/Readers/Docx.hs b/tests/Tests/Readers/Docx.hs
index 273f03f4d..3a13641a9 100644
--- a/tests/Tests/Readers/Docx.hs
+++ b/tests/Tests/Readers/Docx.hs
@@ -44,7 +44,12 @@ tests = [ testGroup "inlines"
           , testCompare
             "handling unicode input"
             "docx.unicode.docx"
-            "docx.unicode.native"]
+            "docx.unicode.native"
+          , testCompare
+            "literal tabs"
+            "docx.tabs.docx"
+            "docx.tabs.native"
+          ]
         , testGroup "blocks"
           [ testCompare
             "headers"
diff --git a/tests/docx.notes.native b/tests/docx.notes.native
index 1e9b6bba4..5a94b1999 100644
--- a/tests/docx.notes.native
+++ b/tests/docx.notes.native
@@ -1,2 +1,2 @@
 [Header 2 ("",[],[]) [Str "A",Space,Str "footnote"]
-,Para [Str "Test",Space,Str "footnote.",Note [Para [Space,Str "My",Space,Str "note."]],Space,Str "Test",Space,Str "endnote.",Note [Para [Space,Str "This",Space,Str "is",Space,Str "an",Space,Str "endnote",Space,Str "at",Space,Str "the",Space,Str "end",Space,Str "of",Space,Str "the",Space,Str "document."]]]]
+,Para [Str "Test",Space,Str "footnote.",Note [Para [Str "My",Space,Str "note."]],Space,Str "Test",Space,Str "endnote.",Note [Para [Str "This",Space,Str "is",Space,Str "an",Space,Str "endnote",Space,Str "at",Space,Str "the",Space,Str "end",Space,Str "of",Space,Str "the",Space,Str "document."]]]]
diff --git a/tests/docx.tabs.docx b/tests/docx.tabs.docx
new file mode 100644
index 000000000..6ff5f4bb1
--- /dev/null
+++ b/tests/docx.tabs.docx
diff --git a/tests/docx.tabs.native b/tests/docx.tabs.native
new file mode 100644
index 000000000..05461f20b
--- /dev/null
+++ b/tests/docx.tabs.native
@@ -0,0 +1,2 @@
+[Para [Str "Some",Space,Str "text",Space,Str "separated",Space,Str "by",Space,Str "a",Space,Str "tab."]
+,Para [Str "Tab-indented",Space,Str "text."]]