Merge pull request #2419 from mb21/bidi

Support bidirectional text output with XeLaTeX, ConTeXt and HTML

README

@@ -959,6 +959,25 @@ as `title`, `author`, and `date`) as well as the following:
     Currently only used by XeTeX through the generated
     `polyglossia-otherlangs` variable.
 
+`dir`
+:   the base direction of the document, either `rtl` (right-to-left)
+    or `ltr` (left-to-right).
+
+    For bidirectional documents, native pandoc `span`s and `div`s
+    with the `dir` attribute (value `rtl` or `ltr`) can be used to
+    override the base direction in some output formats.
+    This may not always be necessary if the final renderer
+    (e.g. the browser, when generating HTML) supports the
+    [Unicode Bidirectional Algorithm][uba-basics].
+
+    When using LaTeX for bidi documents, only the XeLaTeX engine
+    is fully supported (see `--latex-engine`).
+    Setting the `dir` variable enables bidirectional text
+    handling in LaTeX and ConTeXt, thus even if the base direction
+    is the default left-to-right, you should set `dir: ltr` for
+    documents that also contain some right-to-left script.
+
+
 `slidy-url`
 :   base URL for Slidy documents (defaults to
     `http://www.w3.org/Talks/Tools/Slidy2`)

src/Text/Pandoc/Writers/ConTeXt.hs

@@ -83,13 +83,18 @@ pandocToConTeXt options (Pandoc meta blocks) = do
                 $ metadata
   let context' =  defField "context-lang" (maybe "" (fromBcp47 . splitBy (=='-')) $
                     getField "lang" context)
-                  context
+                $ defField "context-dir" (toContextDir $ getField "dir" context)
+                $ context
   return $ if writerStandalone options
               then renderTemplate' (writerTemplate options) context'
               else main
 
--- escape things as needed for ConTeXt
+toContextDir :: Maybe String -> String
+toContextDir (Just "rtl") = "r2l"
+toContextDir (Just "ltr") = "l2r"
+toContextDir _            = ""
 
+-- | escape things as needed for ConTeXt
 escapeCharForConTeXt :: WriterOptions -> Char -> String
 escapeCharForConTeXt opts ch =
  let ligatures = writerTeXLigatures opts in
@@ -151,13 +156,18 @@ blockToConTeXt (CodeBlock _ str) =
   -- blankline because \stoptyping can't have anything after it, inc. '}'
 blockToConTeXt (RawBlock "context" str) = return $ text str <> blankline
 blockToConTeXt (RawBlock _ _ ) = return empty
-blockToConTeXt (Div (ident,_,_) bs) = do
+blockToConTeXt (Div (ident,_,kvs) bs) = do
   contents <- blockListToConTeXt bs
-  if null ident
-     then return contents
-     else return $
-          ("\\reference" <> brackets (text $ toLabel ident) <> braces empty <>
-            "%") $$ contents
+  let contents' = if null ident
+                     then contents
+                     else ("\\reference" <> brackets (text $ toLabel ident) <>
+                            braces empty <> "%") $$ contents
+  let align dir = blankline <> "\\startalignment[" <> dir <> "]"
+                    $$ contents' $$ "\\stopalignment" <> blankline
+  return $ case lookup "dir" kvs of
+             Just "rtl" -> align "righttoleft"
+             Just "ltr" -> align "lefttoright"
+             _          -> contents'
 blockToConTeXt (BulletList lst) = do
   contents <- mapM listItemToConTeXt lst
   return $ ("\\startitemize" <> if isTightList lst
@@ -335,7 +345,12 @@ inlineToConTeXt (Note contents) = do
               then text "\\footnote{" <> nest 2 contents' <> char '}'
               else text "\\startbuffer " <> nest 2 contents' <>
                    text "\\stopbuffer\\footnote{\\getbuffer}"
-inlineToConTeXt (Span _ ils) = inlineListToConTeXt ils
+inlineToConTeXt (Span (_,_,kvs) ils) = do
+  contents <- inlineListToConTeXt ils
+  return $ case lookup "dir" kvs of
+             Just "rtl" -> braces $ "\\righttoleft " <> contents
+             Just "ltr" -> braces $ "\\lefttoright " <> contents
+             _          -> contents
 
 -- | Craft the section header, inserting the secton reference, if supplied.
 sectionHeader :: Attr

src/Text/Pandoc/Writers/LaTeX.hs

@@ -187,6 +187,9 @@ pandocToLaTeX options (Pandoc meta blocks) = do
         $ defField "polyglossia-otherlangs"
             (maybe [] (map $ fst . toPolyglossia . splitBy (=='-')) $
             getField "otherlangs" context)
+        $ defField "latex-dir-rtl" (case (getField "dir" context)::Maybe String of
+                                      Just "rtl" -> True
+                                      _          -> False)
         $ context
   return $ if writerStandalone options
               then renderTemplate' template context'
@@ -323,14 +326,19 @@ isLineBreakOrSpace _ = False
 blockToLaTeX :: Block     -- ^ Block to convert
              -> State WriterState Doc
 blockToLaTeX Null = return empty
-blockToLaTeX (Div (identifier,classes,_) bs) = do
+blockToLaTeX (Div (identifier,classes,kvs) bs) = do
   beamer <- writerBeamer `fmap` gets stOptions
   ref <- toLabel identifier
   let linkAnchor = if null identifier
                       then empty
                       else "\\hyperdef{}" <> braces (text ref) <>
                            braces ("\\label" <> braces (text ref))
-  contents <- blockListToLaTeX bs
+  contents' <- blockListToLaTeX bs
+  let align dir = inCmd "begin" dir $$ contents' $$ inCmd "end" dir
+  let contents = case lookup "dir" kvs of
+                   Just "rtl" -> align "RTL"
+                   Just "ltr" -> align "LTR"
+                   _          -> contents'
   if beamer && "notes" `elem` classes  -- speaker notes
      then return $ "\\note" <> braces contents
      else return (linkAnchor $$ contents)
@@ -727,10 +735,12 @@ isQuoted _ = False
 -- | Convert inline element to LaTeX
 inlineToLaTeX :: Inline    -- ^ Inline to convert
               -> State WriterState Doc
-inlineToLaTeX (Span (id',classes,_) ils) = do
+inlineToLaTeX (Span (id',classes,kvs) ils) = do
   let noEmph = "csl-no-emph" `elem` classes
   let noStrong = "csl-no-strong" `elem` classes
   let noSmallCaps = "csl-no-smallcaps" `elem` classes
+  let rtl = ("dir","rtl") `elem` kvs
+  let ltr = ("dir","ltr") `elem` kvs
   ref <- toLabel id'
   let linkAnchor = if null id'
                       then empty
@@ -740,7 +750,9 @@ inlineToLaTeX (Span (id',classes,_) ils) = do
     ((if noEmph then inCmd "textup" else id) .
      (if noStrong then inCmd "textnormal" else id) .
      (if noSmallCaps then inCmd "textnormal" else id) .
-     (if not (noEmph || noStrong || noSmallCaps)
+     (if rtl then inCmd "RL" else id) .
+     (if ltr then inCmd "LR" else id) .
+     (if not (noEmph || noStrong || noSmallCaps || rtl || ltr)
          then braces
          else id)) `fmap` inlineListToLaTeX ils
 inlineToLaTeX (Emph lst) =