Plain writer: use _(..) or ^(..) for super/subscript...

...unless unicode super/subscripted characters are available.

3 files changed, 34 insertions, 27 deletions

diff --git a/src/Text/Pandoc/Writers/Markdown.hs b/src/Text/Pandoc/Writers/Markdown.hs
index c1a02e609..88dd53808 100644
--- a/src/Text/Pandoc/Writers/Markdown.hs
+++ b/src/Text/Pandoc/Writers/Markdown.hs
@@ -915,8 +915,11 @@ inlineToMarkdown opts (Superscript lst) =
                 then "^" <> contents <> "^"
                 else if isEnabled Ext_raw_html opts
                          then "<sup>" <> contents <> "</sup>"
-                         else text $ map toSuperscript
-                                   $ render Nothing contents
+                         else
+                           let rendered = render Nothing contents
+                           in  case mapM toSuperscript rendered of
+                                    Just r -> text r
+                                    Nothing -> text $ "^(" ++ rendered ++ ")"
 inlineToMarkdown opts (Subscript lst) =
   local (\env -> env {envEscapeSpaces = True}) $ do
     contents <- inlineListToMarkdown opts lst
@@ -924,8 +927,11 @@ inlineToMarkdown opts (Subscript lst) =
                 then "~" <> contents <> "~"
                 else if isEnabled Ext_raw_html opts
                          then "<sub>" <> contents <> "</sub>"
-                         else text $ map toSubscript
-                                   $ render Nothing contents
+                         else
+                           let rendered = render Nothing contents
+                           in  case mapM toSubscript rendered of
+                                    Just r -> text r
+                                    Nothing -> text $ "_(" ++ rendered ++ ")"
 inlineToMarkdown opts (SmallCaps lst) = do
   plain <- asks envPlain
   if not plain &&
@@ -1120,28 +1126,29 @@ makeMathPlainer = walk go
   go (Emph xs) = Span nullAttr xs
   go x         = x
 
-toSuperscript :: Char -> Char
-toSuperscript '1' = '\x00B9'
-toSuperscript '2' = '\x00B2'
-toSuperscript '3' = '\x00B3'
-toSuperscript '+' = '\x207A'
-toSuperscript '-' = '\x207B'
-toSuperscript '=' = '\x207C'
-toSuperscript '(' = '\x207D'
-toSuperscript ')' = '\x207E'
+toSuperscript :: Char -> Maybe Char
+toSuperscript '1' = Just '\x00B9'
+toSuperscript '2' = Just '\x00B2'
+toSuperscript '3' = Just '\x00B3'
+toSuperscript '+' = Just '\x207A'
+toSuperscript '-' = Just '\x207B'
+toSuperscript '=' = Just '\x207C'
+toSuperscript '(' = Just '\x207D'
+toSuperscript ')' = Just '\x207E'
 toSuperscript c
   | c >= '0' && c <= '9' =
-                 chr (0x2070 + (ord c - 48))
-  | otherwise = c
+                 Just $ chr (0x2070 + (ord c - 48))
+  | isSpace c = Just c
+  | otherwise = Nothing
 
-toSubscript :: Char -> Char
-toSubscript '+' = '\x208A'
-toSubscript '-' = '\x208B'
-toSubscript '=' = '\x208C'
-toSubscript '(' = '\x208D'
-toSubscript ')' = '\x208E'
+toSubscript :: Char -> Maybe Char
+toSubscript '+' = Just '\x208A'
+toSubscript '-' = Just '\x208B'
+toSubscript '=' = Just '\x208C'
+toSubscript '(' = Just '\x208D'
+toSubscript ')' = Just '\x208E'
 toSubscript c
   | c >= '0' && c <= '9' =
-                 chr (0x2080 + (ord c - 48))
-  | otherwise = c
-
+                 Just $ chr (0x2080 + (ord c - 48))
+  | isSpace c = Just c
+  | otherwise = Nothing
diff --git a/test/writer.opml b/test/writer.opml
index 0b621823e..51b0cb2d5 100644
--- a/test/writer.opml
+++ b/test/writer.opml
@@ -46,7 +46,7 @@
 </outline>
 <outline text="HTML Blocks" _note="Simple block on one line:&#10;&#10;foo&#10;&#10;And nested without indentation:&#10;&#10;foo&#10;&#10;bar&#10;&#10;Interpreted markdown in a table:&#10;&#10;This is *emphasized*&#10;And this is **strong**&#10;Here’s a simple block:&#10;&#10;foo&#10;&#10;This should be a code block, though:&#10;&#10;    &lt;div&gt;&#10;        foo&#10;    &lt;/div&gt;&#10;&#10;As should this:&#10;&#10;    &lt;div&gt;foo&lt;/div&gt;&#10;&#10;Now, nested:&#10;&#10;foo&#10;&#10;This should just be an HTML comment:&#10;&#10;Multiline:&#10;&#10;Code block:&#10;&#10;    &lt;!-- Comment --&gt;&#10;&#10;Just plain comment, with trailing spaces on the line:&#10;&#10;Code:&#10;&#10;    &lt;hr /&gt;&#10;&#10;Hr’s:&#10;&#10;------------------------------------------------------------------------">
 </outline>
-<outline text="Inline Markup" _note="This is *emphasized*, and so *is this*.&#10;&#10;This is **strong**, and so **is this**.&#10;&#10;An *[emphasized link](/url)*.&#10;&#10;***This is strong and em.***&#10;&#10;So is ***this*** word.&#10;&#10;***This is strong and em.***&#10;&#10;So is ***this*** word.&#10;&#10;This is code: `&gt;`, `$`, `\`, `\$`, `&lt;html&gt;`.&#10;&#10;This is *strikeout*.&#10;&#10;Superscripts: abcd a*hello* ahello there.&#10;&#10;Subscripts: H₂O, H₂₃O, Hmany of themO.&#10;&#10;These should not be superscripts or subscripts, because of the unescaped&#10;spaces: a^b c^d, a~b c~d.&#10;&#10;------------------------------------------------------------------------">
+<outline text="Inline Markup" _note="This is *emphasized*, and so *is this*.&#10;&#10;This is **strong**, and so **is this**.&#10;&#10;An *[emphasized link](/url)*.&#10;&#10;***This is strong and em.***&#10;&#10;So is ***this*** word.&#10;&#10;***This is strong and em.***&#10;&#10;So is ***this*** word.&#10;&#10;This is code: `&gt;`, `$`, `\`, `\$`, `&lt;html&gt;`.&#10;&#10;This is *strikeout*.&#10;&#10;Superscripts: a^(bc)d a^(*hello*) a^(hello there).&#10;&#10;Subscripts: H₂O, H₂₃O, H_(many of them)O.&#10;&#10;These should not be superscripts or subscripts, because of the unescaped&#10;spaces: a^b c^d, a~b c~d.&#10;&#10;------------------------------------------------------------------------">
 </outline>
 <outline text="Smart quotes, ellipses, dashes" _note="“Hello,” said the spider. “‘Shelob’ is my name.”&#10;&#10;‘A’, ‘B’, and ‘C’ are letters.&#10;&#10;‘Oak,’ ‘elm,’ and ‘beech’ are names of trees. So is ‘pine.’&#10;&#10;‘He said, “I want to go.”’ Were you alive in the 70’s?&#10;&#10;Here is some quoted ‘`code`’ and a “[quoted&#10;link](http://example.com/?foo=1&amp;bar=2)”.&#10;&#10;Some dashes: one—two — three—four — five.&#10;&#10;Dashes between numbers: 5–7, 255–66, 1987–1999.&#10;&#10;Ellipses…and…and….&#10;&#10;------------------------------------------------------------------------">
 </outline>
diff --git a/test/writer.plain b/test/writer.plain
index f34af9100..175efb608 100644
--- a/test/writer.plain
+++ b/test/writer.plain
@@ -448,9 +448,9 @@ This is code: >, $, \, \$, <html>.
 
 ~~This is _strikeout_.~~
 
-Superscripts: abcd a_hello_ ahello there.
+Superscripts: a^(bc)d a^(_hello_) a^(hello there).
 
-Subscripts: H₂O, H₂₃O, Hmany of themO.
+Subscripts: H₂O, H₂₃O, H_(many of them)O.
 
 These should not be superscripts or subscripts, because of the unescaped
 spaces: a^b c^d, a~b c~d.