cf306f3 · typeable/pandoc

git.iko.soy/typeable/pandoc

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

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

commitcf306f34e5e23a1416b598766ca73d63d7367283 authorJohn MacFarlane <jgm@berkeley.edu> dateTue 21 Mar 2017, 15:41 (10 years ago) parent430e2db

3 files changed+34−27

compare with master

unifiedsplit

src/Text/Pandoc/Writers/Markdown.hs+31−24 test/writer.opml+1−1 test/writer.plain+2−2

src/Text/Pandoc/Writers/Markdown.hs+31−24

@@ -915,8 +915,11 @@ inlineToMarkdown opts (Superscript lst) =

915915 then "^" <> contents <> "^"

916916 else if isEnabled Ext_raw_html opts

917917 then "<sup>" <> contents <> "</sup>"

918− else text $ map toSuperscript

919− $ render Nothing contents

918+ else

919+ let rendered = render Nothing contents

920+ in case mapM toSuperscript rendered of

921+ Just r -> text r

922+ Nothing -> text $ "^(" ++ rendered ++ ")"

920923inlineToMarkdown opts (Subscript lst) =

921924 local (\env -> env {envEscapeSpaces = True}) $ do

922925 contents <- inlineListToMarkdown opts lst

@@ -924,8 +927,11 @@ inlineToMarkdown opts (Subscript lst) =

924927 then "~" <> contents <> "~"

925928 else if isEnabled Ext_raw_html opts

926929 then "<sub>" <> contents <> "</sub>"

927− else text $ map toSubscript

928− $ render Nothing contents

930+ else

931+ let rendered = render Nothing contents

932+ in case mapM toSubscript rendered of

933+ Just r -> text r

934+ Nothing -> text $ "_(" ++ rendered ++ ")"

929935inlineToMarkdown opts (SmallCaps lst) = do

930936 plain <- asks envPlain

931937 if not plain &&

@@ -1120,28 +1126,29 @@ makeMathPlainer = walk go

11201126 go (Emph xs) = Span nullAttr xs

11211127 go x = x

11221128

1123−toSuperscript :: Char -> Char

1124−toSuperscript '1' = '\x00B9'

1125−toSuperscript '2' = '\x00B2'

1126−toSuperscript '3' = '\x00B3'

1127−toSuperscript '+' = '\x207A'

1128−toSuperscript '-' = '\x207B'

1129−toSuperscript '=' = '\x207C'

1130−toSuperscript '(' = '\x207D'

1131−toSuperscript ')' = '\x207E'

1129+toSuperscript :: Char -> Maybe Char

1130+toSuperscript '1' = Just '\x00B9'

1131+toSuperscript '2' = Just '\x00B2'

1132+toSuperscript '3' = Just '\x00B3'

1133+toSuperscript '+' = Just '\x207A'

1134+toSuperscript '-' = Just '\x207B'

1135+toSuperscript '=' = Just '\x207C'

1136+toSuperscript '(' = Just '\x207D'

1137+toSuperscript ')' = Just '\x207E'

11321138toSuperscript c

11331139 | c >= '0' && c <= '9' =

1134− chr (0x2070 + (ord c - 48))

1135− | otherwise = c

1140+ Just $ chr (0x2070 + (ord c - 48))

1141+ | isSpace c = Just c

1142+ | otherwise = Nothing

11361143

1137−toSubscript :: Char -> Char

1138−toSubscript '+' = '\x208A'

1139−toSubscript '-' = '\x208B'

1140−toSubscript '=' = '\x208C'

1141−toSubscript '(' = '\x208D'

1142−toSubscript ')' = '\x208E'

1144+toSubscript :: Char -> Maybe Char

1145+toSubscript '+' = Just '\x208A'

1146+toSubscript '-' = Just '\x208B'

1147+toSubscript '=' = Just '\x208C'

1148+toSubscript '(' = Just '\x208D'

1149+toSubscript ')' = Just '\x208E'

11431150toSubscript c

11441151 | c >= '0' && c <= '9' =

1145− chr (0x2080 + (ord c - 48))

1146− | otherwise = c

1147−

1152+ Just $ chr (0x2080 + (ord c - 48))

1153+ | isSpace c = Just c

1154+ | otherwise = Nothing

test/writer.opml+1−1

@@ -46,7 +46,7 @@

4646</outline>

4747<outline text="HTML Blocks" _note="Simple block on one line:

foo

And nested without indentation:

foo

bar

Interpreted markdown in a table:

This is *emphasized*
And this is **strong**
Here’s a simple block:

foo

This should be a code block, though:

 <div>
 foo
 </div>

As should this:

 <div>foo</div>

Now, nested:

foo

This should just be an HTML comment:

Multiline:

Code block:

 

Just plain comment, with trailing spaces on the line:

Code:

 <hr />

Hr’s:

------------------------------------------------------------------------">

4848</outline>

49−<outline text="Inline Markup" _note="This is *emphasized*, and so *is this*.

This is **strong**, and so **is this**.

An *[emphasized link](/url)*.

***This is strong and em.***

So is ***this*** word.

***This is strong and em.***

So is ***this*** word.

This is code: `>`, `$`, `\`, `\$`, `<html>`.

This is *strikeout*.

Superscripts: abcd a*hello* ahello there.

Subscripts: H₂O, H₂₃O, Hmany of themO.

These should not be superscripts or subscripts, because of the unescaped
spaces: a^b c^d, a~b c~d.

------------------------------------------------------------------------">

49+<outline text="Inline Markup" _note="This is *emphasized*, and so *is this*.

This is **strong**, and so **is this**.

An *[emphasized link](/url)*.

***This is strong and em.***

So is ***this*** word.

***This is strong and em.***

So is ***this*** word.

This is code: `>`, `$`, `\`, `\$`, `<html>`.

This is *strikeout*.

Superscripts: a^(bc)d a^(*hello*) a^(hello there).

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.

------------------------------------------------------------------------">

5050</outline>

5151<outline text="Smart quotes, ellipses, dashes" _note="“Hello,” said the spider. “‘Shelob’ is my name.”

‘A’, ‘B’, and ‘C’ are letters.

‘Oak,’ ‘elm,’ and ‘beech’ are names of trees. So is ‘pine.’

‘He said, “I want to go.”’ Were you alive in the 70’s?

Here is some quoted ‘`code`’ and a “[quoted
link](http://example.com/?foo=1&bar=2)”.

Some dashes: one—two — three—four — five.

Dashes between numbers: 5–7, 255–66, 1987–1999.

Ellipses…and…and….

------------------------------------------------------------------------">

5252</outline>

test/writer.plain+2−2

@@ -448,9 +448,9 @@ This is code: >, $, \, \$, <html>.

448448

449449~~This is _strikeout_.~~

450450

451−Superscripts: abcd a_hello_ ahello there.

451+Superscripts: a^(bc)d a^(_hello_) a^(hello there).

452452

453−Subscripts: H₂O, H₂₃O, Hmany of themO.

453+Subscripts: H₂O, H₂₃O, H_(many of them)O.

454454

455455These should not be superscripts or subscripts, because of the unescaped

456456spaces: a^b c^d, a~b c~d.