Org reader: support special strings export option

Parsing of special strings (like '...' as ellipsis or '--' as en dash)
can be toggled using the `-` option.

4 files changed, 36 insertions, 12 deletions

diff --git a/src/Text/Pandoc/Readers/Org/Blocks.hs b/src/Text/Pandoc/Readers/Org/Blocks.hs
index 896f1da1a..75e564f2f 100644
--- a/src/Text/Pandoc/Readers/Org/Blocks.hs
+++ b/src/Text/Pandoc/Readers/Org/Blocks.hs
@@ -502,7 +502,7 @@ exportSetting = choice
   [ booleanSetting "^" setExportSubSuperscripts
   , booleanSetting "'" setExportSmartQuotes
   , booleanSetting "*" setExportEmphasizedText
-  , ignoredSetting "-"
+  , booleanSetting "-" setExportSpecialStrings
   , ignoredSetting ":"
   , ignoredSetting "<"
   , ignoredSetting "\\n"
diff --git a/src/Text/Pandoc/Readers/Org/Inlines.hs b/src/Text/Pandoc/Readers/Org/Inlines.hs
index fdefd2a3f..001aeb569 100644
--- a/src/Text/Pandoc/Readers/Org/Inlines.hs
+++ b/src/Text/Pandoc/Readers/Org/Inlines.hs
@@ -728,8 +728,12 @@ smart = do
   doubleQuoted <|> singleQuoted <|>
     choice (map (return <$>) [orgApostrophe, orgDash, orgEllipses])
   where
-    orgDash = dash <* updatePositions '-'
-    orgEllipses = ellipses <* updatePositions '.'
+    orgDash = do
+      guard =<< getExportSetting exportSpecialStrings
+      dash <* updatePositions '-'
+    orgEllipses = do
+      guard =<< getExportSetting exportSpecialStrings
+      ellipses <* updatePositions '.'
     orgApostrophe =
           (char '\'' <|> char '\8217') <* updateLastPreCharPos
                                        <* updateLastForbiddenCharPos
@@ -737,7 +741,7 @@ smart = do
 
 singleQuoted :: OrgParser (F Inlines)
 singleQuoted = try $ do
-  guard . exportSmartQuotes . orgStateExportSettings =<< getState
+  guard =<< getExportSetting exportSmartQuotes
   singleQuoteStart
   updatePositions '\''
   withQuoteContext InSingleQuote $
@@ -749,7 +753,7 @@ singleQuoted = try $ do
 -- in the same paragraph.
 doubleQuoted :: OrgParser (F Inlines)
 doubleQuoted = try $ do
-  guard . exportSmartQuotes . orgStateExportSettings =<< getState
+  guard =<< getExportSetting exportSmartQuotes
   doubleQuoteStart
   updatePositions '"'
   contents <- mconcat <$> many (try $ notFollowedBy doubleQuoteEnd >> inline)
diff --git a/src/Text/Pandoc/Readers/Org/ParserState.hs b/src/Text/Pandoc/Readers/Org/ParserState.hs
index 2fa5d920b..0c58183f9 100644
--- a/src/Text/Pandoc/Readers/Org/ParserState.hs
+++ b/src/Text/Pandoc/Readers/Org/ParserState.hs
@@ -45,6 +45,7 @@ module Text.Pandoc.Readers.Org.ParserState
   , setExportDrawers
   , setExportEmphasizedText
   , setExportSmartQuotes
+  , setExportSpecialStrings
   , setExportSubSuperscripts
   , modifyExportSettings
   , optionsToParserState
@@ -84,8 +85,9 @@ data ExportSettings = ExportSettings
   -- ^ Specify drawer names which should be exported.  @Left@ names are
   -- explicitly excluded from the resulting output while @Right@ means that
   -- only the listed drawer names should be included.
-  , exportEmphasizedText  :: Bool -- ^ Parse emphasized text.
-  , exportSmartQuotes     :: Bool -- ^ Parse quotes, ellipses, apostrophs smartly
+  , exportEmphasizedText  :: Bool -- ^ Parse emphasized text
+  , exportSmartQuotes     :: Bool -- ^ Parse quotes smartly
+  , exportSpecialStrings  :: Bool -- ^ Parse ellipses and dashes smartly
   , exportSubSuperscripts :: Bool -- ^ TeX-like syntax for sub- and superscripts
   }
 
@@ -160,6 +162,7 @@ defaultExportSettings = ExportSettings
   { exportDrawers = Left ["LOGBOOK"]
   , exportEmphasizedText = True
   , exportSmartQuotes = True
+  , exportSpecialStrings = True
   , exportSubSuperscripts = True
   }
 
@@ -182,15 +185,20 @@ setExportDrawers val es = es { exportDrawers = val }
 setExportEmphasizedText :: ExportSettingSetter Bool
 setExportEmphasizedText val es = es { exportEmphasizedText = val }
 
+-- | Set export options for parsing of smart quotes.
+setExportSmartQuotes :: ExportSettingSetter Bool
+setExportSmartQuotes val es = es { exportSmartQuotes = val }
+
+-- | Set export options for parsing of special strings (like em/en dashes or
+-- ellipses).
+setExportSpecialStrings :: ExportSettingSetter Bool
+setExportSpecialStrings val es = es { exportSpecialStrings = val }
+
 -- | Set export options for sub/superscript parsing.  The short syntax will
 -- not be parsed if this is set set to @False@.
 setExportSubSuperscripts :: ExportSettingSetter Bool
 setExportSubSuperscripts val es = es { exportSubSuperscripts = val }
 
--- | Set export options for parsing of smart quotes.
-setExportSmartQuotes :: ExportSettingSetter Bool
-setExportSmartQuotes val es = es { exportSmartQuotes = val }
-
 -- | Modify a parser state
 modifyExportSettings :: ExportSettingSetter a -> a -> OrgParserState -> OrgParserState
 modifyExportSettings setter val state =
diff --git a/src/Text/Pandoc/Readers/Org/Parsing.hs b/src/Text/Pandoc/Readers/Org/Parsing.hs
index 0b6b876d8..8cf0c696c 100644
--- a/src/Text/Pandoc/Readers/Org/Parsing.hs
+++ b/src/Text/Pandoc/Readers/Org/Parsing.hs
@@ -37,6 +37,7 @@ module Text.Pandoc.Readers.Org.Parsing
   , skipSpaces1
   , inList
   , withContext
+  , getExportSetting
   , updateLastForbiddenCharPos
   , updateLastPreCharPos
   , orgArgKey
@@ -174,9 +175,13 @@ withContext context parser = do
   return result
 
 --
--- Parser state update functions
+-- Parser state functions
 --
 
+-- | Get an export setting.
+getExportSetting :: (ExportSettings -> a) -> OrgParser a
+getExportSetting s = s . orgStateExportSettings <$> getState
+
 -- | Set the current position as the last position at which a forbidden char
 -- was found (i.e. a character which is not allowed at the inner border of
 -- markup).
@@ -190,13 +195,20 @@ updateLastPreCharPos :: OrgParser ()
 updateLastPreCharPos = getPosition >>= \p ->
   updateState $ \s -> s{ orgStateLastPreCharPos = Just p}
 
+--
+-- Org key-value parsing
+--
+
+-- | Read the key of a plist style key-value list.
 orgArgKey :: OrgParser String
 orgArgKey = try $
   skipSpaces *> char ':'
              *> many1 orgArgWordChar
 
+-- | Read the value of a plist style key-value list.
 orgArgWord :: OrgParser String
 orgArgWord = many1 orgArgWordChar
 
+-- | Chars treated as part of a word in plists.
 orgArgWordChar :: OrgParser Char
 orgArgWordChar = alphaNum <|> oneOf "-_"