f567929 · typeable/pandoc

git.iko.soy/typeable/pandoc

Org reader: support special strings export option

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

commitf56792927fbcebfd25593c488476fee69d078b48 authorAlbert Krewinkel <albert@zeitkraut.de> dateFri 3 Jun 2016, 11:10 (10 years ago) parentd4de845

4 files changed+36−12

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Org/Blocks.hs+1−1 src/Text/Pandoc/Readers/Org/Inlines.hs+8−4 src/Text/Pandoc/Readers/Org/ParserState.hs+14−6 src/Text/Pandoc/Readers/Org/Parsing.hs+13−1

src/Text/Pandoc/Readers/Org/Blocks.hs+1−1

@@ -502,7 +502,7 @@ exportSetting = choice

502502 [ booleanSetting "^" setExportSubSuperscripts

503503 , booleanSetting "'" setExportSmartQuotes

504504 , booleanSetting "*" setExportEmphasizedText

505− , ignoredSetting "-"

505+ , booleanSetting "-" setExportSpecialStrings

506506 , ignoredSetting ":"

507507 , ignoredSetting "<"

508508 , ignoredSetting "\\n"

src/Text/Pandoc/Readers/Org/Inlines.hs+8−4

@@ -728,8 +728,12 @@ smart = do

728728 doubleQuoted <|> singleQuoted <|>

729729 choice (map (return <$>) [orgApostrophe, orgDash, orgEllipses])

730730 where

731− orgDash = dash <* updatePositions '-'

732− orgEllipses = ellipses <* updatePositions '.'

731+ orgDash = do

732+ guard =<< getExportSetting exportSpecialStrings

733+ dash <* updatePositions '-'

734+ orgEllipses = do

735+ guard =<< getExportSetting exportSpecialStrings

736+ ellipses <* updatePositions '.'

733737 orgApostrophe =

734738 (char '\'' <|> char '\8217') <* updateLastPreCharPos

735739 <* updateLastForbiddenCharPos

@@ -737,7 +741,7 @@ smart = do

737741

738742singleQuoted :: OrgParser (F Inlines)

739743singleQuoted = try $ do

740− guard . exportSmartQuotes . orgStateExportSettings =<< getState

744+ guard =<< getExportSetting exportSmartQuotes

741745 singleQuoteStart

742746 updatePositions '\''

743747 withQuoteContext InSingleQuote $

@@ -749,7 +753,7 @@ singleQuoted = try $ do

749753-- in the same paragraph.

750754doubleQuoted :: OrgParser (F Inlines)

751755doubleQuoted = try $ do

752− guard . exportSmartQuotes . orgStateExportSettings =<< getState

756+ guard =<< getExportSetting exportSmartQuotes

753757 doubleQuoteStart

754758 updatePositions '"'

755759 contents <- mconcat <$> many (try $ notFollowedBy doubleQuoteEnd >> inline)

src/Text/Pandoc/Readers/Org/ParserState.hs+14−6

@@ -45,6 +45,7 @@ module Text.Pandoc.Readers.Org.ParserState

4545 , setExportDrawers

4646 , setExportEmphasizedText

4747 , setExportSmartQuotes

48+ , setExportSpecialStrings

4849 , setExportSubSuperscripts

4950 , modifyExportSettings

5051 , optionsToParserState

@@ -84,8 +85,9 @@ data ExportSettings = ExportSettings

8485 -- ^ Specify drawer names which should be exported. @Left@ names are

8586 -- explicitly excluded from the resulting output while @Right@ means that

8687 -- only the listed drawer names should be included.

87− , exportEmphasizedText :: Bool -- ^ Parse emphasized text.

88− , exportSmartQuotes :: Bool -- ^ Parse quotes, ellipses, apostrophs smartly

88+ , exportEmphasizedText :: Bool -- ^ Parse emphasized text

89+ , exportSmartQuotes :: Bool -- ^ Parse quotes smartly

90+ , exportSpecialStrings :: Bool -- ^ Parse ellipses and dashes smartly

8991 , exportSubSuperscripts :: Bool -- ^ TeX-like syntax for sub- and superscripts

9092 }

9193

@@ -160,6 +162,7 @@ defaultExportSettings = ExportSettings

160162 { exportDrawers = Left ["LOGBOOK"]

161163 , exportEmphasizedText = True

162164 , exportSmartQuotes = True

165+ , exportSpecialStrings = True

163166 , exportSubSuperscripts = True

164167 }

165168

@@ -182,15 +185,20 @@ setExportDrawers val es = es { exportDrawers = val }

182185setExportEmphasizedText :: ExportSettingSetter Bool

183186setExportEmphasizedText val es = es { exportEmphasizedText = val }

184187

188+-- | Set export options for parsing of smart quotes.

189+setExportSmartQuotes :: ExportSettingSetter Bool

190+setExportSmartQuotes val es = es { exportSmartQuotes = val }

191+

192+-- | Set export options for parsing of special strings (like em/en dashes or

193+-- ellipses).

194+setExportSpecialStrings :: ExportSettingSetter Bool

195+setExportSpecialStrings val es = es { exportSpecialStrings = val }

196+

185197-- | Set export options for sub/superscript parsing. The short syntax will

186198-- not be parsed if this is set set to @False@.

187199setExportSubSuperscripts :: ExportSettingSetter Bool

188200setExportSubSuperscripts val es = es { exportSubSuperscripts = val }

189201

190−-- | Set export options for parsing of smart quotes.

191−setExportSmartQuotes :: ExportSettingSetter Bool

192−setExportSmartQuotes val es = es { exportSmartQuotes = val }

193−

194202-- | Modify a parser state

195203modifyExportSettings :: ExportSettingSetter a -> a -> OrgParserState -> OrgParserState

196204modifyExportSettings setter val state =

src/Text/Pandoc/Readers/Org/Parsing.hs+13−1

@@ -37,6 +37,7 @@ module Text.Pandoc.Readers.Org.Parsing

3737 , skipSpaces1

3838 , inList

3939 , withContext

40+ , getExportSetting

4041 , updateLastForbiddenCharPos

4142 , updateLastPreCharPos

4243 , orgArgKey

@@ -174,9 +175,13 @@ withContext context parser = do

174175 return result

175176

176177--

177−-- Parser state update functions

178+-- Parser state functions

178179--

179180

181+-- | Get an export setting.

182+getExportSetting :: (ExportSettings -> a) -> OrgParser a

183+getExportSetting s = s . orgStateExportSettings <$> getState

184+

180185-- | Set the current position as the last position at which a forbidden char

181186-- was found (i.e. a character which is not allowed at the inner border of

182187-- markup).

@@ -190,13 +195,20 @@ updateLastPreCharPos :: OrgParser ()

190195updateLastPreCharPos = getPosition >>= \p ->

191196 updateState $ \s -> s{ orgStateLastPreCharPos = Just p}

192197

198+--

199+-- Org key-value parsing

200+--

201+

202+-- | Read the key of a plist style key-value list.

193203orgArgKey :: OrgParser String

194204orgArgKey = try $

195205 skipSpaces *> char ':'

196206 *> many1 orgArgWordChar

197207

208+-- | Read the value of a plist style key-value list.

198209orgArgWord :: OrgParser String

199210orgArgWord = many1 orgArgWordChar

200211

212+-- | Chars treated as part of a word in plists.

201213orgArgWordChar :: OrgParser Char

202214orgArgWordChar = alphaNum <|> oneOf "-_"