9e6e9a7 · typeable/pandoc

git.iko.soy/typeable/pandoc

Org reader: honor tex export option

The `tex` export option can be set with `#+OPTION: tex:nil` and allows three settings: - `t` causes LaTeX fragments to be parsed as TeX or added as raw TeX, - `nil` removes all LaTeX fragments from the document, and - `verbatim` treats LaTeX as text. The default is `t`. Closes: #4070

commit9e6e9a72218e8c408e151bf8b169f44a8c55eb40 authorAlbert Krewinkel <albert@zeitkraut.de> dateThu 25 Jun 2020, 20:04 (6 years ago) parentf1c678a

5 files changed+148−30

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Org/Blocks.hs+18−14 src/Text/Pandoc/Readers/Org/ExportSettings.hs+30−6 src/Text/Pandoc/Readers/Org/Inlines.hs+17−10 src/Text/Pandoc/Readers/Org/ParserState.hs+10 test/Tests/Readers/Org/Directive.hs+73

src/Text/Pandoc/Readers/Org/Blocks.hs+18−14

@@ -34,7 +34,7 @@ import Text.Pandoc.Shared (compactify, compactifyDL, safeRead)

3434import Control.Monad (foldM, guard, mplus, mzero, void)

3535import Data.Char (isSpace)

3636import Data.Default (Default)

37−import Data.List (foldl')

37+import Data.List (foldl', intersperse)

3838import Data.Maybe (fromMaybe, isJust, isNothing)

3939import Data.Text (Text)

4040

@@ -739,22 +739,26 @@ rowToContent tbl row =

739739--

740740-- LaTeX fragments

741741--

742−latexFragment :: Monad m => OrgParser m (F Blocks)

742+latexFragment :: PandocMonad m => OrgParser m (F Blocks)

743743latexFragment = try $ do

744744 envName <- latexEnvStart

745− content <- mconcat <$> manyTill anyLineNewline (latexEnd envName)

746− returnF $ B.rawBlock "latex" (content `inLatexEnv` envName)

745+ texOpt <- getExportSetting exportWithLatex

746+ let envStart = "\\begin{" <> envName <> "}"

747+ let envEnd = "\\end{" <> envName <> "}"

748+ envLines <- do

749+ content <- manyTill anyLine (latexEnd envName)

750+ return $ envStart : content ++ [envEnd]

751+ returnF $ case texOpt of

752+ TeXExport -> B.rawBlock "latex" . T.unlines $ envLines

753+ TeXIgnore -> mempty

754+ TeXVerbatim -> B.para . mconcat . intersperse B.softbreak $

755+ map B.str envLines

747756 where

748− c `inLatexEnv` e = mconcat [ "\\begin{", e, "}\n"

749− , c

750− , "\\end{", e, "}\n"

751− ]

752−

753−latexEnd :: Monad m => Text -> OrgParser m ()

754−latexEnd envName = try $

755− () <$ skipSpaces

756− <* textStr ("\\end{" <> envName <> "}")

757− <* blankline

757+ latexEnd :: Monad m => Text -> OrgParser m ()

758+ latexEnd envName = try . void

759+ $ skipSpaces

760+ <* textStr ("\\end{" <> envName <> "}")

761+ <* blankline

758762

759763

760764--

src/Text/Pandoc/Readers/Org/ExportSettings.hs+30−6

@@ -20,7 +20,7 @@ import Text.Pandoc.Readers.Org.Parsing

2020import Control.Monad (mzero, void)

2121import Data.Char (toLower)

2222import Data.Maybe (listToMaybe)

23−import Data.Text (Text)

23+import Data.Text (Text, unpack)

2424

2525-- | Read and handle space separated org-mode export settings.

2626exportSettings :: PandocMonad m => OrgParser m ()

@@ -59,7 +59,7 @@ exportSetting = choice

5959 , ignoredSetting "stat"

6060 , booleanSetting "tags" (\val es -> es { exportWithTags = val })

6161 , ignoredSetting "tasks"

62− , ignoredSetting "tex"

62+ , texSetting "tex" (\val es -> es { exportWithLatex = val })

6363 , ignoredSetting "timestamp"

6464 , ignoredSetting "title"

6565 , ignoredSetting "toc"

@@ -68,6 +68,8 @@ exportSetting = choice

6868 , ignoreAndWarn

6969 ] <?> "export setting"

7070

71+-- | Generic handler for export settings. Takes a parser which converts

72+-- the plain option text into a data structure.

7173genericExportSetting :: Monad m

7274 => OrgParser m a

7375 -> Text

@@ -101,10 +103,8 @@ archivedTreeSetting :: Monad m

101103archivedTreeSetting =

102104 genericExportSetting $ archivedTreesHeadlineSetting <|> archivedTreesBoolean

103105 where

104− archivedTreesHeadlineSetting = try $ do

105− _ <- string "headline"

106− lookAhead (newline <|> spaceChar)

107− return ArchivedTreesHeadlineOnly

106+ archivedTreesHeadlineSetting =

107+ ArchivedTreesHeadlineOnly <$ optionString "headline"

108108

109109 archivedTreesBoolean = try $ do

110110 exportBool <- elispBoolean

@@ -143,6 +143,22 @@ complementableListSetting = genericExportSetting $ choice

143143 char '"'

144144 *> manyTillChar alphaNum (char '"')

145145

146+-- | Parses either @t@, @nil@, or @verbatim@ into a 'TeXExport' value.

147+texSetting :: Monad m

148+ => Text

149+ -> ExportSettingSetter TeXExport

150+ -> OrgParser m ()

151+texSetting = genericExportSetting $ texVerbatim <|> texBoolean

152+ where

153+ texVerbatim = TeXVerbatim <$ optionString "verbatim"

154+

155+ texBoolean = try $ do

156+ exportBool <- elispBoolean

157+ return $

158+ if exportBool

159+ then TeXExport

160+ else TeXIgnore

161+

146162-- | Read but ignore the export setting.

147163ignoredSetting :: Monad m => Text -> OrgParser m ()

148164ignoredSetting s = try (() <$ textStr s <* char ':' <* many1 nonspaceChar)

@@ -164,3 +180,11 @@ elispBoolean = try $ do

164180 "{}" -> False

165181 "()" -> False

166182 _ -> True

183+

184+-- | Try to parse a literal string as the option value. Returns the

185+-- string on success.

186+optionString :: Monad m => Text -> OrgParser m Text

187+optionString s = try $ do

188+ _ <- string (unpack s)

189+ lookAhead (newline <|> spaceChar)

190+ return s

src/Text/Pandoc/Readers/Org/Inlines.hs+17−10

@@ -1,3 +1,4 @@

1+{-# LANGUAGE LambdaCase #-}

12{-# LANGUAGE OverloadedStrings #-}

23{- |

34 Module : Text.Pandoc.Readers.Org.Inlines

@@ -788,15 +789,17 @@ simpleSubOrSuperText = try $ do

788789inlineLaTeX :: PandocMonad m => OrgParser m (F Inlines)

789790inlineLaTeX = try $ do

790791 cmd <- inlineLaTeXCommand

791− ils <- (lift . lift) $ parseAsInlineLaTeX cmd

792+ texOpt <- getExportSetting exportWithLatex

793+ ils <- parseAsInlineLaTeX cmd texOpt

792794 maybe mzero returnF $

793− parseAsMathMLSym cmd `mplus` parseAsMath cmd `mplus` ils

795+ parseAsMathMLSym cmd `mplus` parseAsMath cmd texOpt `mplus` ils

794796 where

795− parseAsMath :: Text -> Maybe Inlines

796− parseAsMath cs = B.fromList <$> texMathToPandoc cs

797−

798− parseAsInlineLaTeX :: PandocMonad m => Text -> m (Maybe Inlines)

799− parseAsInlineLaTeX cs = maybeRight <$> runParserT inlineCommand state "" cs

797+ parseAsInlineLaTeX :: PandocMonad m

798+ => Text -> TeXExport -> OrgParser m (Maybe Inlines)

799+ parseAsInlineLaTeX cs = \case

800+ TeXExport -> maybeRight <$> runParserT inlineCommand state "" cs

801+ TeXIgnore -> return (Just mempty)

802+ TeXVerbatim -> return (Just $ B.str cs)

800803

801804 parseAsMathMLSym :: Text -> Maybe Inlines

802805 parseAsMathMLSym cs = B.str <$> MathMLEntityMap.getUnicode (clean cs)

@@ -807,8 +810,12 @@ inlineLaTeX = try $ do

807810 state = def{ stateOptions = def{ readerExtensions =

808811 enableExtension Ext_raw_tex (readerExtensions def) } }

809812

810− texMathToPandoc :: Text -> Maybe [Inline]

811− texMathToPandoc cs = maybeRight (readTeX cs) >>= writePandoc DisplayInline

813+ parseAsMath :: Text -> TeXExport -> Maybe Inlines

814+ parseAsMath cs = \case

815+ TeXExport -> maybeRight (readTeX cs) >>=

816+ fmap B.fromList . writePandoc DisplayInline

817+ TeXIgnore -> Just mempty

818+ TeXVerbatim -> Just $ B.str cs

812819

813820maybeRight :: Either a b -> Maybe b

814821maybeRight = either (const Nothing) Just

@@ -820,7 +827,7 @@ inlineLaTeXCommand = try $ do

820827 parsed <- (lift . lift) $ runParserT rawLaTeXInline st "source" rest

821828 case parsed of

822829 Right cs -> do

823− -- drop any trailing whitespace, those are not be part of the command as

830+ -- drop any trailing whitespace, those are not part of the command as

824831 -- far as org mode is concerned.

825832 let cmdNoSpc = T.dropWhileEnd isSpace cs

826833 let len = T.length cmdNoSpc

src/Text/Pandoc/Readers/Org/ParserState.hs+10

@@ -35,6 +35,7 @@ module Text.Pandoc.Readers.Org.ParserState

3535 , returnF

3636 , ExportSettings (..)

3737 , ArchivedTreesOption (..)

38+ , TeXExport (..)

3839 , optionsToParserState

3940 ) where

4041

@@ -231,6 +232,13 @@ data ArchivedTreesOption =

231232 | ArchivedTreesNoExport -- ^ Exclude archived trees from exporting

232233 | ArchivedTreesHeadlineOnly -- ^ Export only the headline, discard the contents

233234

235+-- | Options for the handling of LaTeX environments and fragments.

236+-- Represents allowed values of Emacs variable @org-export-with-latex@.

237+data TeXExport

238+ = TeXExport -- ^ Include raw TeX in the output

239+ | TeXIgnore -- ^ Ignore raw TeX

240+ | TeXVerbatim -- ^ Keep everything in verbatim

241+

234242-- | Export settings <http://orgmode.org/manual/Export-settings.html>

235243-- These settings can be changed via OPTIONS statements.

236244data ExportSettings = ExportSettings

@@ -249,6 +257,7 @@ data ExportSettings = ExportSettings

249257 , exportWithAuthor :: Bool -- ^ Include author in final meta-data

250258 , exportWithCreator :: Bool -- ^ Include creator in final meta-data

251259 , exportWithEmail :: Bool -- ^ Include email in final meta-data

260+ , exportWithLatex :: TeXExport -- ^ Handling of raw TeX commands

252261 , exportWithPlanning :: Bool -- ^ Keep planning info after headlines

253262 , exportWithTags :: Bool -- ^ Keep tags as part of headlines

254263 , exportWithTodoKeywords :: Bool -- ^ Keep TODO keywords in headers

@@ -270,6 +279,7 @@ defaultExportSettings = ExportSettings

270279 , exportWithAuthor = True

271280 , exportWithCreator = True

272281 , exportWithEmail = True

282+ , exportWithLatex = TeXExport

273283 , exportWithPlanning = False

274284 , exportWithTags = True

275285 , exportWithTodoKeywords = True

test/Tests/Readers/Org/Directive.hs+73

@@ -162,6 +162,79 @@ tests =

162162 ] =?>

163163 headerWith ("headline", [], mempty) 1 "Headline"

164164

165+ , testGroup "LaTeX"

166+ [ testGroup "Include LaTeX fragments"

167+ [ "Inline command" =:

168+ T.unlines [ "#+OPTIONS: tex:t"

169+ , "Hello \\emph{Name}"

170+ ] =?>

171+ para ("Hello" <> space <> emph "Name")

172+

173+ , "Alpha" =:

174+ T.unlines [ "#+OPTIONS: tex:t"

175+ , "\\alpha"

176+ ] =?>

177+ para "α"

178+

179+ , "equation environment" =:

180+ T.unlines [ "#+OPTIONS: tex:t"

181+ , "\\begin{equation}"

182+ , "f(x) = x^2"

183+ , "\\end{equation}"

184+ ] =?>

185+ rawBlock "latex" (T.unlines [ "\\begin{equation}"

186+ , "f(x) = x^2"

187+ , "\\end{equation}"

188+ ])

189+ ]

190+

191+ , testGroup "Ignore LaTeX fragments"

192+ [ "Inline command" =:

193+ T.unlines [ "#+OPTIONS: tex:nil"

194+ , "Hello \\emph{Emphasised}"

195+ ] =?>

196+ para "Hello"

197+

198+ , "MathML symbol (alpha)" =:

199+ T.unlines [ "#+OPTIONS: tex:nil"

200+ , "\\alpha"

201+ ] =?>

202+ para "α"

203+

204+ , "equation environment" =:

205+ T.unlines [ "#+OPTIONS: tex:nil"

206+ , "\\begin{equation}"

207+ , "f(x) = x^2"

208+ , "\\end{equation}"

209+ ] =?>

210+ (mempty :: Blocks)

211+ ]

212+

213+ , testGroup "Verbatim LaTeX"

214+ [ "Inline command" =:

215+ T.unlines [ "#+OPTIONS: tex:verbatim"

216+ , "Hello \\emph{Emphasised}"

217+ ] =?>

218+ para "Hello \\emph{Emphasised}"

219+

220+ , "MathML symbol (alpha)" =:

221+ T.unlines [ "#+OPTIONS: tex:verbatim"

222+ , "\\alpha"

223+ ] =?>

224+ para "α"

225+

226+ , "equation environment" =:

227+ T.unlines [ "#+OPTIONS: tex:verbatim"

228+ , "\\begin{equation}"

229+ , "f(x) = x^2"

230+ , "\\end{equation}"

231+ ] =?>

232+ para (str "\\begin{equation}" <> softbreak <>

233+ str "f(x) = x^2" <> softbreak <>

234+ str "\\end{equation}")

235+ ]

236+ ]

237+

165238 , testGroup "planning information"

166239 [ "include planning info after headlines" =:

167240 T.unlines [ "#+OPTIONS: p:t"