06bc037 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

LaTeX reader: Improve handling of raw LaTeX (for markdown etc.).

06bc037John MacFarlane9 years ago

src/Text/Pandoc/Readers/LaTeX.hs+32−31

compare with master

unifiedsplit

@@ -242,21 +242,30 @@ withVerbatimMode parser = do

242 return result

243

244rawLaTeXParser :: (PandocMonad m, HasMacros s, HasReaderOptions s)

245− => LP m a -> ParserT String s m (a, String)

245+ => LP m a -> LP m a -> ParserT String s m (a, String)

246−rawLaTeXParser parser = do

246+rawLaTeXParser parser valParser = do

247 inp <- getInput

248 let toks = tokenize "source" $ T.pack inp

249 pstate <- getState

250− let lstate = def{ sOptions = extractReaderOptions pstate

250+ let lstate = def{ sOptions = extractReaderOptions pstate }

251− , sMacros = extractMacros pstate }

251+ let lstate' = lstate { sMacros = extractMacros pstate }

252− let rawparser = (,) <$> withRaw parser <*> getState

252+ let rawparser = (,) <$> withRaw valParser <*> getState

253− res <- lift $ runParserT rawparser lstate "chunk" toks

253+ res' <- lift $ runParserT (snd <$> withRaw parser) lstate "chunk" toks

254− case res of

254+ case res' of

255 Left _ -> mzero

256+ Right toks' -> do

257+ res <- lift $ runParserT (do doMacros 0

258+ -- retokenize, applying macros

259+ ts <- many (satisfyTok (const True))

260+ setInput ts

261+ rawparser)

262+ lstate' "chunk" toks'

263+ case res of

264+ Left _ -> mzero

256− Right ((val, raw), st) -> do

265+ Right ((val, raw), st) -> do

257− updateState (updateMacros (sMacros st <>))

266+ updateState (updateMacros (sMacros st <>))

258− rawstring <- takeP (T.length (untokenize raw))

267+ _ <- takeP (T.length (untokenize toks'))

259− return (val, rawstring)

268+ return (val, T.unpack (untokenize raw))

260

269

261applyMacros :: (PandocMonad m, HasMacros s, HasReaderOptions s)

270applyMacros :: (PandocMonad m, HasMacros s, HasReaderOptions s)

262 => String -> ParserT String s m String

271 => String -> ParserT String s m String

@@ -277,19 +286,18 @@ rawLaTeXBlock = do

277 lookAhead (try (char '\\' >> letter))

286 lookAhead (try (char '\\' >> letter))

278 -- we don't want to apply newly defined latex macros to their own

287 -- we don't want to apply newly defined latex macros to their own

279 -- definitions:

288 -- definitions:

280− snd <$> rawLaTeXParser macroDef

289+ snd <$> rawLaTeXParser (environment <|> macroDef <|> blockCommand) block

281− <|> ((snd <$> rawLaTeXParser (environment <|> blockCommand)) >>= applyMacros)

282

290

283rawLaTeXInline :: (PandocMonad m, HasMacros s, HasReaderOptions s)

291rawLaTeXInline :: (PandocMonad m, HasMacros s, HasReaderOptions s)

284 => ParserT String s m String

292 => ParserT String s m String

285rawLaTeXInline = do

293rawLaTeXInline = do

286 lookAhead (try (char '\\' >> letter))

294 lookAhead (try (char '\\' >> letter))

287− rawLaTeXParser (inlineEnvironment <|> inlineCommand') >>= applyMacros . snd

295+ snd <$> rawLaTeXParser (inlineEnvironment <|> inlineCommand') inline

288

296

289inlineCommand :: PandocMonad m => ParserT String ParserState m Inlines

297inlineCommand :: PandocMonad m => ParserT String ParserState m Inlines

290inlineCommand = do

298inlineCommand = do

291 lookAhead (try (char '\\' >> letter))

299 lookAhead (try (char '\\' >> letter))

292− fst <$> rawLaTeXParser (inlineEnvironment <|> inlineCommand')

300+ fst <$> rawLaTeXParser (inlineEnvironment <|> inlineCommand') inline

293

301

294tokenize :: SourceName -> Text -> [Tok]

302tokenize :: SourceName -> Text -> [Tok]

295tokenize sourcename = totoks (initialPos sourcename)

303tokenize sourcename = totoks (initialPos sourcename)

@@ -1703,6 +1711,9 @@ treatAsBlock = Set.fromList

1703 , "clearpage"

1711 , "clearpage"

1704 , "pagebreak"

1712 , "pagebreak"

1705 , "titleformat"

1713 , "titleformat"

1714+ , "listoffigures"

1715+ , "listoftables"

1716+ , "write"

1706 ]

1717 ]

1707

1718

1708isInlineCommand :: Text -> Bool

1719isInlineCommand :: Text -> Bool

@@ -2165,19 +2176,6 @@ environments = M.fromList

2165 codeBlockWith attr <$> verbEnv "lstlisting")

2176 codeBlockWith attr <$> verbEnv "lstlisting")

2166 , ("minted", minted)

2177 , ("minted", minted)

2167 , ("obeylines", obeylines)

2178 , ("obeylines", obeylines)

2168− , ("displaymath", mathEnvWith para Nothing "displaymath")

2169− , ("equation", mathEnvWith para Nothing "equation")

2170− , ("equation*", mathEnvWith para Nothing "equation*")

2171− , ("gather", mathEnvWith para (Just "gathered") "gather")

2172− , ("gather*", mathEnvWith para (Just "gathered") "gather*")

2173− , ("multline", mathEnvWith para (Just "gathered") "multline")

2174− , ("multline*", mathEnvWith para (Just "gathered") "multline*")

2175− , ("eqnarray", mathEnvWith para (Just "aligned") "eqnarray")

2176− , ("eqnarray*", mathEnvWith para (Just "aligned") "eqnarray*")

2177− , ("align", mathEnvWith para (Just "aligned") "align")

2178− , ("align*", mathEnvWith para (Just "aligned") "align*")

2179− , ("alignat", mathEnvWith para (Just "aligned") "alignat")

2180− , ("alignat*", mathEnvWith para (Just "aligned") "alignat*")

2181 , ("tikzpicture", rawVerbEnv "tikzpicture")

2179 , ("tikzpicture", rawVerbEnv "tikzpicture")

2182 -- etoolbox

2180 -- etoolbox

2183 , ("ifstrequal", ifstrequal)

2181 , ("ifstrequal", ifstrequal)

@@ -2188,11 +2186,14 @@ environments = M.fromList

2188 ]

2186 ]

2189

2187

2190environment :: PandocMonad m => LP m Blocks

2188environment :: PandocMonad m => LP m Blocks

2191−environment = do

2189+environment = try $ do

2192 controlSeq "begin"

2190 controlSeq "begin"

2193 name <- untokenize <$> braced

2191 name <- untokenize <$> braced

2194− M.findWithDefault mzero name environments

2192+ M.findWithDefault mzero name environments <|>

2193+ if M.member name (inlineEnvironments

2194+ :: M.Map Text (LP PandocPure Inlines))

2195+ then mzero

2195− <|> rawEnv name

2196+ else rawEnv name

2196

2197

2197env :: PandocMonad m => Text -> LP m a -> LP m a

2198env :: PandocMonad m => Text -> LP m a -> LP m a

2198env name p = p <* end_ name

2199env name p = p <* end_ name