git.iko.soy/typeable/pandoc
on master help sign in
LaTeX reader: tokenize before pulling tokens,

rather than after. This has some performance penalty but is more reliable. Closes #4408.

commit1db585689ad6de5c829851a1e7b3aa0192e4e808 authorJohn MacFarlane <jgm@berkeley.edu> dateMon 15 Oct 2018, 14:52 (8 years ago) parent7e9e24b
2 files changed+16−15
src/Text/Pandoc/Readers/LaTeX.hs+2−4 src/Text/Pandoc/Readers/LaTeX/Parsing.hs+14−11
src/Text/Pandoc/Readers/LaTeX.hs+2−4
@@ -1480,7 +1480,7 @@ authors = try $ do
14801480
14811481macroDef :: (Monoid a, PandocMonad m) => LP m a
14821482macroDef =
1483− mempty <$ ((commandDef <|> environmentDef) <* doMacros)
1483+ mempty <$ (commandDef <|> environmentDef)
14841484 where commandDef = do
14851485 (name, macro') <- newcommand <|> letmacro <|> defmacro
14861486 guardDisabled Ext_latex_macros <|>
@@ -1501,7 +1501,7 @@ letmacro :: PandocMonad m => LP m (Text, Macro)
15011501letmacro = do
15021502 controlSeq "let"
15031503 (name, contents) <- withVerbatimMode $ do
1504− Tok _ (CtrlSeq name) _ <- withVerbatimMode anyControlSeq
1504+ Tok _ (CtrlSeq name) _ <- anyControlSeq
15051505 optional $ symbol '='
15061506 spaces
15071507 -- we first parse in verbatim mode, and then expand macros,
@@ -1521,7 +1521,6 @@ defmacro = try $
15211521 Tok _ (CtrlSeq name) _ <- anyControlSeq
15221522 argspecs <- many (argspecArg <|> argspecPattern)
15231523 contents <- bracedOrToken
1524− doMacros -- after all this verbatim mode
15251524 return (name, Macro ExpandWhenUsed argspecs Nothing contents)
15261525
15271526argspecArg :: PandocMonad m => LP m ArgSpec
@@ -1559,7 +1558,6 @@ newcommand = do
15591558 case M.lookup name macros of
15601559 Just _ -> report $ MacroAlreadyDefined (T.unpack txt) pos
15611560 Nothing -> return ()
1562− doMacros -- after all this verbatim mode
15631561 return (name, Macro ExpandWhenUsed argspecs optarg contents)
15641562
15651563newenvironment :: PandocMonad m => LP m (Text, Macro, Macro)
src/Text/Pandoc/Readers/LaTeX/Parsing.hs+14−11
@@ -143,6 +143,7 @@ data LaTeXState = LaTeXState{ sOptions :: ReaderOptions
143143 , sLabels :: M.Map String [Inline]
144144 , sHasChapters :: Bool
145145 , sToggles :: M.Map String Bool
146+ , sExpanded :: Bool
146147 }
147148 deriving Show
148149
@@ -164,6 +165,7 @@ defaultLaTeXState = LaTeXState{ sOptions = def
164165 , sLabels = M.empty
165166 , sHasChapters = False
166167 , sToggles = M.empty
168+ , sExpanded = False
167169 }
168170
169171instance PandocMonad m => HasQuoteContext LaTeXState m where
@@ -249,8 +251,7 @@ rawLaTeXParser retokenize parser valParser = do
249251applyMacros :: (PandocMonad m, HasMacros s, HasReaderOptions s)
250252 => String -> ParserT String s m String
251253applyMacros s = (guardDisabled Ext_latex_macros >> return s) <|>
252− do let retokenize = doMacros *>
253− (toksToString <$> many (satisfyTok (const True)))
254+ do let retokenize = toksToString <$> many (satisfyTok (const True))
254255 pstate <- getState
255256 let lstate = def{ sOptions = extractReaderOptions pstate
256257 , sMacros = extractMacros pstate }
@@ -258,6 +259,7 @@ applyMacros s = (guardDisabled Ext_latex_macros >> return s) <|>
258259 case res of
259260 Left e -> fail (show e)
260261 Right s' -> return s'
262+
261263tokenize :: SourceName -> Text -> [Tok]
262264tokenize sourcename = totoks (initialPos sourcename)
263265
@@ -371,11 +373,10 @@ toksToString :: [Tok] -> String
371373toksToString = T.unpack . untokenize
372374
373375satisfyTok :: PandocMonad m => (Tok -> Bool) -> LP m Tok
374−satisfyTok f =
375− try $ do
376− res <- tokenPrim (T.unpack . untoken) updatePos matcher
376+satisfyTok f = do
377377 doMacros -- apply macros on remaining input stream
378− return res
378+ updateState $ \st -> st{ sExpanded = False }
379+ tokenPrim (T.unpack . untoken) updatePos matcher
379380 where matcher t | f t = Just t
380381 | otherwise = Nothing
381382 updatePos :: SourcePos -> Tok -> [Tok] -> SourcePos
@@ -384,12 +385,14 @@ satisfyTok f =
384385
385386doMacros :: PandocMonad m => LP m ()
386387doMacros = do
388+ expanded <- sExpanded <$> getState
387389 verbatimMode <- sVerbatimMode <$> getState
388− unless verbatimMode $ do
389− mbNewInp <- getInput >>= doMacros' 1
390− case mbNewInp of
391− Nothing -> return ()
392− Just inp -> setInput inp
390+ unless (expanded || verbatimMode) $ do
391+ mbNewInp <- getInput >>= doMacros' 1
392+ case mbNewInp of
393+ Nothing -> return ()
394+ Just inp -> setInput inp
395+ updateState $ \st -> st{ sExpanded = True }
393396
394397doMacros' :: PandocMonad m => Int -> [Tok] -> LP m (Maybe [Tok])
395398doMacros' n inp = do