041cfbd · typeable/pandoc

git.iko.soy/typeable/pandoc

LaTeX untokenize: Ensure space between control sequence and following letter.

Closes #5836.

commit041cfbd5eeffd2bc79447c4894652fc3c108c40d authorJohn MacFarlane <jgm@berkeley.edu> dateSat 2 Nov 2019, 12:20 (7 years ago) parent6c9a20b

2 files changed+15−2

compare with master

unifiedsplit

src/Text/Pandoc/Readers/LaTeX.hs+1 src/Text/Pandoc/Readers/LaTeX/Parsing.hs+14−2

src/Text/Pandoc/Readers/LaTeX.hs+1

@@ -618,6 +618,7 @@ accent combiningAccent fallBack = try $ do

618618 [Space] -> return $ str [fromMaybe combiningAccent fallBack]

619619 [] -> return $ str [fromMaybe combiningAccent fallBack]

620620 _ -> return ils

621+

621622mathDisplay :: String -> Inlines

622623mathDisplay = displayMath . trimMath

623624

src/Text/Pandoc/Readers/LaTeX/Parsing.hs+14−2

@@ -92,6 +92,7 @@ import Text.Pandoc.Readers.LaTeX.Types (ExpansionPoint (..), Macro (..),

9292 ArgSpec (..), Tok (..), TokType (..))

9393import Text.Pandoc.Shared

9494import Text.Parsec.Pos

95+-- import Debug.Trace

9596

9697newtype DottedNum = DottedNum [Int]

9798 deriving (Show)

@@ -350,10 +351,21 @@ isLowerHex :: Char -> Bool

350351isLowerHex x = x >= '0' && x <= '9' || x >= 'a' && x <= 'f'

351352

352353untokenize :: [Tok] -> Text

353−untokenize = mconcat . map untoken

354+untokenize = foldr untokenAccum mempty

355+

356+untokenAccum :: Tok -> Text -> Text

357+untokenAccum (Tok _ (CtrlSeq _) t) accum =

358+ -- insert space to prevent breaking a control sequence; see #5836

359+ case (T.unsnoc t, T.uncons accum) of

360+ (Just (_,c), Just (d,_))

361+ | isLetter c

362+ , isLetter d

363+ -> t <> " " <> accum

364+ _ -> t <> accum

365+untokenAccum (Tok _ _ t) accum = t <> accum

354366

355367untoken :: Tok -> Text

356−untoken (Tok _ _ t) = t

368+untoken t = untokenAccum t mempty

357369

358370toksToString :: [Tok] -> String

359371toksToString = T.unpack . untokenize