From 041cfbd5eeffd2bc79447c4894652fc3c108c40d Mon Sep 17 00:00:00 2001
From: John MacFarlane <jgm@berkeley.edu>
Date: Sat, 2 Nov 2019 12:20:04 -0700
Subject: LaTeX untokenize: Ensure space between control sequence and following
 letter.

Closes #5836.
---
 src/Text/Pandoc/Readers/LaTeX.hs         |  1 +
 src/Text/Pandoc/Readers/LaTeX/Parsing.hs | 16 ++++++++++++++--
 2 files changed, 15 insertions(+), 2 deletions(-)

(limited to 'src/Text')

diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs
index b4e9de258..94fd9427b 100644
--- a/src/Text/Pandoc/Readers/LaTeX.hs
+++ b/src/Text/Pandoc/Readers/LaTeX.hs
@@ -618,6 +618,7 @@ accent combiningAccent fallBack = try $ do
        [Space]           -> return $ str [fromMaybe combiningAccent fallBack]
        []                -> return $ str [fromMaybe combiningAccent fallBack]
        _                 -> return ils
+
 mathDisplay :: String -> Inlines
 mathDisplay = displayMath . trimMath
 
diff --git a/src/Text/Pandoc/Readers/LaTeX/Parsing.hs b/src/Text/Pandoc/Readers/LaTeX/Parsing.hs
index af354843a..14cb408b0 100644
--- a/src/Text/Pandoc/Readers/LaTeX/Parsing.hs
+++ b/src/Text/Pandoc/Readers/LaTeX/Parsing.hs
@@ -92,6 +92,7 @@ import Text.Pandoc.Readers.LaTeX.Types (ExpansionPoint (..), Macro (..),
                                         ArgSpec (..), Tok (..), TokType (..))
 import Text.Pandoc.Shared
 import Text.Parsec.Pos
+-- import Debug.Trace
 
 newtype DottedNum = DottedNum [Int]
   deriving (Show)
@@ -350,10 +351,21 @@ isLowerHex :: Char -> Bool
 isLowerHex x = x >= '0' && x <= '9' || x >= 'a' && x <= 'f'
 
 untokenize :: [Tok] -> Text
-untokenize = mconcat . map untoken
+untokenize = foldr untokenAccum mempty
+
+untokenAccum :: Tok -> Text -> Text
+untokenAccum (Tok _ (CtrlSeq _) t) accum =
+  -- insert space to prevent breaking a control sequence; see #5836
+  case (T.unsnoc t, T.uncons accum) of
+    (Just (_,c), Just (d,_))
+      | isLetter c
+      , isLetter d
+      -> t <> " " <> accum
+    _ -> t <> accum
+untokenAccum (Tok _ _ t) accum = t <> accum
 
 untoken :: Tok -> Text
-untoken (Tok _ _ t) = t
+untoken t = untokenAccum t mempty
 
 toksToString :: [Tok] -> String
 toksToString = T.unpack . untokenize
-- 
cgit v1.2.3