Remove splitSentences from T.P.Shared [API change].

We used to attempt automatic sentence splitting in man and ms
output, since sentence-ending periods need to be followed by
two spaces or a newline in these formats.

But it's just too difficult to do this reliably.

Now that you can set `--wrap=preserve`, that option provides a
better way to ensure that sentences end at the end of a line,
if that's important to you.

3 files changed, 4 insertions, 50 deletions

diff --git a/src/Text/Pandoc/Shared.hs b/src/Text/Pandoc/Shared.hs
index 6806a78c6..bb40ae2cc 100644
--- a/src/Text/Pandoc/Shared.hs
+++ b/src/Text/Pandoc/Shared.hs
@@ -68,7 +68,6 @@ module Text.Pandoc.Shared (
                      makeMeta,
                      eastAsianLineBreakFilter,
                      htmlSpanLikeElements,
-                     splitSentences,
                      filterIpynbOutput,
                      -- * TagSoup HTML handling
                      renderTags',
@@ -707,49 +706,6 @@ eastAsianLineBreakFilter = bottomUp go
 htmlSpanLikeElements :: Set.Set T.Text
 htmlSpanLikeElements = Set.fromList ["kbd", "mark", "dfn"]
 
--- | Returns the first sentence in a list of inlines, and the rest.
-breakSentence :: [Inline] -> ([Inline], [Inline])
-breakSentence xs =
-  case break isStr xs of
-    (ys, Str t:zs)
-      | Just (t',t'') <- breakOnSentenceEnder t
-      , not (T.null t'') || startsWithSpace zs
-      -> (ys ++ [Str t'],
-          case T.stripStart t'' of
-            t''' | T.null t''' -> zs
-                 | otherwise -> Str t''' : zs)
-      | otherwise ->
-        let (as, bs) = breakSentence zs
-         in (ys ++ Str t : as, bs)
-    _ -> (xs, [])
- where
-  isStr (Str _) = True
-  isStr _ = False
-  breakOnSentenceEnder t = -- ". " ".)" "? " "?)" ".$" "! " " !)"
-    let (x,y) = T.break isSentenceEndPunct t
-     in if T.null y
-        then Nothing
-        else case T.uncons (T.drop 1 y) of
-          Nothing -> Just (t, mempty)
-          Just (c,_) | c == ' ' || c == '\r' || c == '\n' || c == ')'
-                  -> Just (x <> T.take 1 y, T.drop 1 y)
-          _ -> (\(w,z) -> (x <> T.take 1 y <> w, z)) <$>
-                  breakOnSentenceEnder (T.drop 1 y)
-  isSentenceEndPunct '.' = True
-  isSentenceEndPunct '?' = True
-  isSentenceEndPunct '!' = True
-  isSentenceEndPunct _   = False
-  startsWithSpace (LineBreak:_) = True
-  startsWithSpace (SoftBreak:_) = True
-  startsWithSpace [] = True
-  startsWithSpace _ = False
-
--- | Split a list of inlines into sentences.
-splitSentences :: [Inline] -> [[Inline]]
-splitSentences xs =
-  let (sent, rest) = breakSentence xs
-  in  if null rest then [sent] else sent : splitSentences rest
-
 -- | Process ipynb output cells.  If mode is Nothing,
 -- remove all output.  If mode is Just format, select
 -- best output for the format.  If format is not ipynb,
diff --git a/src/Text/Pandoc/Writers/Man.hs b/src/Text/Pandoc/Writers/Man.hs
index b3607cdd0..058273250 100644
--- a/src/Text/Pandoc/Writers/Man.hs
+++ b/src/Text/Pandoc/Writers/Man.hs
@@ -109,10 +109,9 @@ blockToMan :: PandocMonad m
 blockToMan _ Null = return empty
 blockToMan opts (Div _ bs) = blockListToMan opts bs
 blockToMan opts (Plain inlines) =
-  liftM vcat $ mapM (inlineListToMan opts) $ splitSentences inlines
+  inlineListToMan opts inlines
 blockToMan opts (Para inlines) = do
-  contents <- liftM vcat $ mapM (inlineListToMan opts) $
-    splitSentences inlines
+  contents <- inlineListToMan opts inlines
   return $ text ".PP" $$ contents
 blockToMan opts (LineBlock lns) =
   blockToMan opts $ linesToPara lns
diff --git a/src/Text/Pandoc/Writers/Ms.hs b/src/Text/Pandoc/Writers/Ms.hs
index d32ed53dc..c090cd08c 100644
--- a/src/Text/Pandoc/Writers/Ms.hs
+++ b/src/Text/Pandoc/Writers/Ms.hs
@@ -143,7 +143,7 @@ blockToMs opts (Div (ident,cls,kvs) bs) = do
        setFirstPara
        return $ anchor $$ res
 blockToMs opts (Plain inlines) =
-  liftM vcat $ mapM (inlineListToMs' opts) $ splitSentences inlines
+  inlineListToMs' opts inlines
 blockToMs opts (Para [Image attr alt (src,_tit)])
   | let ext = takeExtension (T.unpack src) in (ext == ".ps" || ext == ".eps") = do
   let (mbW,mbH) = (inPoints opts <$> dimension Width attr,
@@ -166,8 +166,7 @@ blockToMs opts (Para [Image attr alt (src,_tit)])
 blockToMs opts (Para inlines) = do
   firstPara <- gets stFirstPara
   resetFirstPara
-  contents <- liftM vcat $ mapM (inlineListToMs' opts) $
-    splitSentences inlines
+  contents <- inlineListToMs' opts inlines
   return $ literal (if firstPara then ".LP" else ".PP") $$ contents
 blockToMs _ b@(RawBlock f str)
   | f == Format "ms" = return $ literal str