From af7e78243682ec7c353dd202590c45c734f86ada Mon Sep 17 00:00:00 2001
From: John MacFarlane <jgm@berkeley.edu>
Date: Sat, 12 Dec 2015 09:31:51 -0800
Subject: [PATCH] Modified readers to emit SoftBreak when appropriate.

---
 src/Text/Pandoc/Readers/CommonMark.hs |  2 +-
 src/Text/Pandoc/Readers/HTML.hs       |  5 ++++-
 src/Text/Pandoc/Readers/LaTeX.hs      | 12 +++++++++---
 src/Text/Pandoc/Readers/MediaWiki.hs  |  3 ++-
 src/Text/Pandoc/Readers/Org.hs        |  2 +-
 src/Text/Pandoc/Readers/RST.hs        |  2 +-
 src/Text/Pandoc/Readers/Txt2Tags.hs   |  2 +-
 stack.yaml                            |  2 +-
 tests/testsuite.native                |  2 +-
 tests/writer.docbook                  |  2 +-
 tests/writer.dokuwiki                 |  2 +-
 tests/writer.html                     |  2 +-
 tests/writer.mediawiki                |  2 +-
 tests/writer.textile                  |  2 +-
 14 files changed, 26 insertions(+), 16 deletions(-)

diff --git a/src/Text/Pandoc/Readers/CommonMark.hs b/src/Text/Pandoc/Readers/CommonMark.hs
index 7f752c446..71c7d05b2 100644
--- a/src/Text/Pandoc/Readers/CommonMark.hs
+++ b/src/Text/Pandoc/Readers/CommonMark.hs
@@ -103,7 +103,7 @@ addInline (Node _ (TEXT t) _) = (map toinl clumps ++)
         toinl (' ':_)    = Space
         toinl xs         = Str xs
 addInline (Node _ LINEBREAK _) = (LineBreak :)
-addInline (Node _ SOFTBREAK _) = (Space :)
+addInline (Node _ SOFTBREAK _) = (SoftBreak :)
 addInline (Node _ (INLINE_HTML t) _) =
   (RawInline (Format "html") (unpack t) :)
 addInline (Node _ (CODE t) _) =
diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs
index 85e9a0743..a34e2fb5c 100644
--- a/src/Text/Pandoc/Readers/HTML.hs
+++ b/src/Text/Pandoc/Readers/HTML.hs
@@ -799,7 +799,10 @@ pBad = do
   return $ B.str [c']
 
 pSpace :: InlinesParser Inlines
-pSpace = many1 (satisfy isSpace) >> return B.space
+pSpace = many1 (satisfy isSpace) >>= \xs ->
+            if '\n' `elem` xs
+               then return B.softbreak
+               else return B.space
 
 --
 -- Constants
diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs
index 6500cb29f..5a4612862 100644
--- a/src/Text/Pandoc/Readers/LaTeX.hs
+++ b/src/Text/Pandoc/Readers/LaTeX.hs
@@ -100,8 +100,13 @@ dimenarg = try $ do
   return $ ch ++ num ++ dim
 
 sp :: LP ()
-sp = skipMany1 $ satisfy (\c -> c == ' ' || c == '\t')
-        <|> try (newline <* lookAhead anyChar <* notFollowedBy blankline)
+sp = whitespace <|> endline
+
+whitespace :: LP ()
+whitespace = skipMany1 $ satisfy (\c -> c == ' ' || c == '\t')
+
+endline :: LP ()
+endline = try (newline >> lookAhead anyChar >> notFollowedBy blankline)
 
 isLowerHex :: Char -> Bool
 isLowerHex x = x >= '0' && x <= '9' || x >= 'a' && x <= 'f'
@@ -196,7 +201,8 @@ singleQuote = do
 
 inline :: LP Inlines
 inline = (mempty <$ comment)
-     <|> (space  <$ sp)
+     <|> (space  <$ whitespace)
+     <|> (softbreak <$ endline)
      <|> inlineText
      <|> inlineCommand
      <|> inlineEnvironment
diff --git a/src/Text/Pandoc/Readers/MediaWiki.hs b/src/Text/Pandoc/Readers/MediaWiki.hs
index 24b3f5c7e..e423720df 100644
--- a/src/Text/Pandoc/Readers/MediaWiki.hs
+++ b/src/Text/Pandoc/Readers/MediaWiki.hs
@@ -555,7 +555,8 @@ inlineHtml :: MWParser Inlines
 inlineHtml = B.rawInline "html" . snd <$> htmlTag isInlineTag'
 
 whitespace :: MWParser Inlines
-whitespace = B.space <$ (skipMany1 spaceChar <|> endline <|> htmlComment)
+whitespace = B.space <$ (skipMany1 spaceChar <|> htmlComment)
+         <|> B.softbreak <$ endline
 
 endline :: MWParser ()
 endline = () <$ try (newline <*
diff --git a/src/Text/Pandoc/Readers/Org.hs b/src/Text/Pandoc/Readers/Org.hs
index 3be47cfd4..99a6927e2 100644
--- a/src/Text/Pandoc/Readers/Org.hs
+++ b/src/Text/Pandoc/Readers/Org.hs
@@ -1103,7 +1103,7 @@ endline = try $ do
   decEmphasisNewlinesCount
   guard =<< newlinesCountWithinLimits
   updateLastPreCharPos
-  return . return $ B.space
+  return . return $ B.softbreak
 
 cite :: OrgParser (F Inlines)
 cite = try $ do
diff --git a/src/Text/Pandoc/Readers/RST.hs b/src/Text/Pandoc/Readers/RST.hs
index 0e5bb2a87..85f34d9d8 100644
--- a/src/Text/Pandoc/Readers/RST.hs
+++ b/src/Text/Pandoc/Readers/RST.hs
@@ -1098,7 +1098,7 @@ endline = try $ do
      then notFollowedBy (anyOrderedListMarker >> spaceChar) >>
           notFollowedBy' bulletListStart
      else return ()
-  return B.space
+  return B.softbreak
 
 --
 -- links
diff --git a/src/Text/Pandoc/Readers/Txt2Tags.hs b/src/Text/Pandoc/Readers/Txt2Tags.hs
index 58841f2ce..c28ce1653 100644
--- a/src/Text/Pandoc/Readers/Txt2Tags.hs
+++ b/src/Text/Pandoc/Readers/Txt2Tags.hs
@@ -550,7 +550,7 @@ endline = try $ do
   notFollowedBy quote
   notFollowedBy list
   notFollowedBy table
-  return $ B.space
+  return $ B.softbreak
 
 str :: T2T Inlines
 str = try $ do
diff --git a/stack.yaml b/stack.yaml
index 5eb4609d4..d034f7265 100644
--- a/stack.yaml
+++ b/stack.yaml
@@ -9,7 +9,7 @@ packages:
 - '.'
 - location:
     git: 'https://github.com/jgm/pandoc-types'
-    commit: c46ab17eb17e16c3401a4ed48fc93e034a92b850
+    commit: 8aba9332f56fd23bc3c06065193abe5ace3e70b4
   extra-dep: true
 - location:
     git: 'https://github.com/jgm/texmath'
diff --git a/tests/testsuite.native b/tests/testsuite.native
index c887e5467..fa234dfc2 100644
--- a/tests/testsuite.native
+++ b/tests/testsuite.native
@@ -260,7 +260,7 @@ Pandoc (Meta {unMeta = fromList [("author",MetaList [MetaInlines [Str "John",Spa
 ,Div ("",[],[])
  [Div ("",[],[])
   [Div ("",[],[])
-   [Plain [Str "foo",SoftBreak]]]]
+   [Plain [Str "foo"]]]]
 ,Para [Str "This",Space,Str "should",Space,Str "just",Space,Str "be",Space,Str "an",Space,Str "HTML",Space,Str "comment:"]
 ,RawBlock (Format "html") "<!-- Comment -->"
 ,Para [Str "Multiline:"]
diff --git a/tests/writer.docbook b/tests/writer.docbook
index f4491c932..eee19cdd9 100644
--- a/tests/writer.docbook
+++ b/tests/writer.docbook
@@ -912,7 +912,7 @@ These should not be escaped:  \$ \\ \&gt; \[ \{
     Now, nested:
   </para>
   <para>
-    foo 
+    foo
   </para>
   <para>
     This should just be an HTML comment:
diff --git a/tests/writer.dokuwiki b/tests/writer.dokuwiki
index 92569a30f..fe1f8296a 100644
--- a/tests/writer.dokuwiki
+++ b/tests/writer.dokuwiki
@@ -350,7 +350,7 @@ As should this:
 </code>
 Now, nested:
 
-foo 
+foo
 
 
 
diff --git a/tests/writer.html b/tests/writer.html
index 10536b714..4a60a7b97 100644
--- a/tests/writer.html
+++ b/tests/writer.html
@@ -362,7 +362,7 @@ And this is <strong>strong</strong>
 <div>
 <div>
 <div>
-foo 
+foo
 </div>
 </div>
 </div>
diff --git a/tests/writer.mediawiki b/tests/writer.mediawiki
index af3df65f9..066606c00 100644
--- a/tests/writer.mediawiki
+++ b/tests/writer.mediawiki
@@ -374,7 +374,7 @@ Now, nested:
 
 <div>
 
-foo 
+foo
 
 </div>
 
diff --git a/tests/writer.textile b/tests/writer.textile
index 01bfae79a..293418ed5 100644
--- a/tests/writer.textile
+++ b/tests/writer.textile
@@ -424,7 +424,7 @@ Now, nested:
 
 <div>
 
-foo 
+foo
 
 </div>