Merge pull request #1680 from shelf/master

Respect indent when parsing Org bullet lists

Author: jgm

src/Text/Pandoc/Readers/Org.hs

@@ -853,12 +853,14 @@ list :: OrgParser (F Blocks)
 list = choice [ definitionList, bulletList, orderedList ] <?> "list"
 
 definitionList :: OrgParser (F Blocks)
-definitionList = fmap B.definitionList . fmap compactify'DL . sequence
-                 <$> many1 (definitionListItem bulletListStart)
+definitionList = try $ do n <- lookAhead (bulletListStart' Nothing)
+                          fmap B.definitionList . fmap compactify'DL . sequence
+                            <$> many1 (definitionListItem $ bulletListStart' (Just n))
 
 bulletList :: OrgParser (F Blocks)
-bulletList = fmap B.bulletList . fmap compactify' . sequence
-             <$> many1 (listItem bulletListStart)
+bulletList = try $ do n <- lookAhead (bulletListStart' Nothing)
+                      fmap B.bulletList . fmap compactify' . sequence
+                        <$> many1 (listItem (bulletListStart' $ Just n))
 
 orderedList :: OrgParser (F Blocks)
 orderedList = fmap B.orderedList . fmap compactify' . sequence
@@ -870,10 +872,27 @@ genericListStart listMarker = try $
   (+) <$> (length <$> many spaceChar)
       <*> (length <$> listMarker <* many1 spaceChar)
 
--- parses bullet list start and returns its length (excl. following whitespace)
+-- parses bullet list marker. maybe we know the indent level
 bulletListStart :: OrgParser Int
-bulletListStart = genericListStart bulletListMarker
-  where bulletListMarker = pure <$> oneOf "*-+"
+bulletListStart = bulletListStart' Nothing
+
+bulletListStart' :: Maybe Int -> OrgParser Int
+-- returns length of bulletList prefix, inclusive of marker
+bulletListStart' Nothing  = do ind <- many spaceChar
+                               oneOf bullets
+                               many1 spaceChar
+                               return $ length ind + 1
+ -- Unindented lists are legal, but they can't use '*' bullets
+ -- We return n to maintain compatibility with the generic listItem
+bulletListStart' (Just n) = do count (n-1) spaceChar
+                               oneOf validBullets
+                               many1 spaceChar
+                               return n
+  where validBullets = if n == 1 then noAsterisks else bullets
+        noAsterisks  = filter (/= '*') bullets
+
+bullets :: String
+bullets = "*+-"
 
 orderedListStart :: OrgParser Int
 orderedListStart = genericListStart orderedListMarker

tests/Tests/Readers/Org.hs

@@ -642,6 +642,33 @@ tests =
                        ]
                      ]
 
+      , "Bullet List with Decreasing Indent" =:
+           ("  - Discovery\n\
+            \ - Human After All\n") =?>
+           mconcat [ bulletList [ plain "Discovery" ]
+                   , bulletList [ plain ("Human" <> space <> "After" <> space <> "All")]
+                   ]
+
+      , "Header follows Bullet List" =:
+          ("  - Discovery\n\
+           \  - Human After All\n\
+           \* Homework") =?>
+          mconcat [ bulletList [ plain "Discovery"
+                               , plain ("Human" <> space <> "After" <> space <> "All")
+                               ]
+                  , header 1 "Homework"
+                  ]
+
+      , "Bullet List Unindented with trailing Header" =:
+          ("- Discovery\n\
+           \- Homework\n\
+           \* NotValidListItem") =?>
+          mconcat [ bulletList [ plain "Discovery"
+                               , plain "Homework"
+                               ]
+                  , header 1 "NotValidListItem"
+                  ]
+
       , "Simple Ordered List" =:
           ("1. Item1\n" ++
            "2. Item2\n") =?>
@@ -737,6 +764,16 @@ tests =
           , ("PCR", [ plain $ spcSep [ "polymerase", "chain", "reaction" ] ])
           ]
 
+      , "Definition List With Trailing Header" =:
+          "- definition :: list\n\
+          \- cool :: defs\n\
+          \* header" =?>
+          mconcat [ definitionList [ ("definition", [plain "list"])
+                                   , ("cool", [plain "defs"])
+                                   ]
+                  , header 1 "header"
+                  ]
+
       , "Loose bullet list" =:
           unlines [ "- apple"
                   , ""